Kirill Yukhin <kirill.yukhin@intel.com>
Michael Zolotukhin <michael.v.zolotukhin@intel.com>
+ * gcc.target/i386/avx512cd-check.h: New file.
+ * gcc.target/i386/avx512cd-vpbroadcastmb2q-1.c: Ditto.
+ * gcc.target/i386/avx512cd-vpbroadcastmb2q-2.c: Ditto.
+ * gcc.target/i386/avx512cd-vpbroadcastmw2d-1.c: Ditto.
+ * gcc.target/i386/avx512cd-vpbroadcastmw2d-2.c: Ditto.
+ * gcc.target/i386/avx512cd-vpconflictd-1.c: Ditto.
+ * gcc.target/i386/avx512cd-vpconflictd-2.c: Ditto.
+ * gcc.target/i386/avx512cd-vpconflictq-1.c: Ditto.
+ * gcc.target/i386/avx512cd-vpconflictq-2.c: Ditto.
+ * gcc.target/i386/avx512cd-vplzcntd-1.c: Ditto.
+ * gcc.target/i386/avx512cd-vplzcntd-2.c: Ditto.
+ * gcc.target/i386/avx512cd-vplzcntq-1.c: Ditto.
+ * gcc.target/i386/avx512cd-vplzcntq-2.c: Ditto.
+ * gcc.target/i386/avx512cd-vptestnmd-1.c: Ditto.
+ * gcc.target/i386/avx512cd-vptestnmd-2.c: Ditto.
+ * gcc.target/i386/avx512cd-vptestnmq-1.c: Ditto.
+ * gcc.target/i386/avx512cd-vptestnmq-2.c: Ditto.
+ * gcc.target/i386/avx512er-vexp2pd-1.c: Ditto.
+ * gcc.target/i386/avx512er-vexp2pd-2.c: Ditto.
+ * gcc.target/i386/avx512er-vexp2ps-1.c: Ditto.
+ * gcc.target/i386/avx512er-vexp2ps-2.c: Ditto.
+ * gcc.target/i386/avx512er-vrcp28pd-1.c: Ditto.
+ * gcc.target/i386/avx512er-vrcp28pd-2.c: Ditto.
+ * gcc.target/i386/avx512er-vrcp28ps-1.c: Ditto.
+ * gcc.target/i386/avx512er-vrcp28ps-2.c: Ditto.
+ * gcc.target/i386/avx512er-vrsqrt28pd-1.c: Ditto.
+ * gcc.target/i386/avx512er-vrsqrt28pd-2.c: Ditto.
+ * gcc.target/i386/avx512er-vrsqrt28ps-1.c: Ditto.
+ * gcc.target/i386/avx512er-vrsqrt28ps-2.c: Ditto.
+ * gcc.target/i386/avx512f-broadcast-gpr-1.c: Ditto.
+ * gcc.target/i386/avx512f-broadcast-gpr-2.c: Ditto.
+ * gcc.target/i386/avx512f-ceil-sfix-vec-1.c: Ditto.
+ * gcc.target/i386/avx512f-ceil-sfix-vec-2.c: Ditto.
+ * gcc.target/i386/avx512f-dummy.c: Ditto.
+ * gcc.target/i386/avx512f-floor-sfix-vec-1.c: Ditto.
+ * gcc.target/i386/avx512f-floor-sfix-vec-2.c: Ditto.
+ * gcc.target/i386/avx512f-gather-1.c: Ditto.
+ * gcc.target/i386/avx512f-gather-2.c: Ditto.
+ * gcc.target/i386/avx512f-gather-3.c: Ditto.
+ * gcc.target/i386/avx512f-gather-4.c: Ditto.
+ * gcc.target/i386/avx512f-gather-5.c: Ditto.
+ * gcc.target/i386/avx512f-i32gatherd512-1.c: Ditto.
+ * gcc.target/i386/avx512f-i32gatherd512-2.c: Ditto.
+ * gcc.target/i386/avx512f-i32gatherpd512-1.c: Ditto.
+ * gcc.target/i386/avx512f-i32gatherpd512-2.c: Ditto.
+ * gcc.target/i386/avx512f-i32gatherps512-1.c: Ditto.
+ * gcc.target/i386/avx512f-i32gatherps512-2.c: Ditto.
+ * gcc.target/i386/avx512f-i32gatherq512-1.c: Ditto.
+ * gcc.target/i386/avx512f-i32gatherq512-2.c: Ditto.
+ * gcc.target/i386/avx512f-i32scatterd512-1.c: Ditto.
+ * gcc.target/i386/avx512f-i32scatterd512-2.c: Ditto.
+ * gcc.target/i386/avx512f-i32scatterpd512-1.c: Ditto.
+ * gcc.target/i386/avx512f-i32scatterpd512-2.c: Ditto.
+ * gcc.target/i386/avx512f-i32scatterps512-1.c: Ditto.
+ * gcc.target/i386/avx512f-i32scatterps512-2.c: Ditto.
+ * gcc.target/i386/avx512f-i32scatterq512-1.c: Ditto.
+ * gcc.target/i386/avx512f-i32scatterq512-2.c: Ditto.
+ * gcc.target/i386/avx512f-i64gatherd512-1.c: Ditto.
+ * gcc.target/i386/avx512f-i64gatherd512-2.c: Ditto.
+ * gcc.target/i386/avx512f-i64gatherpd512-1.c: Ditto.
+ * gcc.target/i386/avx512f-i64gatherpd512-2.c: Ditto.
+ * gcc.target/i386/avx512f-i64gatherps512-1.c: Ditto.
+ * gcc.target/i386/avx512f-i64gatherps512-2.c: Ditto.
+ * gcc.target/i386/avx512f-i64gatherq512-1.c: Ditto.
+ * gcc.target/i386/avx512f-i64gatherq512-2.c: Ditto.
+ * gcc.target/i386/avx512f-i64scatterd512-1.c: Ditto.
+ * gcc.target/i386/avx512f-i64scatterd512-2.c: Ditto.
+ * gcc.target/i386/avx512f-i64scatterpd512-1.c: Ditto.
+ * gcc.target/i386/avx512f-i64scatterpd512-2.c: Ditto.
+ * gcc.target/i386/avx512f-i64scatterps512-1.c: Ditto.
+ * gcc.target/i386/avx512f-i64scatterps512-2.c: Ditto.
+ * gcc.target/i386/avx512f-i64scatterq512-1.c: Ditto.
+ * gcc.target/i386/avx512f-i64scatterq512-2.c: Ditto.
+ * gcc.target/i386/avx512f-inline-asm.c: Ditto.
+ * gcc.target/i386/avx512f-kandnw-1.c: Ditto.
+ * gcc.target/i386/avx512f-kandw-1.c: Ditto.
+ * gcc.target/i386/avx512f-klogic-2.c: Ditto.
+ * gcc.target/i386/avx512f-knotw-1.c: Ditto.
+ * gcc.target/i386/avx512f-kortestw-1.c: Ditto.
+ * gcc.target/i386/avx512f-kortestw-2.c: Ditto.
+ * gcc.target/i386/avx512f-korw-1.c: Ditto.
+ * gcc.target/i386/avx512f-kunpckbw-1.c: Ditto.
+ * gcc.target/i386/avx512f-kxnorw-1.c: Ditto.
+ * gcc.target/i386/avx512f-kxorw-1.c: Ditto.
+ * gcc.target/i386/avx512f-rounding.c: Ditto.
+ * gcc.target/i386/avx512f-set-v16sf-1.c: Ditto.
+ * gcc.target/i386/avx512f-set-v16sf-2.c: Ditto.
+ * gcc.target/i386/avx512f-set-v16sf-3.c: Ditto.
+ * gcc.target/i386/avx512f-set-v16sf-4.c: Ditto.
+ * gcc.target/i386/avx512f-set-v16sf-5.c: Ditto.
+ * gcc.target/i386/avx512f-set-v16si-1.c: Ditto.
+ * gcc.target/i386/avx512f-set-v16si-2.c: Ditto.
+ * gcc.target/i386/avx512f-set-v16si-3.c: Ditto.
+ * gcc.target/i386/avx512f-set-v16si-4.c: Ditto.
+ * gcc.target/i386/avx512f-set-v16si-5.c: Ditto.
+ * gcc.target/i386/avx512f-set-v8df-1.c: Ditto.
+ * gcc.target/i386/avx512f-set-v8df-2.c: Ditto.
+ * gcc.target/i386/avx512f-set-v8df-3.c: Ditto.
+ * gcc.target/i386/avx512f-set-v8df-4.c: Ditto.
+ * gcc.target/i386/avx512f-set-v8df-5.c: Ditto.
+ * gcc.target/i386/avx512f-set-v8di-1.c: Ditto.
+ * gcc.target/i386/avx512f-set-v8di-2.c: Ditto.
+ * gcc.target/i386/avx512f-set-v8di-3.c: Ditto.
+ * gcc.target/i386/avx512f-set-v8di-4.c: Ditto.
+ * gcc.target/i386/avx512f-set-v8di-5.c: Ditto.
+ * gcc.target/i386/avx512f-setzero-pd-1.c: Ditto.
+ * gcc.target/i386/avx512f-setzero-ps-1.c: Ditto.
+ * gcc.target/i386/avx512f-setzero-si512-1.c: Ditto.
+ * gcc.target/i386/avx512f-vaddpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vaddpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vaddps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vaddps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vaddsd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vaddsd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vaddss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vaddss-2.c: Ditto.
+ * gcc.target/i386/avx512f-valignd-1.c: Ditto.
+ * gcc.target/i386/avx512f-valignd-2.c: Ditto.
+ * gcc.target/i386/avx512f-valignq-1.c: Ditto.
+ * gcc.target/i386/avx512f-valignq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vblendmpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vblendmpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vblendmps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vblendmps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vbroadcastf32x4-1.c: Ditto.
+ * gcc.target/i386/avx512f-vbroadcastf32x4-2.c: Ditto.
+ * gcc.target/i386/avx512f-vbroadcastf64x4-1.c: Ditto.
+ * gcc.target/i386/avx512f-vbroadcastf64x4-2.c: Ditto.
+ * gcc.target/i386/avx512f-vbroadcasti32x4-1.c: Ditto.
+ * gcc.target/i386/avx512f-vbroadcasti32x4-2.c: Ditto.
+ * gcc.target/i386/avx512f-vbroadcasti64x4-1.c: Ditto.
+ * gcc.target/i386/avx512f-vbroadcasti64x4-2.c: Ditto.
+ * gcc.target/i386/avx512f-vbroadcastsd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vbroadcastsd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vbroadcastss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vbroadcastss-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcmppd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcmppd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcmpps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcmpps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcmpsd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcmpsd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcmpss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcmpss-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcomisd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcomiss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcompresspd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcompresspd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcompressps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcompressps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtdq2pd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtdq2pd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtdq2ps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtdq2ps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtpd2dq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtpd2dq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtpd2ps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtpd2ps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtpd2udq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtpd2udq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtph2ps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtph2ps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtps2dq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtps2dq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtps2pd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtps2pd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtps2ph-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtps2ph-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtps2udq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtps2udq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtsd2si-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtsd2si64-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtsd2ss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtsd2ss-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtsd2usi-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtsd2usi-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtsd2usi64-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtsd2usi64-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtsi2sd64-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtsi2ss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtsi2ss64-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtss2sd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtss2sd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtss2si-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtss2si64-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtss2usi-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtss2usi-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtss2usi64-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtss2usi64-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttpd2dq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttpd2dq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttpd2udq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttpd2udq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttps2dq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttps2dq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttps2udq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttps2udq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttsd2si-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttsd2si-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttsd2si64-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttsd2si64-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttsd2usi-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttsd2usi-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttsd2usi64-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttsd2usi64-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttss2si-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttss2si-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttss2si64-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttss2si64-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttss2usi-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttss2usi-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttss2usi64-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvttss2usi64-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtudq2pd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtudq2pd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtudq2ps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtudq2ps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtusi2sd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtusi2sd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtusi2sd64-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtusi2sd64-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtusi2ss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtusi2ss-2.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtusi2ss64-1.c: Ditto.
+ * gcc.target/i386/avx512f-vcvtusi2ss64-2.c: Ditto.
+ * gcc.target/i386/avx512f-vdivpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vdivpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vdivps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vdivps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vdivsd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vdivsd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vdivss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vdivss-2.c: Ditto.
+ * gcc.target/i386/avx512f-vec-init.c: Ditto.
+ * gcc.target/i386/avx512f-vec-unpack.c: Ditto.
+ * gcc.target/i386/avx512f-vexpandpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vexpandpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vexpandps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vexpandps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vextractf32x4-1.c: Ditto.
+ * gcc.target/i386/avx512f-vextractf32x4-2.c: Ditto.
+ * gcc.target/i386/avx512f-vextractf64x4-1.c: Ditto.
+ * gcc.target/i386/avx512f-vextractf64x4-2.c: Ditto.
+ * gcc.target/i386/avx512f-vextracti32x4-1.c: Ditto.
+ * gcc.target/i386/avx512f-vextracti32x4-2.c: Ditto.
+ * gcc.target/i386/avx512f-vextracti64x4-1.c: Ditto.
+ * gcc.target/i386/avx512f-vextracti64x4-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfixupimmpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfixupimmpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfixupimmps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfixupimmps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfixupimmsd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfixupimmsd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfixupimmss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfixupimmss-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfmaddXXXpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfmaddXXXpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfmaddXXXps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfmaddXXXps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfmaddXXXsd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfmaddXXXsd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfmaddXXXss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfmaddXXXss-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfmaddsubXXXpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfmaddsubXXXpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfmaddsubXXXps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfmaddsubXXXps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfmsubXXXpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfmsubXXXpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfmsubXXXps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfmsubXXXps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfmsubXXXsd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfmsubXXXsd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfmsubXXXss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfmsubXXXss-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfmsubaddXXXpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfmsubaddXXXpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfmsubaddXXXps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfmsubaddXXXps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfnmaddXXXpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfnmaddXXXpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfnmaddXXXps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfnmaddXXXps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfnmaddXXXsd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfnmaddXXXsd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfnmaddXXXss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfnmaddXXXss-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfnmsubXXXpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfnmsubXXXpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfnmsubXXXps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfnmsubXXXps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfnmsubXXXsd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfnmsubXXXsd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vfnmsubXXXss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vfnmsubXXXss-2.c: Ditto.
+ * gcc.target/i386/avx512f-vgetexppd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vgetexppd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vgetexpps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vgetexpps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vgetexpsd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vgetexpsd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vgetexpss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vgetexpss-2.c: Ditto.
+ * gcc.target/i386/avx512f-vgetmantpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vgetmantpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vgetmantps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vgetmantps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vgetmantsd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vgetmantsd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vgetmantss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vgetmantss-2.c: Ditto.
+ * gcc.target/i386/avx512f-vinsertf32x4-1.c: Ditto.
+ * gcc.target/i386/avx512f-vinsertf32x4-2.c: Ditto.
+ * gcc.target/i386/avx512f-vinsertf64x4-1.c: Ditto.
+ * gcc.target/i386/avx512f-vinsertf64x4-2.c: Ditto.
+ * gcc.target/i386/avx512f-vinserti32x4-1.c: Ditto.
+ * gcc.target/i386/avx512f-vinserti32x4-2.c: Ditto.
+ * gcc.target/i386/avx512f-vinserti64x4-1.c: Ditto.
+ * gcc.target/i386/avx512f-vinserti64x4-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmaxpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmaxpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmaxps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmaxps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmaxsd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmaxsd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmaxss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmaxss-2.c: Ditto.
+ * gcc.target/i386/avx512f-vminpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vminpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vminps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vminps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vminsd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vminsd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vminss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vminss-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmovapd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmovapd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmovaps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmovaps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmovddup-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmovddup-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmovdqa32-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmovdqa32-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmovdqa64-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmovdqa64-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmovdqu32-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmovdqu32-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmovdqu64-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmovdqu64-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmovntdq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmovntdq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmovntpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmovntpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmovntps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmovntps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmovsd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmovsd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmovshdup-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmovshdup-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmovsldup-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmovsldup-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmovss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmovss-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmovupd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmovupd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmovups-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmovups-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmulpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmulpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmulps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmulps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmulsd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmulsd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vmulss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vmulss-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpabsd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpabsd512-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpabsq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpabsq512-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpaddd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpaddd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpaddq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpaddq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpandd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpandd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpandnd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpandnd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpandnq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpandnq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpandq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpandq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpblendmd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpblendmd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpblendmq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpblendmq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpbroadcastd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpbroadcastd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpbroadcastq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpbroadcastq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpcmpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpcmpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpcmpeqd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpcmpeqd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpcmpeqq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpcmpeqq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpcmpgtd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpcmpgtd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpcmpgtq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpcmpgtq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpcmpq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpcmpq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpcmpud-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpcmpud-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpcmpuq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpcmpuq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpcompressd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpcompressd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpcompressq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpcompressq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpermd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpermd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpermi2d-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpermi2d-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpermi2pd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpermi2pd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpermi2ps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpermi2ps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpermi2q-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpermi2q-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpermilpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpermilpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpermilpdi-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpermilpdi-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpermilps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpermilps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpermilpsi-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpermilpsi-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpermpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpermpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpermpdi-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpermpdi-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpermps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpermps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpermq-imm-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpermq-imm-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpermq-var-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpermq-var-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpermt2d-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpermt2d-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpermt2pd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpermt2pd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpermt2ps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpermt2ps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpermt2q-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpermt2q-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpexpandd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpexpandd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpexpandq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpexpandq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmaxsd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmaxsd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmaxsq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmaxsq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmaxud-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmaxud-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmaxuq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmaxuq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpminsd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpminsd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpminsq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpminsq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpminud-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpminud-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpminuq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpminuq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovdb-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovdb-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovdw-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovdw-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovqb-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovqb-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovqd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovqd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovqw-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovqw-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovsdb-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovsdb-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovsdw-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovsdw-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovsqb-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovsqb-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovsqd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovsqd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovsqw-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovsqw-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovsxbd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovsxbd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovsxbq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovsxbq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovsxdq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovsxdq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovsxwd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovsxwd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovsxwq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovsxwq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovusdb-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovusdb-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovusdw-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovusdw-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovusqb-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovusqb-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovusqd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovusqd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovusqw-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovusqw-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovzxbd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovzxbd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovzxbq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovzxbq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovzxdq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovzxdq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovzxwd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovzxwd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovzxwq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmovzxwq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmuldq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmuldq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmulld-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmulld-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpmuludq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpmuludq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpord-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpord-2.c: Ditto.
+ * gcc.target/i386/avx512f-vporq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vporq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vprold-1.c: Ditto.
+ * gcc.target/i386/avx512f-vprold-2.c: Ditto.
+ * gcc.target/i386/avx512f-vprolq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vprolq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vprolvd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vprolvd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vprolvq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vprolvq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vprord-1.c: Ditto.
+ * gcc.target/i386/avx512f-vprord-2.c: Ditto.
+ * gcc.target/i386/avx512f-vprorq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vprorq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vprorvd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vprorvd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vprorvq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vprorvq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpshufd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpshufd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpslld-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpslld-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpslldi-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpslldi-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpsllq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpsllq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpsllqi-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpsllqi-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpsllvd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpsllvd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpsllvq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpsllvq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpsllvq512-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpsllvq512-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpsrad-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpsrad-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpsradi-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpsradi-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpsraq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpsraq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpsraqi-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpsraqi-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpsravd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpsravd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpsravq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpsravq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpsravq512-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpsravq512-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpsrld-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpsrld-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpsrldi-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpsrldi-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpsrlq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpsrlq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpsrlqi-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpsrlqi-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpsrlvd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpsrlvd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpsrlvq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpsrlvq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpsrlvq512-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpsrlvq512-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpsubd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpsubd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpsubq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpsubq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpternlogd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpternlogd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpternlogq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpternlogq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vptestmd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vptestmd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vptestmq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vptestmq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpunpckhdq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpunpckhdq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpunpckhqdq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpunpckhqdq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpunpckldq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpunpckldq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpunpcklqdq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpunpcklqdq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpxord-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpxord-2.c: Ditto.
+ * gcc.target/i386/avx512f-vpxorq-1.c: Ditto.
+ * gcc.target/i386/avx512f-vpxorq-2.c: Ditto.
+ * gcc.target/i386/avx512f-vrcp14pd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vrcp14pd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vrcp14ps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vrcp14ps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vrcp14sd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vrcp14sd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vrcp14ss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vrcp14ss-2.c: Ditto.
+ * gcc.target/i386/avx512f-vrndscalepd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vrndscalepd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vrndscaleps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vrndscaleps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vrndscalesd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vrndscalesd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vrndscaless-1.c: Ditto.
+ * gcc.target/i386/avx512f-vrndscaless-2.c: Ditto.
+ * gcc.target/i386/avx512f-vrsqrt14pd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vrsqrt14pd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vrsqrt14ps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vrsqrt14ps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vrsqrt14sd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vrsqrt14sd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vrsqrt14ss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vrsqrt14ss-2.c: Ditto.
+ * gcc.target/i386/avx512f-vscalefpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vscalefpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vscalefps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vscalefps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vscalefsd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vscalefsd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vscalefss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vscalefss-2.c: Ditto.
+ * gcc.target/i386/avx512f-vshuff32x4-1.c: Ditto.
+ * gcc.target/i386/avx512f-vshuff32x4-2.c: Ditto.
+ * gcc.target/i386/avx512f-vshuff64x2-1.c: Ditto.
+ * gcc.target/i386/avx512f-vshuff64x2-2.c: Ditto.
+ * gcc.target/i386/avx512f-vshufi32x4-1.c: Ditto.
+ * gcc.target/i386/avx512f-vshufi32x4-2.c: Ditto.
+ * gcc.target/i386/avx512f-vshufi64x2-1.c: Ditto.
+ * gcc.target/i386/avx512f-vshufi64x2-2.c: Ditto.
+ * gcc.target/i386/avx512f-vshufpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vshufpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vshufps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vshufps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vsqrtpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vsqrtpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vsqrtps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vsqrtps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vsqrtsd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vsqrtsd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vsqrtss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vsqrtss-2.c: Ditto.
+ * gcc.target/i386/avx512f-vsubpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vsubpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vsubps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vsubps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vsubsd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vsubsd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vsubss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vsubss-2.c: Ditto.
+ * gcc.target/i386/avx512f-vucomisd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vucomiss-1.c: Ditto.
+ * gcc.target/i386/avx512f-vunpckhpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vunpckhpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vunpckhps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vunpckhps-2.c: Ditto.
+ * gcc.target/i386/avx512f-vunpcklpd-1.c: Ditto.
+ * gcc.target/i386/avx512f-vunpcklpd-2.c: Ditto.
+ * gcc.target/i386/avx512f-vunpcklps-1.c: Ditto.
+ * gcc.target/i386/avx512f-vunpcklps-2.c: Ditto.
+ * gcc.target/i386/avx512f_cond_move.c: Ditto.
+ * gcc.target/i386/avx512f_evex_reg_asm-1.c: Ditto.
+ * gcc.target/i386/avx512f_evex_reg_asm-2.c: Ditto.
+ * gcc.target/i386/avx512pf-vgatherpf0dps-1.c: Ditto.
+ * gcc.target/i386/avx512pf-vgatherpf0qps-1.c: Ditto.
+ * gcc.target/i386/avx512pf-vgatherpf1dps-1.c: Ditto.
+ * gcc.target/i386/avx512pf-vgatherpf1qps-1.c: Ditto.
+ * gcc.target/i386/avx512pf-vscatterpf0dps-1.c: Ditto.
+ * gcc.target/i386/avx512pf-vscatterpf0qps-1.c: Ditto.
+ * gcc.target/i386/avx512pf-vscatterpf1dps-1.c: Ditto.
+ * gcc.target/i386/avx512pf-vscatterpf1qps-1.c: Ditto.
+ * gcc.target/i386/sse-12.c: Updated options.
+ * gcc.target/i386/sse-13.c: Updated options, added defines for
+ __builtin_ia32_addpd512_mask, __builtin_ia32_addps512_mask,
+ __builtin_ia32_addsd_mask, __builtin_ia32_addss_mask,
+ __builtin_ia32_alignd512_mask, __builtin_ia32_alignq512_mask,
+ __builtin_ia32_cmpd512_mask, __builtin_ia32_cmppd512_mask,
+ __builtin_ia32_cmpps512_mask, __builtin_ia32_cmpq512_mask,
+ __builtin_ia32_cmpsd_mask, __builtin_ia32_cmpss_mask,
+ __builtin_ia32_cvtdq2ps512_mask, __builtin_ia32_cvtpd2dq512_mask,
+ __builtin_ia32_cvtpd2ps512_mask, __builtin_ia32_cvtpd2udq512_mask,
+ __builtin_ia32_cvtps2dq512_mask, __builtin_ia32_cvtps2pd512_mask,
+ __builtin_ia32_cvtps2udq512_mask, __builtin_ia32_cvtsd2ss_mask,
+ __builtin_ia32_cvtsi2sd64, __builtin_ia32_cvtsi2ss32,
+ __builtin_ia32_cvtsi2ss64, __builtin_ia32_cvtss2sd_mask,
+ __builtin_ia32_cvttpd2dq512_mask, __builtin_ia32_cvttpd2udq512_mask,
+ __builtin_ia32_cvttps2dq512_mask, __builtin_ia32_cvttps2udq512_mask,
+ __builtin_ia32_cvtudq2ps512_mask, __builtin_ia32_cvtusi2sd64,
+ __builtin_ia32_cvtusi2ss32, __builtin_ia32_cvtusi2ss64,
+ __builtin_ia32_divpd512_mask, __builtin_ia32_divps512_mask,
+ __builtin_ia32_divsd_mask, __builtin_ia32_divss_mask,
+ __builtin_ia32_extractf32x4_mask, __builtin_ia32_extractf64x4_mask,
+ __builtin_ia32_extracti32x4_mask, __builtin_ia32_extracti64x4_mask,
+ __builtin_ia32_fixupimmpd512_mask, __builtin_ia32_fixupimmpd512_maskz,
+ __builtin_ia32_fixupimmps512_mask, __builtin_ia32_fixupimmps512_maskz,
+ __builtin_ia32_fixupimmsd_mask, __builtin_ia32_fixupimmsd_maskz,
+ __builtin_ia32_fixupimmss_mask, __builtin_ia32_fixupimmss_maskz,
+ __builtin_ia32_gatherdiv8df, __builtin_ia32_gatherdiv8di,
+ __builtin_ia32_gatherdiv16sf, __builtin_ia32_gatherdiv16si,
+ __builtin_ia32_gathersiv16sf, __builtin_ia32_gathersiv16si,
+ __builtin_ia32_gathersiv8df, __builtin_ia32_gathersiv8di,
+ __builtin_ia32_getexppd512_mask, __builtin_ia32_getexpps512_mask,
+ __builtin_ia32_getexpsd128_mask, __builtin_ia32_getexpss128_mask,
+ __builtin_ia32_getmantpd512_mask, __builtin_ia32_getmantps512_mask,
+ __builtin_ia32_getmantsd_mask, __builtin_ia32_getmantss_mask,
+ __builtin_ia32_insertf32x4_mask, __builtin_ia32_insertf64x4_mask,
+ __builtin_ia32_inserti32x4_mask, __builtin_ia32_inserti64x4_mask,
+ __builtin_ia32_maxpd512_mask, __builtin_ia32_maxps512_mask,
+ __builtin_ia32_maxsd_mask, __builtin_ia32_maxss_mask,
+ __builtin_ia32_minpd512_mask, __builtin_ia32_minps512_mask,
+ __builtin_ia32_minsd_mask, __builtin_ia32_minss_mask,
+ __builtin_ia32_mulpd512_mask, __builtin_ia32_mulps512_mask,
+ __builtin_ia32_mulsd_mask, __builtin_ia32_mulss_mask,
+ __builtin_ia32_permdf512_mask, __builtin_ia32_permdi512_mask,
+ __builtin_ia32_prold512_mask, __builtin_ia32_prolq512_mask,
+ __builtin_ia32_prord512_mask, __builtin_ia32_prorq512_mask,
+ __builtin_ia32_pshufd512_mask, __builtin_ia32_pslldi512_mask,
+ __builtin_ia32_psllqi512_mask, __builtin_ia32_psradi512_mask,
+ __builtin_ia32_psraqi512_mask, __builtin_ia32_psrldi512_mask,
+ __builtin_ia32_psrlqi512_mask, __builtin_ia32_pternlogd512_mask,
+ __builtin_ia32_pternlogd512_maskz, __builtin_ia32_pternlogq512_mask,
+ __builtin_ia32_pternlogq512_maskz, __builtin_ia32_rndscalepd_mask,
+ __builtin_ia32_rndscaleps_mask, __builtin_ia32_rndscalesd_mask,
+ __builtin_ia32_rndscaless_mask, __builtin_ia32_scalefpd512_mask,
+ __builtin_ia32_scalefps512_mask, __builtin_ia32_scalefsd_mask,
+ __builtin_ia32_scalefss_mask, __builtin_ia32_scatterdiv8df,
+ __builtin_ia32_scatterdiv8di, __builtin_ia32_scatterdiv16sf,
+ __builtin_ia32_scatterdiv16si, __builtin_ia32_scattersiv16sf,
+ __builtin_ia32_scattersiv16si, __builtin_ia32_scattersiv8df,
+ __builtin_ia32_scattersiv8di, __builtin_ia32_shuf_f32x4_mask,
+ __builtin_ia32_shuf_f64x2_mask, __builtin_ia32_shuf_i32x4_mask,
+ __builtin_ia32_shuf_i64x2_mask, __builtin_ia32_shufpd512_mask,
+ __builtin_ia32_shufps512_mask, __builtin_ia32_sqrtpd512_mask,
+ __builtin_ia32_sqrtps512_mask, __builtin_ia32_sqrtsd_mask,
+ __builtin_ia32_sqrtss_mask, __builtin_ia32_subpd512_mask,
+ __builtin_ia32_subps512_mask, __builtin_ia32_subsd_mask,
+ __builtin_ia32_subss_mask, __builtin_ia32_ucmpd512_mask,
+ __builtin_ia32_ucmpq512_mask, __builtin_ia32_vcomisd,
+ __builtin_ia32_vcomiss, __builtin_ia32_vcvtph2ps512_mask,
+ __builtin_ia32_vcvtps2ph512_mask, __builtin_ia32_vcvtsd2si32,
+ __builtin_ia32_vcvtsd2si64, __builtin_ia32_vcvtsd2usi32,
+ __builtin_ia32_vcvtsd2usi64, __builtin_ia32_vcvtss2si32,
+ __builtin_ia32_vcvtss2si64, __builtin_ia32_vcvtss2usi32,
+ __builtin_ia32_vcvtss2usi64, __builtin_ia32_vcvttsd2si32,
+ __builtin_ia32_vcvttsd2si64, __builtin_ia32_vcvttsd2usi32,
+ __builtin_ia32_vcvttsd2usi64, __builtin_ia32_vcvttss2si32,
+ __builtin_ia32_vcvttss2si64, __builtin_ia32_vcvttss2usi32,
+ __builtin_ia32_vcvttss2usi64, __builtin_ia32_vfmaddpd512_mask,
+ __builtin_ia32_vfmaddpd512_mask3, __builtin_ia32_vfmaddpd512_maskz,
+ __builtin_ia32_vfmaddps512_mask, __builtin_ia32_vfmaddps512_mask3,
+ __builtin_ia32_vfmaddps512_maskz, __builtin_ia32_vfmaddsd3_mask,
+ __builtin_ia32_vfmaddsd3_mask3, __builtin_ia32_vfmaddsd3_maskz,
+ __builtin_ia32_vfmaddss3_mask, __builtin_ia32_vfmaddss3_mask3,
+ __builtin_ia32_vfmaddss3_maskz, __builtin_ia32_vfmaddsubpd512_mask,
+ __builtin_ia32_vfmaddsubpd512_mask3,
+ __builtin_ia32_vfmaddsubpd512_maskz,
+ __builtin_ia32_vfmaddsubps512_mask,
+ __builtin_ia32_vfmaddsubps512_mask3,
+ __builtin_ia32_vfmaddsubps512_maskz,
+ __builtin_ia32_vfmsubaddpd512_mask3,
+ __builtin_ia32_vfmsubaddps512_mask3, __builtin_ia32_vfmsubpd512_mask3,
+ __builtin_ia32_vfmsubps512_mask3, __builtin_ia32_vfmsubsd3_mask3,
+ __builtin_ia32_vfmsubss3_mask3, __builtin_ia32_vfnmaddpd512_mask,
+ __builtin_ia32_vfnmaddps512_mask, __builtin_ia32_vfnmsubpd512_mask,
+ __builtin_ia32_vfnmsubpd512_mask3, __builtin_ia32_vfnmsubps512_mask,
+ __builtin_ia32_vfnmsubps512_mask3, __builtin_ia32_vpermilpd512_mask,
+ __builtin_ia32_vpermilps512_mask, __builtin_ia32_exp2ps_mask,
+ __builtin_ia32_exp2pd_mask, __builtin_ia32_exp2ps_mask,
+ __builtin_ia32_exp2pd_mask, __builtin_ia32_rsqrt28ps_mask,
+ __builtin_ia32_rsqrt28pd_mask, __builtin_ia32_gatherpfdps,
+ __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdps,
+ __builtin_ia32_scatterpfqps, __builtin_ia32_addpd512_mask,
+ __builtin_ia32_addps512_mask, __builtin_ia32_addsd_mask,
+ __builtin_ia32_addss_mask, __builtin_ia32_alignd512_mask,
+ __builtin_ia32_alignq512_mask, __builtin_ia32_cmpd512_mask,
+ __builtin_ia32_cmppd512_mask, __builtin_ia32_cmpps512_mask,
+ __builtin_ia32_cmpq512_mask, __builtin_ia32_cmpsd_mask,
+ __builtin_ia32_cmpss_mask, __builtin_ia32_cvtdq2ps512_mask,
+ __builtin_ia32_cvtpd2dq512_mask, __builtin_ia32_cvtpd2ps512_mask,
+ __builtin_ia32_cvtpd2udq512_mask, __builtin_ia32_cvtps2dq512_mask,
+ __builtin_ia32_cvtps2pd512_mask, __builtin_ia32_cvtps2udq512_mask,
+ __builtin_ia32_cvtsd2ss_mask, __builtin_ia32_cvtsi2sd64,
+ __builtin_ia32_cvtsi2ss32, __builtin_ia32_cvtsi2ss64,
+ __builtin_ia32_cvtss2sd_mask, __builtin_ia32_cvttpd2dq512_mask,
+ __builtin_ia32_cvttpd2udq512_mask, __builtin_ia32_cvttps2dq512_mask,
+ __builtin_ia32_cvttps2udq512_mask, __builtin_ia32_cvtudq2ps512_mask,
+ __builtin_ia32_cvtusi2sd64, __builtin_ia32_cvtusi2ss32,
+ __builtin_ia32_cvtusi2ss64, __builtin_ia32_divpd512_mask,
+ __builtin_ia32_divps512_mask, __builtin_ia32_divsd_mask,
+ __builtin_ia32_divss_mask, __builtin_ia32_extractf32x4_mask,
+ __builtin_ia32_extractf64x4_mask, __builtin_ia32_extracti32x4_mask,
+ __builtin_ia32_extracti64x4_mask, __builtin_ia32_fixupimmpd512_mask,
+ __builtin_ia32_fixupimmpd512_maskz, __builtin_ia32_fixupimmps512_mask,
+ __builtin_ia32_fixupimmps512_maskz, __builtin_ia32_fixupimmsd_mask,
+ __builtin_ia32_fixupimmsd_maskz, __builtin_ia32_fixupimmss_mask,
+ __builtin_ia32_fixupimmss_maskz, __builtin_ia32_gatherdiv8df,
+ __builtin_ia32_gatherdiv8di, __builtin_ia32_gatherdiv16sf,
+ __builtin_ia32_gatherdiv16si, __builtin_ia32_gathersiv16sf,
+ __builtin_ia32_gathersiv16si, __builtin_ia32_gathersiv8df,
+ __builtin_ia32_gathersiv8di, __builtin_ia32_getexppd512_mask,
+ __builtin_ia32_getexpps512_mask, __builtin_ia32_getexpsd128_mask,
+ __builtin_ia32_getexpss128_mask, __builtin_ia32_getmantpd512_mask,
+ __builtin_ia32_getmantps512_mask, __builtin_ia32_getmantsd_mask,
+ __builtin_ia32_getmantss_mask, __builtin_ia32_insertf32x4_mask,
+ __builtin_ia32_insertf64x4_mask, __builtin_ia32_inserti32x4_mask,
+ __builtin_ia32_inserti64x4_mask, __builtin_ia32_maxpd512_mask,
+ __builtin_ia32_maxps512_mask, __builtin_ia32_maxsd_mask,
+ __builtin_ia32_maxss_mask, __builtin_ia32_minpd512_mask,
+ __builtin_ia32_minps512_mask, __builtin_ia32_minsd_mask,
+ __builtin_ia32_minss_mask, __builtin_ia32_mulpd512_mask,
+ __builtin_ia32_mulps512_mask, __builtin_ia32_mulsd_mask,
+ __builtin_ia32_mulss_mask, __builtin_ia32_permdf512_mask,
+ __builtin_ia32_permdi512_mask, __builtin_ia32_prold512_mask,
+ __builtin_ia32_prolq512_mask, __builtin_ia32_prord512_mask,
+ __builtin_ia32_prorq512_mask, __builtin_ia32_pshufd512_mask,
+ __builtin_ia32_pslldi512_mask, __builtin_ia32_psllqi512_mask,
+ __builtin_ia32_psradi512_mask, __builtin_ia32_psraqi512_mask,
+ __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlqi512_mask,
+ __builtin_ia32_pternlogd512_mask, __builtin_ia32_pternlogd512_maskz,
+ __builtin_ia32_pternlogq512_mask, __builtin_ia32_pternlogq512_maskz,
+ __builtin_ia32_rndscalepd_mask, __builtin_ia32_rndscaleps_mask,
+ __builtin_ia32_rndscalesd_mask, __builtin_ia32_rndscaless_mask,
+ __builtin_ia32_scalefpd512_mask, __builtin_ia32_scalefps512_mask,
+ __builtin_ia32_scalefsd_mask, __builtin_ia32_scalefss_mask,
+ __builtin_ia32_scatterdiv8df, __builtin_ia32_scatterdiv8di,
+ __builtin_ia32_scatterdiv16sf, __builtin_ia32_scatterdiv16si,
+ __builtin_ia32_scattersiv16sf, __builtin_ia32_scattersiv16si,
+ __builtin_ia32_scattersiv8df, __builtin_ia32_scattersiv8di,
+ __builtin_ia32_shuf_f32x4_mask, __builtin_ia32_shuf_f64x2_mask,
+ __builtin_ia32_shuf_i32x4_mask, __builtin_ia32_shuf_i64x2_mask,
+ __builtin_ia32_shufpd512_mask, __builtin_ia32_shufps512_mask,
+ __builtin_ia32_sqrtpd512_mask, __builtin_ia32_sqrtps512_mask,
+ __builtin_ia32_sqrtsd_mask, __builtin_ia32_sqrtss_mask,
+ __builtin_ia32_subpd512_mask, __builtin_ia32_subps512_mask,
+ __builtin_ia32_subsd_mask, __builtin_ia32_subss_mask,
+ __builtin_ia32_ucmpd512_mask, __builtin_ia32_ucmpq512_mask,
+ __builtin_ia32_vcomisd, __builtin_ia32_vcomiss,
+ __builtin_ia32_vcvtph2ps512_mask, __builtin_ia32_vcvtps2ph512_mask,
+ __builtin_ia32_vcvtsd2si32, __builtin_ia32_vcvtsd2si64,
+ __builtin_ia32_vcvtsd2usi32, __builtin_ia32_vcvtsd2usi64,
+ __builtin_ia32_vcvtss2si32, __builtin_ia32_vcvtss2si64,
+ __builtin_ia32_vcvtss2usi32, __builtin_ia32_vcvtss2usi64,
+ __builtin_ia32_vcvttsd2si32, __builtin_ia32_vcvttsd2si64,
+ __builtin_ia32_vcvttsd2usi32, __builtin_ia32_vcvttsd2usi64,
+ __builtin_ia32_vcvttss2si32, __builtin_ia32_vcvttss2si64,
+ __builtin_ia32_vcvttss2usi32, __builtin_ia32_vcvttss2usi64,
+ __builtin_ia32_vfmaddpd512_mask, __builtin_ia32_vfmaddpd512_mask3,
+ __builtin_ia32_vfmaddpd512_maskz, __builtin_ia32_vfmaddps512_mask,
+ __builtin_ia32_vfmaddps512_mask3, __builtin_ia32_vfmaddps512_maskz,
+ __builtin_ia32_vfmaddsd3_mask, __builtin_ia32_vfmaddsd3_mask3,
+ __builtin_ia32_vfmaddsd3_maskz, __builtin_ia32_vfmaddss3_mask,
+ __builtin_ia32_vfmaddss3_mask3, __builtin_ia32_vfmaddss3_maskz,
+ __builtin_ia32_vfmaddsubpd512_mask,
+ __builtin_ia32_vfmaddsubpd512_mask3,
+ __builtin_ia32_vfmaddsubpd512_maskz,
+ __builtin_ia32_vfmaddsubps512_mask,
+ __builtin_ia32_vfmaddsubps512_mask3,
+ __builtin_ia32_vfmaddsubps512_maskz,
+ __builtin_ia32_vfmsubaddpd512_mask3,
+ __builtin_ia32_vfmsubaddps512_mask3, __builtin_ia32_vfmsubpd512_mask3,
+ __builtin_ia32_vfmsubps512_mask3, __builtin_ia32_vfmsubsd3_mask3,
+ __builtin_ia32_vfmsubss3_mask3, __builtin_ia32_vfnmaddpd512_mask,
+ __builtin_ia32_vfnmaddps512_mask, __builtin_ia32_vfnmsubpd512_mask,
+ __builtin_ia32_vfnmsubpd512_mask3, __builtin_ia32_vfnmsubps512_mask,
+ __builtin_ia32_vfnmsubps512_mask3, __builtin_ia32_vpermilpd512_mask,
+ __builtin_ia32_vpermilps512_mask, __builtin_ia32_gatherpfdps,
+ __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdps,
+ __builtin_ia32_scatterpfqps, __builtin_ia32_exp2pd_mask,
+ __builtin_ia32_exp2ps_mask, __builtin_ia32_rcp28pd_mask,
+ __builtin_ia32_rcp28ps_mask, __builtin_ia32_rsqrt28pd_mask,
+ __builtin_ia32_rsqrt28ps_mask.
+ * gcc.target/i386/sse-14.c (test_1y): New.
+ (test_2y): Ditto.
+ (test_2vx): Ditto.
+ (test_3x): Ditto.
+ (test_3v): Ditto.
+ (test_3vx): Ditto.
+ (test_4x): Ditto.
+ (test_4y): Ditto.
+ (test_4v): Ditto.
+ (pragma GCC target): Add avx512f, avx512er, avx512cd, avx512pf.
+ (tests): Add _mm512_cvt_roundepi32_ps, _mm512_cvt_roundepu32_ps,
+ _mm512_cvt_roundpd_epi32, _mm512_cvt_roundpd_epu32,
+ _mm512_cvt_roundpd_ps, _mm512_cvt_roundph_ps,
+ _mm512_cvt_roundps_epi32, _mm512_cvt_roundps_epu32,
+ _mm512_cvt_roundps_pd, _mm512_cvtps_ph, _mm512_cvtt_roundpd_epi32,
+ _mm512_cvtt_roundpd_epu32, _mm512_cvtt_roundps_epi32,
+ _mm512_cvtt_roundps_epu32, _mm512_extractf32x4_ps,
+ _mm512_extractf64x4_pd, _mm512_extracti32x4_epi32,
+ _mm512_extracti64x4_epi64, _mm512_getexp_round_pd,
+ _mm512_getexp_round_ps, _mm512_getmant_round_pd,
+ _mm512_getmant_round_ps, _mm512_permute_pd, _mm512_permute_ps,
+ _mm512_permutex_epi64, _mm512_permutex_pd, _mm512_rol_epi32,
+ _mm512_rol_epi64, _mm512_ror_epi32, _mm512_ror_epi64,
+ _mm512_shuffle_epi32, _mm512_slli_epi32, _mm512_slli_epi64,
+ _mm512_sqrt_round_pd, _mm512_sqrt_round_ps, _mm512_srai_epi32,
+ _mm512_srai_epi64, _mm512_srli_epi32, _mm512_srli_epi64,
+ _mm_cvt_roundsd_i32, _mm_cvt_roundsd_u32, _mm_cvt_roundss_i32,
+ _mm_cvt_roundss_u32, _mm_cvtt_roundsd_i32, _mm_cvtt_roundsd_u32,
+ _mm_cvtt_roundss_i32, _mm_cvtt_roundss_u32, _mm512_getmant_pd,
+ _mm512_getmant_ps, _mm_cvt_roundi32_ss, _mm512_add_round_pd,
+ _mm512_add_round_ps, _mm512_alignr_epi32, _mm512_alignr_epi64,
+ _mm512_cmp_epi32_mask, _mm512_cmp_epi64_mask, _mm512_cmp_epu32_mask,
+ _mm512_cmp_epu64_mask, _mm512_cmp_pd_mask, _mm512_cmp_ps_mask,
+ _mm512_div_round_pd, _mm512_div_round_ps, _mm512_i32gather_epi32,
+ _mm512_i32gather_epi64, _mm512_i32gather_pd, _mm512_i32gather_ps,
+ _mm512_i64gather_epi32, _mm512_i64gather_epi64, _mm512_i64gather_pd,
+ _mm512_i64gather_ps, _mm512_insertf32x4, _mm512_insertf64x4,
+ _mm512_inserti32x4, _mm512_inserti64x4,
+ _mm512_maskz_cvt_roundepi32_ps, _mm512_maskz_cvt_roundepu32_ps,
+ _mm512_maskz_cvt_roundpd_epi32, _mm512_maskz_cvt_roundpd_epu32,
+ _mm512_maskz_cvt_roundpd_ps, _mm512_maskz_cvt_roundph_ps,
+ _mm512_maskz_cvt_roundps_epi32, _mm512_maskz_cvt_roundps_epu32,
+ _mm512_maskz_cvt_roundps_pd, _mm512_maskz_cvtps_ph,
+ _mm512_maskz_cvtt_roundpd_epi32, _mm512_maskz_cvtt_roundpd_epu32,
+ _mm512_maskz_cvtt_roundps_epi32, _mm512_maskz_cvtt_roundps_epu32,
+ _mm512_maskz_extractf32x4_ps, _mm512_maskz_extractf64x4_pd,
+ _mm512_maskz_extracti32x4_epi32, _mm512_maskz_extracti64x4_epi64,
+ _mm512_maskz_getexp_round_pd, _mm512_maskz_getexp_round_ps,
+ _mm512_maskz_getmant_round_pd, _mm512_maskz_getmant_round_ps,
+ _mm512_maskz_permute_pd, _mm512_maskz_permute_ps,
+ _mm512_maskz_permutex_epi64, _mm512_maskz_permutex_pd,
+ _mm512_maskz_rol_epi32, _mm512_maskz_rol_epi64,
+ _mm512_maskz_ror_epi32, _mm512_maskz_ror_epi64,
+ _mm512_maskz_shuffle_epi32, _mm512_maskz_slli_epi32,
+ _mm512_maskz_slli_epi64, _mm512_maskz_sqrt_round_pd,
+ _mm512_maskz_sqrt_round_ps, _mm512_maskz_srai_epi32,
+ _mm512_maskz_srai_epi64, _mm512_maskz_srli_epi32,
+ _mm512_maskz_srli_epi64, _mm512_max_round_pd, _mm512_max_round_ps,
+ _mm512_min_round_pd, _mm512_min_round_ps, _mm512_mul_round_pd,
+ _mm512_mul_round_ps, _mm512_scalef_round_pd, _mm512_scalef_round_ps,
+ _mm512_shuffle_f32x4, _mm512_shuffle_f64x2, _mm512_shuffle_i32x4,
+ _mm512_shuffle_i64x2, _mm512_shuffle_pd, _mm512_shuffle_ps,
+ _mm512_sub_round_pd, _mm512_sub_round_ps, _mm_add_round_sd,
+ _mm_add_round_ss, _mm_cmp_sd_mask, _mm_cmp_ss_mask,
+ _mm_cvt_roundi64_sd, _mm_cvt_roundi64_ss, _mm_cvt_roundsd_ss,
+ _mm_cvt_roundss_sd, _mm_cvt_roundu32_ss, _mm_cvt_roundu64_sd,
+ _mm_cvt_roundu64_ss, _mm_div_round_sd, _mm_div_round_ss,
+ _mm_getexp_round_sd, _mm_getexp_round_ss, _mm_getmant_round_sd,
+ _mm_getmant_round_ss, _mm_mul_round_sd, _mm_mul_round_ss,
+ _mm_scalef_round_sd, _mm_scalef_round_ss, _mm_sqrt_round_sd,
+ _mm_sqrt_round_ss, _mm_sub_round_sd, _mm_sub_round_ss,
+ _mm512_cmp_round_pd_mask, _mm512_cmp_round_ps_mask,
+ _mm512_maskz_roundscale_round_pd, _mm512_maskz_roundscale_round_ps,
+ _mm_cmp_round_sd_mask, _mm_cmp_round_ss_mask, _mm_comi_round_sd,
+ _mm_comi_round_ss, _mm_roundscale_round_sd, _mm_roundscale_round_ss,
+ _mm512_fmadd_round_pd, _mm512_fmadd_round_ps,
+ _mm512_fmaddsub_round_pd, _mm512_fmaddsub_round_ps,
+ _mm512_fmsub_round_pd, _mm512_fmsub_round_ps,
+ _mm512_fmsubadd_round_pd, _mm512_fmsubadd_round_ps,
+ _mm512_fnmadd_round_pd, _mm512_fnmadd_round_ps,
+ _mm512_fnmsub_round_pd, _mm512_fnmsub_round_ps,
+ _mm512_mask_cmp_epi32_mask, _mm512_mask_cmp_epi64_mask,
+ _mm512_mask_cmp_epu32_mask, _mm512_mask_cmp_epu64_mask,
+ _mm512_mask_cmp_pd_mask, _mm512_mask_cmp_ps_mask,
+ _mm512_mask_cvt_roundepi32_ps, _mm512_mask_cvt_roundepu32_ps,
+ _mm512_mask_cvt_roundpd_epi32, _mm512_mask_cvt_roundpd_epu32,
+ _mm512_mask_cvt_roundpd_ps, _mm512_mask_cvt_roundph_ps,
+ _mm512_mask_cvt_roundps_epi32, _mm512_mask_cvt_roundps_epu32,
+ _mm512_mask_cvt_roundps_pd, _mm512_mask_cvtps_ph,
+ _mm512_mask_cvtt_roundpd_epi32, _mm512_mask_cvtt_roundpd_epu32,
+ _mm512_mask_cvtt_roundps_epi32, _mm512_mask_cvtt_roundps_epu32,
+ _mm512_mask_extractf32x4_ps, _mm512_mask_extractf64x4_pd,
+ _mm512_mask_extracti32x4_epi32, _mm512_mask_extracti64x4_epi64,
+ _mm512_mask_getexp_round_pd, _mm512_mask_getexp_round_ps,
+ _mm512_mask_getmant_round_pd, _mm512_mask_getmant_round_ps,
+ _mm512_mask_permute_pd, _mm512_mask_permute_ps,
+ _mm512_mask_permutex_epi64, _mm512_mask_permutex_pd,
+ _mm512_mask_rol_epi32, _mm512_mask_rol_epi64, _mm512_mask_ror_epi32,
+ _mm512_mask_ror_epi64, _mm512_mask_shuffle_epi32,
+ _mm512_mask_slli_epi32, _mm512_mask_slli_epi64,
+ _mm512_mask_sqrt_round_pd, _mm512_mask_sqrt_round_ps,
+ _mm512_mask_srai_epi32, _mm512_mask_srai_epi64,
+ _mm512_mask_srli_epi32, _mm512_mask_srli_epi64,
+ _mm512_maskz_add_round_pd, _mm512_maskz_add_round_ps,
+ _mm512_maskz_alignr_epi32, _mm512_maskz_alignr_epi64,
+ _mm512_maskz_div_round_pd, _mm512_maskz_div_round_ps,
+ _mm512_maskz_insertf32x4, _mm512_maskz_insertf64x4,
+ _mm512_maskz_inserti32x4, _mm512_maskz_inserti64x4,
+ _mm512_maskz_max_round_pd, _mm512_maskz_max_round_ps,
+ _mm512_maskz_min_round_pd, _mm512_maskz_min_round_ps,
+ _mm512_maskz_mul_round_pd, _mm512_maskz_mul_round_ps,
+ _mm512_maskz_scalef_round_pd, _mm512_maskz_scalef_round_ps,
+ _mm512_maskz_shuffle_f32x4, _mm512_maskz_shuffle_f64x2,
+ _mm512_maskz_shuffle_i32x4, _mm512_maskz_shuffle_i64x2,
+ _mm512_maskz_shuffle_pd, _mm512_maskz_shuffle_ps,
+ _mm512_maskz_sub_round_pd, _mm512_maskz_sub_round_ps,
+ _mm512_ternarylogic_epi32, _mm512_ternarylogic_epi64,
+ _mm_fmadd_round_sd, _mm_fmadd_round_ss, _mm_fmsub_round_sd,
+ _mm_fmsub_round_ss, _mm_fnmadd_round_sd, _mm_fnmadd_round_ss,
+ _mm_fnmsub_round_sd, _mm_fnmsub_round_ss, _mm_mask_cmp_sd_mask,
+ _mm_mask_cmp_ss_mask, _mm_maskz_add_round_sd, _mm_maskz_add_round_ss,
+ _mm_maskz_cvt_roundsd_ss, _mm_maskz_cvt_roundss_sd,
+ _mm_maskz_div_round_sd, _mm_maskz_div_round_ss,
+ _mm_maskz_getexp_round_sd, _mm_maskz_getexp_round_ss,
+ _mm_maskz_getmant_round_sd, _mm_maskz_getmant_round_ss,
+ _mm_maskz_mul_round_sd, _mm_maskz_mul_round_ss,
+ _mm_maskz_scalef_round_sd, _mm_maskz_scalef_round_ss,
+ _mm_maskz_sqrt_round_sd, _mm_maskz_sqrt_round_ss,
+ _mm_maskz_sub_round_sd, _mm_maskz_sub_round_ss,
+ _mm512_i32scatter_epi32, _mm512_i32scatter_epi64,
+ _mm512_i32scatter_pd, _mm512_i32scatter_ps, _mm512_i64scatter_epi32,
+ _mm512_i64scatter_epi64, _mm512_i64scatter_pd, _mm512_i64scatter_ps,
+ _mm512_mask_roundscale_round_pd, _mm512_mask_roundscale_round_ps,
+ _mm512_mask_cmp_round_pd_mask, _mm512_mask_cmp_round_ps_mask,
+ _mm_fixupimm_round_sd, _mm_fixupimm_round_ss,
+ _mm_mask_cmp_round_sd_mask, _mm_mask_cmp_round_ss_mask,
+ _mm_maskz_roundscale_round_sd, _mm_maskz_roundscale_round_ss,
+ _mm512_mask3_fmadd_round_pd, _mm512_mask3_fmadd_round_ps,
+ _mm512_mask3_fmaddsub_round_pd, _mm512_mask3_fmaddsub_round_ps,
+ _mm512_mask3_fmsub_round_pd, _mm512_mask3_fmsub_round_ps,
+ _mm512_mask3_fmsubadd_round_pd, _mm512_mask3_fmsubadd_round_ps,
+ _mm512_mask3_fnmadd_round_pd, _mm512_mask3_fnmadd_round_ps,
+ _mm512_mask3_fnmsub_round_pd, _mm512_mask3_fnmsub_round_ps,
+ _mm512_mask_add_round_pd, _mm512_mask_add_round_ps,
+ _mm512_mask_alignr_epi32, _mm512_mask_alignr_epi64,
+ _mm512_mask_div_round_pd, _mm512_mask_div_round_ps,
+ _mm512_mask_fmadd_round_pd, _mm512_mask_fmadd_round_ps,
+ _mm512_mask_fmaddsub_round_pd, _mm512_mask_fmaddsub_round_ps,
+ _mm512_mask_fmsub_round_pd, _mm512_mask_fmsub_round_ps,
+ _mm512_mask_fmsubadd_round_pd, _mm512_mask_fmsubadd_round_ps,
+ _mm512_mask_fnmadd_round_pd, _mm512_mask_fnmadd_round_ps,
+ _mm512_mask_fnmsub_round_pd, _mm512_mask_fnmsub_round_ps,
+ _mm512_mask_i32gather_epi32, _mm512_mask_i32gather_epi64,
+ _mm512_mask_i32gather_pd, _mm512_mask_i32gather_ps,
+ _mm512_mask_i64gather_epi32, _mm512_mask_i64gather_epi64,
+ _mm512_mask_i64gather_pd, _mm512_mask_i64gather_ps,
+ _mm512_mask_insertf32x4, _mm512_mask_insertf64x4,
+ _mm512_mask_inserti32x4, _mm512_mask_inserti64x4,
+ _mm512_mask_max_round_pd, _mm512_mask_max_round_ps,
+ _mm512_mask_min_round_pd, _mm512_mask_min_round_ps,
+ _mm512_mask_mul_round_pd, _mm512_mask_mul_round_ps,
+ _mm512_mask_scalef_round_pd, _mm512_mask_scalef_round_ps,
+ _mm512_mask_shuffle_f32x4, _mm512_mask_shuffle_f64x2,
+ _mm512_mask_shuffle_i32x4, _mm512_mask_shuffle_i64x2,
+ _mm512_mask_shuffle_pd, _mm512_mask_shuffle_ps,
+ _mm512_mask_sub_round_pd, _mm512_mask_sub_round_ps,
+ _mm512_mask_ternarylogic_epi32, _mm512_mask_ternarylogic_epi64,
+ _mm512_maskz_fmadd_round_pd, _mm512_maskz_fmadd_round_ps,
+ _mm512_maskz_fmaddsub_round_pd, _mm512_maskz_fmaddsub_round_ps,
+ _mm512_maskz_fmsub_round_pd, _mm512_maskz_fmsub_round_ps,
+ _mm512_maskz_fmsubadd_round_pd, _mm512_maskz_fmsubadd_round_ps,
+ _mm512_maskz_fnmadd_round_pd, _mm512_maskz_fnmadd_round_ps,
+ _mm512_maskz_fnmsub_round_pd, _mm512_maskz_fnmsub_round_ps,
+ _mm512_maskz_ternarylogic_epi32, _mm512_maskz_ternarylogic_epi64,
+ _mm_mask3_fmadd_round_sd, _mm_mask3_fmadd_round_ss,
+ _mm_mask3_fmsub_round_sd, _mm_mask3_fmsub_round_ss,
+ _mm_mask3_fnmadd_round_sd, _mm_mask3_fnmadd_round_ss,
+ _mm_mask3_fnmsub_round_sd, _mm_mask3_fnmsub_round_ss,
+ _mm_mask_add_round_sd, _mm_mask_add_round_ss, _mm_mask_cvt_roundsd_ss,
+ _mm_mask_cvt_roundss_sd, _mm_mask_div_round_sd, _mm_mask_div_round_ss,
+ _mm_mask_fmadd_round_sd, _mm_mask_fmadd_round_ss,
+ _mm_mask_fmsub_round_sd, _mm_mask_fmsub_round_ss,
+ _mm_mask_fnmadd_round_sd, _mm_mask_fnmadd_round_ss,
+ _mm_mask_fnmsub_round_sd, _mm_mask_fnmsub_round_ss,
+ _mm_mask_getexp_round_sd, _mm_mask_getexp_round_ss,
+ _mm_mask_getmant_round_sd, _mm_mask_getmant_round_ss,
+ _mm_mask_mul_round_sd, _mm_mask_mul_round_ss,
+ _mm_mask_scalef_round_sd, _mm_mask_scalef_round_ss,
+ _mm_mask_sqrt_round_sd, _mm_mask_sqrt_round_ss, _mm_mask_sub_round_sd,
+ _mm_mask_sub_round_ss, _mm_maskz_fmadd_round_sd,
+ _mm_maskz_fmadd_round_ss, _mm_maskz_fmsub_round_sd,
+ _mm_maskz_fmsub_round_ss, _mm_maskz_fnmadd_round_sd,
+ _mm_maskz_fnmadd_round_ss, _mm_maskz_fnmsub_round_sd,
+ _mm_maskz_fnmsub_round_ss, _mm512_mask_i32scatter_epi32,
+ _mm512_mask_i32scatter_epi64, _mm512_mask_i32scatter_pd,
+ _mm512_mask_i32scatter_ps, _mm512_mask_i64scatter_epi32,
+ _mm512_mask_i64scatter_epi64, _mm512_mask_i64scatter_pd,
+ _mm512_mask_i64scatter_ps, _mm_mask_getmant_sd, _mm_mask_getmant_ss,
+ _mm_mask_roundscale_round_sd, _mm_mask_roundscale_round_ss,
+ _mm512_mask_fixupimm_round_pd, _mm512_mask_fixupimm_round_ps,
+ _mm512_maskz_fixupimm_round_pd, _mm512_maskz_fixupimm_round_ps,
+ _mm_mask_fixupimm_round_sd, _mm_mask_fixupimm_round_ss,
+ _mm_maskz_fixupimm_round_sd, _mm_maskz_fixupimm_round_ss,
+ _mm512_mask_prefetch_i32gather_ps, _mm512_mask_prefetch_i32scatter_ps,
+ _mm512_mask_prefetch_i64gather_ps, _mm512_mask_prefetch_i64scatter_ps,
+ _mm512_exp2a23_round_pd, _mm512_exp2a23_round_ps,
+ _mm512_rcp28_round_pd, _mm512_rcp28_round_ps, _mm512_rsqrt28_round_pd,
+ _mm512_rsqrt28_round_ps, _mm512_maskz_exp2a23_round_pd,
+ _mm512_maskz_exp2a23_round_ps, _mm512_maskz_rcp28_round_pd,
+ _mm512_maskz_rcp28_round_ps, _mm512_maskz_rsqrt28_round_pd,
+ _mm512_maskz_rsqrt28_round_ps, _mm512_mask_exp2a23_round_pd,
+ _mm512_mask_exp2a23_round_ps, _mm512_mask_rcp28_round_pd,
+ _mm512_mask_rcp28_round_ps, _mm512_mask_rsqrt28_round_pd,
+ _mm512_mask_rsqrt28_round_ps.
+ * gcc.target/i386/testimm-10.c: New file.
+ * gcc.target/i386/testround-1.c: Ditto.
+ * gcc.target/i386/testround-2.c: Ditto.
+ * gcc.target/x86_64/abi/avx512f/test_m512_returning.c: Ditto.
+ * gcc.target/x86_64/abi/avx512f/test_passing_m512.c: Ditto.
+ * gcc.target/x86_64/abi/avx512f/test_passing_structs.c: Ditto.
+ * gcc.target/x86_64/abi/avx512f/test_passing_unions.c: Ditto.
+ * gcc.target/i386/avx512cd-check.h: Ditto.
+ * gcc.target/i386/avx512er-check.h: Ditto.
+ * gcc.target/i386/avx512f-check.h: Ditto.
+ * gcc.target/i386/avx512f-helper.h: Ditto.
+ * gcc.target/i386/avx512f-mask-type.h: Ditto.
+ * gcc.target/i386/avx512f-os-support.h: Ditto.
+ * gcc.target/i386/i386.exp (check_effective_target_avx512f): New.
+ (check_effective_target_avx512cd): Ditto.
+ (check_effective_target_avx512er): Ditto.
+ * gcc.target/i386/m128-check.h (CHECK_FP_EXP): Ditto.
+ * gcc.target/i386/m512-check.h: Ditto.
+ * gcc.target/x86_64/abi/avx512f/abi-avx512f.exp: New file.
+ * gcc.target/x86_64/abi/avx512f/args.h: Ditto.
+ * gcc.target/x86_64/abi/avx512f/asm-support.S: Ditto.
+ * gcc.target/x86_64/abi/avx512f/avx512f-check.h: Ditto.
+ * lib/target-supports.exp (check_effective_target_avx512f): New.
+
+2013-12-31 Alexander Ivchenko <alexander.ivchenko@intel.com>
+ Maxim Kuznetsov <maxim.kuznetsov@intel.com>
+ Sergey Lega <sergey.s.lega@intel.com>
+ Anna Tikhonova <anna.tikhonova@intel.com>
+ Ilya Tocar <ilya.tocar@intel.com>
+ Andrey Turetskiy <andrey.turetskiy@intel.com>
+ Ilya Verbin <ilya.verbin@intel.com>
+ Kirill Yukhin <kirill.yukhin@intel.com>
+ Michael Zolotukhin <michael.v.zolotukhin@intel.com>
+
* gcc.target/i386/avx-1.c: Extend to AVX-512.
* gcc.target/i386/sse-22.c: Ditto.
* gcc.target/i386/sse-23.c: Ditto.
/* { dg-do compile { target i?86-*-* x86_64-*-* } } */
-/* { dg-options "-O -pedantic-errors -march=k8 -msse4a -m3dnow -mavx -mavx2 -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi -mbmi2 -mtbm -mlwp -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -mprfchw -madx -mfxsr -mxsaveopt" } */
+/* { dg-options "-O -pedantic-errors -march=k8 -msse4a -m3dnow -mavx -mavx2 -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi -mbmi2 -mtbm -mlwp -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -mprfchw -madx -mfxsr -mxsaveopt -mavx512f -mavx512er -mavx512cd" } */
/* Test that {,x,e,p,t,s,w,a,b,i}mmintrin.h, mm3dnow.h, fma4intrin.h,
xopintrin.h, abmintrin.h, bmiintrin.h, tbmintrin.h, lwpintrin.h,
/* { dg-do compile { target i?86-*-* x86_64-*-* } } */
-/* { dg-options "-O -fkeep-inline-functions -march=k8 -msse4a -m3dnow -mavx -mavx2 -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi -mbmi2 -mtbm -mlwp -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -mprfchw -madx -mfxsr -mxsaveopt" } */
+/* { dg-options "-O -fkeep-inline-functions -march=k8 -msse4a -m3dnow -mavx -mavx2 -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi -mbmi2 -mtbm -mlwp -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -mprfchw -madx -mfxsr -mxsaveopt -mavx512f -mavx512er -mavx512cd" } */
/* Test that {,x,e,p,t,s,w,a,b,i}mmintrin.h, mm3dnow.h, fma4intrin.h,
xopintrin.h, abmintrin.h, bmiintrin.h, tbmintrin.h, lwpintrin.h,
--- /dev/null
+#include <stdlib.h>
+#include "cpuid.h"
+#include "m512-check.h"
+#include "avx512f-os-support.h"
+
+static void avx512cd_test (void);
+
+static void __attribute__ ((noinline)) do_test (void)
+{
+ avx512cd_test ();
+}
+
+int
+main ()
+{
+ unsigned int eax, ebx, ecx, edx;
+
+ if (!__get_cpuid (1, &eax, &ebx, &ecx, &edx))
+ return 0;
+
+ if ((ecx & bit_OSXSAVE) == (bit_OSXSAVE))
+ {
+ if (__get_cpuid_max (0, NULL) < 7)
+ return 0;
+
+ __cpuid_count (7, 0, eax, ebx, ecx, edx);
+
+ if ((avx512f_os_support ()) && ((ebx & (bit_AVX512CD)) == (bit_AVX512CD)))
+ {
+ do_test ();
+#ifdef DEBUG
+ printf ("PASSED\n");
+#endif
+ return 0;
+ }
+#ifdef DEBUG
+ printf ("SKIPPED\n");
+#endif
+ }
+#ifdef DEBUG
+ else
+ printf ("SKIPPED\n");
+#endif
+
+ return 0;
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512cd -O2" } */
+/* { dg-final { scan-assembler "vpbroadcastmb2q\[ \\t\]+\[^\n\]*k\[1-7\]\[^\n\]*%zmm\[0-7\]" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m8;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_broadcastmb_epi64 (m8);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512cd" } */
+/* { dg-require-effective-target avx512cd } */
+
+#define HAVE_512
+#define AVX512CD
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+
+CALC (long long *res, __mmask8 src)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ res[i] = src;
+}
+
+static void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_q) res;
+ long long res_ref[SIZE];
+ __mmask8 src;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ res.a[i] = -1;
+ }
+
+ res.x = INTRINSIC (_broadcastmb_epi64) (src);
+
+ CALC (res_ref, src);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512cd -O2" } */
+/* { dg-final { scan-assembler "vpbroadcastmw2d\[ \\t\]+\[^\n\]*k\[1-7\]\[^\n\]*%zmm\[0-7\]" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m16;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_broadcastmw_epi32 (m16);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512cd" } */
+/* { dg-require-effective-target avx512cd } */
+
+#define HAVE_512
+#define AVX512CD
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+
+CALC (int *res, __mmask16 src)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ res[i] = src;
+}
+
+static void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_d) res;
+ int res_ref[SIZE];
+ __mmask16 src;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ res.a[i] = -1;
+ }
+
+ res.x = INTRINSIC (_broadcastmw_epi32) (src);
+
+ CALC (res_ref, src);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512cd -O2" } */
+/* { dg-final { scan-assembler-times "vpconflictd\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpconflictd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpconflictd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i s;
+volatile __m512i res;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_conflict_epi32 (s);
+ res = _mm512_mask_conflict_epi32 (res, 2, s);
+ res = _mm512_maskz_conflict_epi32 (2, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512cd" } */
+/* { dg-require-effective-target avx512cd } */
+
+#define HAVE_512
+#define AVX512CD
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (int *s, int *r)
+{
+ int i, j;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = 0;
+ for (j = 0; j < i; j++)
+ {
+ r[i] |= s[j] == s[i] ? 1 << j : 0;
+ }
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) s, res1, res2, res3;
+ int res_ref[SIZE];
+ MASK_TYPE mask = MASK_VALUE;
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 1234 * (i % 5);
+ res1.a[i] = DEFAULT_VALUE;
+ res2.a[i] = DEFAULT_VALUE;
+ res3.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_conflict_epi32) (s.x);
+ res2.x = INTRINSIC (_mask_conflict_epi32) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_conflict_epi32) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512cd -O2" } */
+/* { dg-final { scan-assembler-times "vpconflictq\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpconflictq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vpconflictq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i s;
+volatile __m512i res;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_conflict_epi64 (s);
+ res = _mm512_mask_conflict_epi64 (res, 2, s);
+ res = _mm512_maskz_conflict_epi64 (2, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512cd" } */
+/* { dg-require-effective-target avx512cd } */
+
+#define HAVE_512
+#define AVX512CD
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (long long *s, long long *r)
+{
+ int i, j;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = 0;
+ for (j = 0; j < i; j++)
+ {
+ r[i] |= s[i] == s[j] ? 1 << j : 0;
+ }
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_q) s, res1, res2, res3;
+ long long res_ref[SIZE];
+ MASK_TYPE mask = MASK_VALUE;
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 12345678 * (i % 5);
+ res1.a[i] = DEFAULT_VALUE;
+ res2.a[i] = DEFAULT_VALUE;
+ res3.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_conflict_epi64) (s.x);
+ res2.x = INTRINSIC (_mask_conflict_epi64) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_conflict_epi64) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512cd -O2" } */
+/* { dg-final { scan-assembler-times "vplzcntd\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vplzcntd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1} } */
+/* { dg-final { scan-assembler-times "vplzcntd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i s;
+volatile __m512i res;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_lzcnt_epi32 (s);
+ res = _mm512_mask_lzcnt_epi32 (res, 2, s);
+ res = _mm512_maskz_lzcnt_epi32 (2, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512cd" } */
+/* { dg-require-effective-target avx512cd } */
+
+#define HAVE_512
+#define AVX512CD
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#include <strings.h>
+
+static void
+CALC (int *s, int *r)
+{
+ int i, res;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ res = 0;
+ while ((res < 32) && (((s[i] >> (31 - res)) & 1) == 0))
+ ++res;
+ r[i] = res;
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) s, res1, res2, res3;
+ int res_ref[SIZE];
+ MASK_TYPE mask = MASK_VALUE;
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 12345678 * (i % 5);
+ res1.a[i] = DEFAULT_VALUE;
+ res2.a[i] = DEFAULT_VALUE;
+ res3.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_lzcnt_epi32) (s.x);
+ res2.x = INTRINSIC (_mask_lzcnt_epi32) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_lzcnt_epi32) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512cd -O2" } */
+/* { dg-final { scan-assembler-times "vplzcntq\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vplzcntq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vplzcntq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i s;
+volatile __m512i res;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_lzcnt_epi64 (s);
+ res = _mm512_maskz_lzcnt_epi64 (2, s);
+ res = _mm512_mask_lzcnt_epi64 (res, 2, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512cd" } */
+/* { dg-require-effective-target avx512cd } */
+
+#define HAVE_512
+#define AVX512CD
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+#include <strings.h>
+
+static void
+CALC (long long *s, long long *r)
+{
+ int i, res;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ res = 0;
+ while ((res < 64) && (((s[i] >> (63 - res)) & 1) == 0))
+ ++res;
+ r[i] = res;
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_q) s, res1, res2, res3;
+ long long res_ref[SIZE];
+ MASK_TYPE mask = MASK_VALUE;
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 12345678 * (i % 5);
+ res1.a[i] = DEFAULT_VALUE;
+ res2.a[i] = DEFAULT_VALUE;
+ res3.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_lzcnt_epi64) (s.x);
+ res2.x = INTRINSIC (_mask_lzcnt_epi64) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_lzcnt_epi64) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512cd -O2" } */
+/* { dg-final { scan-assembler-times "vptestnmd\[ \\t\]+\[^\n\]*%zmm\[0-7\]\[^\n^k\]*k\[1-7\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vptestnmd\[ \\t\]+\[^\n\]*%zmm\[0-7\]\[^\n^k\]*k\[1-7\]\{" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m16;
+
+void extern
+avx512cd_test (void)
+{
+ m16 = _mm512_testn_epi32_mask (x, x);
+ m16 = _mm512_mask_testn_epi32_mask (3, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512cd" } */
+/* { dg-require-effective-target avx512cd } */
+
+#define AVX512CD
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (MASK_TYPE *res, int *src1, int *src2)
+{
+ int i;
+ *res = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (!(src1[i] & src2[i]))
+ *res = *res | one << i;
+}
+
+static void
+TEST (void)
+{
+ int i, sign = 1;
+ UNION_TYPE (AVX512F_LEN, i_d) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+ res1 = 0;
+ res2 = 0;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = i * i * sign;
+ src2.a[i] = i + 20;
+ sign = -sign;
+ }
+
+ res1 = INTRINSIC (_testn_epi32_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_testn_epi32_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res1 != res_ref)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res2 != res_ref)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512cd -O2" } */
+/* { dg-final { scan-assembler-times "vptestnmq\[ \\t\]+\[^\n\]*%zmm\[0-7\]\[^\n^k\]*k\[1-7\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vptestnmq\[ \\t\]+\[^\n\]*%zmm\[0-7\]\[^\n^k\]*k\[1-7\]\{" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m8;
+
+void extern
+avx512cd_test (void)
+{
+ m8 = _mm512_testn_epi64_mask (x, x);
+ m8 = _mm512_mask_testn_epi64_mask (3, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512cd" } */
+/* { dg-require-effective-target avx512cd } */
+
+#define AVX512CD
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (MASK_TYPE *res, long long *src1, long long *src2)
+{
+ int i;
+ *res = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (!(src1[i] & src2[i]))
+ *res = *res | one << i;
+}
+
+static void
+TEST (void)
+{
+ int i, sign = 1;
+ UNION_TYPE (AVX512F_LEN, i_q) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+ res1 = 0;
+ res2 = 0;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = i * i * sign;
+ src2.a[i] = i + 20;
+ sign = -sign;
+ }
+
+ res1 = INTRINSIC (_testn_epi64_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_testn_epi64_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res1 != res_ref)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res2 != res_ref)
+ abort ();
+}
--- /dev/null
+#include <stdlib.h>
+#include "cpuid.h"
+#include "m512-check.h"
+#include "avx512f-os-support.h"
+
+static void avx512er_test (void);
+
+static void __attribute__ ((noinline)) do_test (void)
+{
+ avx512er_test ();
+}
+
+int
+main ()
+{
+ unsigned int eax, ebx, ecx, edx;
+
+ if (!__get_cpuid (1, &eax, &ebx, &ecx, &edx))
+ return 0;
+
+ if ((ecx & bit_OSXSAVE) == (bit_OSXSAVE))
+ {
+ if (__get_cpuid_max (0, NULL) < 7)
+ return 0;
+
+ __cpuid_count (7, 0, eax, ebx, ecx, edx);
+
+ if ((avx512f_os_support ()) && ((ebx & bit_AVX512ER) == bit_AVX512ER))
+ {
+ do_test ();
+#ifdef DEBUG
+ printf ("PASSED\n");
+#endif
+ return 0;
+ }
+#ifdef DEBUG
+ printf ("SKIPPED\n");
+#endif
+ }
+#ifdef DEBUG
+ else
+ printf ("SKIPPED\n");
+#endif
+
+ return 0;
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512er -O2" } */
+/* { dg-final { scan-assembler-times "vexp2pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[\\n\]" 2 } } */
+/* { dg-final { scan-assembler-times "vexp2pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vexp2pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vexp2pd\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vexp2pd\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vexp2pd\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __mmask8 m;
+
+void extern
+avx512er_test (void)
+{
+ x = _mm512_exp2a23_pd (x);
+ x = _mm512_mask_exp2a23_pd (x, m, x);
+ x = _mm512_maskz_exp2a23_pd (m, x);
+ x = _mm512_exp2a23_round_pd (x, _MM_FROUND_TO_NEAREST_INT);
+ x = _mm512_mask_exp2a23_round_pd (x, m, x, _MM_FROUND_TO_NEG_INF);
+ x = _mm512_maskz_exp2a23_round_pd (m, x, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-require-effective-target avx512er } */
+/* { dg-options "-O2 -mavx512er" } */
+
+#include "avx512er-check.h"
+#include "avx512f-mask-type.h"
+#include "avx512f-helper.h"
+#include <math.h>
+
+void static
+compute_vexp2pd (double *s, double *r)
+{
+ int i;
+ for (i = 0; i < 8; i++)
+ r[i] = pow (2.0, s[i]);
+}
+
+void static
+avx512er_test (void)
+{
+ union512d src, res1, res2, res3;
+ __mmask8 mask = MASK_VALUE;
+ double res_ref[8];
+ int i;
+
+ for (i = 0; i < 8; i++)
+ {
+ src.a[i] = 179.345 - 6.5645 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = _mm512_exp2a23_pd (src.x);
+ res2.x = _mm512_mask_exp2a23_pd (res2.x, mask, src.x);
+ res3.x = _mm512_maskz_exp2a23_pd (mask, src.x);
+
+ compute_vexp2pd (src.a, res_ref);
+
+ if (check_rough_union512d (res1, res_ref, 0.0001))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, 8);
+ if (check_rough_union512d (res2, res_ref, 0.0001))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, 8);
+ if (check_rough_union512d (res3, res_ref, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512er -O2" } */
+/* { dg-final { scan-assembler-times "vexp2ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[\\n\]" 2 } } */
+/* { dg-final { scan-assembler-times "vexp2ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vexp2ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vexp2ps\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vexp2ps\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vexp2ps\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __mmask16 m;
+
+void extern
+avx512er_test (void)
+{
+ x = _mm512_exp2a23_ps (x);
+ x = _mm512_mask_exp2a23_ps (x, m, x);
+ x = _mm512_maskz_exp2a23_ps (m, x);
+ x = _mm512_exp2a23_round_ps (x, _MM_FROUND_TO_NEAREST_INT);
+ x = _mm512_mask_exp2a23_round_ps (x, m, x, _MM_FROUND_TO_POS_INF);
+ x = _mm512_maskz_exp2a23_round_ps (m, x, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-require-effective-target avx512er } */
+/* { dg-options "-O2 -mavx512er" } */
+
+#include "avx512er-check.h"
+#include "avx512f-mask-type.h"
+#include "avx512f-helper.h"
+#include <math.h>
+
+void static
+compute_vexp2ps (float *s, float *r)
+{
+ int i;
+ for (i = 0; i < 16; i++)
+ r[i] = pow (2.0, s[i]);
+}
+
+void static
+avx512er_test (void)
+{
+ union512 src, res1, res2, res3;
+ __mmask16 mask = MASK_VALUE;
+ float res_ref[16];
+ int i;
+
+ for (i = 0; i < 16; i++)
+ {
+ src.a[i] = 179.345 - 6.5645 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = _mm512_exp2a23_ps (src.x);
+ res2.x = _mm512_mask_exp2a23_ps (res2.x, mask, src.x);
+ res3.x = _mm512_maskz_exp2a23_ps (mask, src.x);
+
+ compute_vexp2ps (src.a, res_ref);
+
+ if (check_rough_union512 (res1, res_ref, 0.0001))
+ abort ();
+
+ MASK_MERGE ()(res_ref, mask, 16);
+ if (check_rough_union512 (res2, res_ref, 0.0001))
+ abort ();
+
+ MASK_ZERO ()(res_ref, mask, 16);
+ if (check_rough_union512 (res3, res_ref, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512er -O2" } */
+/* { dg-final { scan-assembler-times "vrcp28pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[\\n\]" 2 } } */
+/* { dg-final { scan-assembler-times "vrcp28pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vrcp28pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vrcp28pd\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vrcp28pd\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vrcp28pd\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __mmask8 m;
+
+void extern
+avx512er_test (void)
+{
+ x = _mm512_rcp28_pd (x);
+ x = _mm512_mask_rcp28_pd (x, m, x);
+ x = _mm512_maskz_rcp28_pd (m, x);
+ x = _mm512_rcp28_round_pd (x, _MM_FROUND_TO_NEAREST_INT);
+ x = _mm512_mask_rcp28_round_pd (x, m, x, _MM_FROUND_TO_NEG_INF);
+ x = _mm512_maskz_rcp28_round_pd (m, x, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-require-effective-target avx512er } */
+/* { dg-options "-O2 -mavx512er" } */
+
+#include "avx512er-check.h"
+#include "avx512f-mask-type.h"
+#include "avx512f-helper.h"
+
+void static
+compute_vrcp28pd (double *s, double *r)
+{
+ int i;
+ for (i = 0; i < 8; i++)
+ r[i] = 1.0 / s[i];
+}
+
+void static
+avx512er_test (void)
+{
+ union512d src, res1, res2, res3;
+ __mmask8 mask = MASK_VALUE;
+ double res_ref[8];
+ int i;
+
+ for (i = 0; i < 8; i++)
+ {
+ src.a[i] = 179.345 - 6.5645 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = _mm512_rcp28_pd (src.x);
+ res2.x = _mm512_mask_rcp28_pd (res2.x, mask, src.x);
+ res3.x = _mm512_maskz_rcp28_pd (mask, src.x);
+
+ compute_vrcp28pd (src.a, res_ref);
+
+ if (check_rough_union512d (res1, res_ref, 0.0001))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, 8);
+ if (check_rough_union512d (res2, res_ref, 0.0001))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, 8);
+ if (check_rough_union512d (res3, res_ref, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512er -O2" } */
+/* { dg-final { scan-assembler-times "vrcp28ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[\\n\]" 2 } } */
+/* { dg-final { scan-assembler-times "vrcp28ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vrcp28ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vrcp28ps\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vrcp28ps\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vrcp28ps\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __mmask16 m;
+
+void extern
+avx512er_test (void)
+{
+ x = _mm512_rcp28_ps (x);
+ x = _mm512_mask_rcp28_ps (x, m, x);
+ x = _mm512_maskz_rcp28_ps (m, x);
+ x = _mm512_rcp28_round_ps (x, _MM_FROUND_TO_NEAREST_INT);
+ x = _mm512_mask_rcp28_round_ps (x, m, x, _MM_FROUND_TO_POS_INF);
+ x = _mm512_maskz_rcp28_round_ps (m, x, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-require-effective-target avx512er } */
+/* { dg-options "-O2 -mavx512er" } */
+
+#include "avx512er-check.h"
+#include "avx512f-mask-type.h"
+#include "avx512f-helper.h"
+
+void static
+compute_vrcp28ps (float *s, float *r)
+{
+ int i;
+ for (i = 0; i < 16; i++)
+ r[i] = 1.0 / s[i];
+}
+
+void static
+avx512er_test (void)
+{
+ union512 src, res1, res2, res3;
+ __mmask16 mask = MASK_VALUE;
+ float res_ref[16];
+ int i;
+
+ for (i = 0; i < 16; i++)
+ {
+ src.a[i] = 179.345 - 6.5645 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = _mm512_rcp28_ps (src.x);
+ res2.x = _mm512_mask_rcp28_ps (res2.x, mask, src.x);
+ res3.x = _mm512_maskz_rcp28_ps (mask, src.x);
+
+ compute_vrcp28ps (src.a, res_ref);
+
+ if (check_rough_union512 (res1, res_ref, 0.0001))
+ abort ();
+
+ MASK_MERGE ()(res_ref, mask, 16);
+ if (check_rough_union512 (res2, res_ref, 0.0001))
+ abort ();
+
+ MASK_ZERO ()(res_ref, mask, 16);
+ if (check_rough_union512 (res3, res_ref, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512er -O2" } */
+/* { dg-final { scan-assembler-times "vrsqrt28pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[\\n\]" 2 } } */
+/* { dg-final { scan-assembler-times "vrsqrt28pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vrsqrt28pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vrsqrt28pd\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vrsqrt28pd\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vrsqrt28pd\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __mmask8 m;
+
+void extern
+avx512er_test (void)
+{
+ x = _mm512_rsqrt28_pd (x);
+ x = _mm512_mask_rsqrt28_pd (x, m, x);
+ x = _mm512_maskz_rsqrt28_pd (m, x);
+ x = _mm512_rsqrt28_round_pd (x, _MM_FROUND_TO_NEAREST_INT);
+ x = _mm512_mask_rsqrt28_round_pd (x, m, x, _MM_FROUND_TO_NEG_INF);
+ x = _mm512_maskz_rsqrt28_round_pd (m, x, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-require-effective-target avx512er } */
+/* { dg-options "-O2 -mavx512er" } */
+
+#include "avx512er-check.h"
+#include "avx512f-mask-type.h"
+#include "avx512f-helper.h"
+#include <math.h>
+
+void static
+compute_vrsqrt28pd (double *s, double *r)
+{
+ int i;
+ for (i = 0; i < 8; i++)
+ r[i] = 1.0 / sqrt (s[i]);
+}
+
+void static
+avx512er_test (void)
+{
+ union512d src, res1, res2, res3;
+ __mmask8 mask = MASK_VALUE;
+ double res_ref[8];
+ int i;
+
+ for (i = 0; i < 8; i++)
+ {
+ src.a[i] = 179.345 - 6.5645 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = _mm512_rsqrt28_pd (src.x);
+ res2.x = _mm512_mask_rsqrt28_pd (res2.x, mask, src.x);
+ res3.x = _mm512_maskz_rsqrt28_pd (mask, src.x);
+
+ compute_vrsqrt28pd (src.a, res_ref);
+
+ if (check_rough_union512d (res1, res_ref, 0.0001))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, 8);
+ if (check_rough_union512d (res2, res_ref, 0.0001))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, 8);
+ if (check_rough_union512d (res3, res_ref, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512er -O2" } */
+/* { dg-final { scan-assembler-times "vrsqrt28ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[\\n\]" 2 } } */
+/* { dg-final { scan-assembler-times "vrsqrt28ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vrsqrt28ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vrsqrt28ps\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vrsqrt28ps\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vrsqrt28ps\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __mmask16 m;
+
+void extern
+avx512er_test (void)
+{
+ x = _mm512_rsqrt28_ps (x);
+ x = _mm512_mask_rsqrt28_ps (x, m, x);
+ x = _mm512_maskz_rsqrt28_ps (m, x);
+ x = _mm512_rsqrt28_round_ps (x, _MM_FROUND_TO_NEAREST_INT);
+ x = _mm512_mask_rsqrt28_round_ps (x, m, x, _MM_FROUND_TO_POS_INF);
+ x = _mm512_maskz_rsqrt28_round_ps (m, x, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-require-effective-target avx512er } */
+/* { dg-options "-O2 -mavx512er" } */
+
+#include "avx512er-check.h"
+#include "avx512f-mask-type.h"
+#include "avx512f-helper.h"
+#include <math.h>
+
+void static
+compute_vrsqrt28ps (float *s, float *r)
+{
+ int i;
+ for (i = 0; i < 16; i++)
+ r[i] = 1.0 / sqrt (s[i]);
+}
+
+void static
+avx512er_test (void)
+{
+ union512 src, res1, res2, res3;
+ __mmask16 mask = MASK_VALUE;
+ float res_ref[16];
+ int i;
+
+ for (i = 0; i < 16; i++)
+ {
+ src.a[i] = 179.345 - 6.5645 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = _mm512_rsqrt28_ps (src.x);
+ res2.x = _mm512_mask_rsqrt28_ps (res2.x, mask, src.x);
+ res3.x = _mm512_maskz_rsqrt28_ps (mask, src.x);
+
+ compute_vrsqrt28ps (src.a, res_ref);
+
+ if (check_rough_union512 (res1, res_ref, 0.0001))
+ abort ();
+
+ MASK_MERGE ()(res_ref, mask, 16);
+ if (check_rough_union512 (res2, res_ref, 0.0001))
+ abort ();
+
+ MASK_ZERO ()(res_ref, mask, 16);
+ if (check_rough_union512 (res3, res_ref, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpbroadcastq\[ \\t\]+%r\[^\n\]+%zmm\[0-9\]\[^\{\]" 1 { target { ! { ia32 } } } } } */
+/* { dg-final { scan-assembler-times "vpbroadcastd\[ \\t\]+%e\[^\n\]+%zmm\[0-9\]\[^\{\]" 1 { target { ! { ia32 } } } } } */
+/* { dg-final { scan-assembler-times "vpbroadcastq\[ \\t\]+\[^\n\]+%zmm\[0-9\]\[^\{\]" 1 { target ia32 } } } */
+/* { dg-final { scan-assembler-times "vpbroadcastd\[ \\t\]+\[^\n\]+%zmm\[0-9\]\[^\{\]" 1 { target ia32 } } } */
+
+#include <x86intrin.h>
+
+__m512i
+foo_1 (long long y)
+{
+ return __extension__ (__m512i)(__v8di){ y, y, y, y, y, y, y, y };
+}
+
+__m512i
+foo_2 (int y)
+{
+ return __extension__ (__m512i)(__v16si){ y, y, y, y, y, y, y, y, y,
+ y, y, y, y, y, y, y };
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+#include "avx512f-broadcast-gpr-1.c"
+
+void
+avx512f_test (void)
+{
+ union512i_q q;
+ union512i_d d;
+ int i;
+
+ q.x = foo_1 (3);
+ d.x = foo_2 (5);
+
+ for (i = 0; i < 8; i++)
+ {
+ if (q.a[i] != 3)
+ abort ();
+ }
+
+ for (i = 0; i < 16; i++)
+ {
+ if (d.a[i] != 5)
+ abort ();
+ }
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -ffast-math -ftree-vectorize -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+/* { dg-skip-if "no M_PI" { vxworks_kernel } } */
+
+#include <math.h>
+#include "avx512f-check.h"
+
+extern double ceil (double);
+
+#define NUM 64
+
+static void
+__attribute__((__target__("fpmath=sse")))
+init_src (double *src)
+{
+ int i, sign = 1;
+ double f = rand ();
+
+ for (i = 0; i < NUM; i++)
+ {
+ src[i] = (i + 1) * f * M_PI * sign;
+ if (i < (NUM / 2))
+ {
+ if ((i % 6) == 0)
+ f = f * src[i];
+ }
+ else if (i == (NUM / 2))
+ f = rand ();
+ else if ((i % 6) == 0)
+ f = 1 / (f * (i + 1) * src[i] * M_PI * sign);
+ sign = -sign;
+ }
+}
+
+static void
+__attribute__((__target__("fpmath=387")))
+avx512f_test (void)
+{
+ double a[NUM];
+ int r[NUM];
+ int i;
+
+ init_src (a);
+
+ for (i = 0; i < NUM; i++)
+ r[i] = (int) ceil (a[i]);
+
+ /* check results: */
+ for (i = 0; i < NUM; i++)
+ if (r[i] != (int) ceil (a[i]))
+ abort();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -ffast-math -ftree-vectorize -mavx512f" } */
+
+#include "avx512f-ceil-sfix-vec-1.c"
+
+/* { dg-final { scan-assembler "vrndscalepd\[^\n\]*zmm\[0-9\]" } } */
+/* { dg-final { scan-assembler "vcvttpd2dq\[^\n\]*zmm\[0-9\]" } } */
--- /dev/null
+#include <stdlib.h>
+#include "cpuid.h"
+#include "m512-check.h"
+#include "avx512f-os-support.h"
+
+static void avx512f_test (void);
+
+static void __attribute__ ((noinline)) do_test (void)
+{
+ avx512f_test ();
+}
+
+int
+main ()
+{
+ unsigned int eax, ebx, ecx, edx;
+
+ if (!__get_cpuid (1, &eax, &ebx, &ecx, &edx))
+ return 0;
+
+ /* Run AVX512F test only if host has AVX512F support. */
+ if ((ecx & bit_OSXSAVE) == (bit_OSXSAVE))
+ {
+ if (__get_cpuid_max (0, NULL) < 7)
+ return 0;
+
+ __cpuid_count (7, 0, eax, ebx, ecx, edx);
+
+ if ((avx512f_os_support ()) && ((ebx & bit_AVX512F) == bit_AVX512F))
+ {
+ do_test ();
+#ifdef DEBUG
+ printf ("PASSED\n");
+#endif
+ return 0;
+ }
+#ifdef DEBUG
+ printf ("SKIPPED\n");
+#endif
+ }
+#ifdef DEBUG
+ else
+ printf ("SKIPPED\n");
+#endif
+
+ return 0;
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+void static
+avx512f_test (void)
+{
+ union512i_q u, s1, s2;
+ long long e[8];
+ volatile int tst = check_union512i_q (u, e);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -ffast-math -ftree-vectorize -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+/* { dg-skip-if "no M_PI" { vxworks_kernel } } */
+
+#include <math.h>
+#include "avx512f-check.h"
+
+extern double floor (double);
+
+#define NUM 64
+
+static void
+__attribute__((__target__("fpmath=sse")))
+init_src (double *src)
+{
+ int i, sign = 1;
+ double f = rand ();
+
+ for (i = 0; i < NUM; i++)
+ {
+ src[i] = (i + 1) * f * M_PI * sign;
+ if (i < (NUM / 2))
+ {
+ if ((i % 6) == 0)
+ f = f * src[i];
+ }
+ else if (i == (NUM / 2))
+ f = rand ();
+ else if ((i % 6) == 0)
+ f = 1 / (f * (i + 1) * src[i] * M_PI * sign);
+ sign = -sign;
+ }
+}
+
+static void
+__attribute__((__target__("fpmath=387")))
+avx512f_test (void)
+{
+ double a[NUM];
+ int r[NUM];
+ int i;
+
+ init_src (a);
+
+ for (i = 0; i < NUM; i++)
+ r[i] = (int) floor (a[i]);
+
+ /* check results: */
+ for (i = 0; i < NUM; i++)
+ if (r[i] != (int) floor (a[i]))
+ abort();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -ffast-math -ftree-vectorize -mavx512f" } */
+
+#include "avx512f-floor-sfix-vec-1.c"
+
+/* { dg-final { scan-assembler "vrndscalepd\[^\n\]*zmm\[0-9\]" } } */
+/* { dg-final { scan-assembler "vcvttpd2dq\[^\n\]*zmm\[0-9\]" } } */
--- /dev/null
+/* { dg-do run } */
+/* { dg-require-effective-target avx512f } */
+/* { dg-options "-O3 -mavx512f" } */
+
+#include "avx512f-check.h"
+
+#define N 1024
+float vf1[N+16], vf2[N];
+double vd1[N+16], vd2[N];
+int vi1[N+16], vi2[N], k[N];
+long long vl1[N+16], vl2[N];
+long l[N];
+
+__attribute__((noinline, noclone)) void
+f1 (void)
+{
+ int i;
+ for (i = 0; i < N; i++)
+ vf2[i] = vf1[k[i]];
+}
+
+__attribute__((noinline, noclone)) void
+f2 (void)
+{
+ int i;
+ for (i = 0; i < N; i++)
+ vi2[i] = vi1[k[i]];
+}
+
+__attribute__((noinline, noclone)) void
+f3 (int x)
+{
+ int i;
+ for (i = 0; i < N; i++)
+ vf2[i] = vf1[k[i] + x];
+}
+
+__attribute__((noinline, noclone)) void
+f4 (int x)
+{
+ int i;
+ for (i = 0; i < N; i++)
+ vi2[i] = vi1[k[i] + x];
+}
+
+__attribute__((noinline, noclone)) void
+f5 (void)
+{
+ int i;
+ for (i = 0; i < N; i++)
+ vd2[i] = vd1[k[i]];
+}
+
+__attribute__((noinline, noclone)) void
+f6 (void)
+{
+ int i;
+ for (i = 0; i < N; i++)
+ vl2[i] = vl1[k[i]];
+}
+
+__attribute__((noinline, noclone)) void
+f7 (int x)
+{
+ int i;
+ for (i = 0; i < N; i++)
+ vd2[i] = vd1[k[i] + x];
+}
+
+__attribute__((noinline, noclone)) void
+f8 (int x)
+{
+ int i;
+ for (i = 0; i < N; i++)
+ vl2[i] = vl1[k[i] + x];
+}
+
+__attribute__((noinline, noclone)) void
+f9 (void)
+{
+ int i;
+ for (i = 0; i < N; i++)
+ vf2[i] = vf1[l[i]];
+}
+
+__attribute__((noinline, noclone)) void
+f10 (void)
+{
+ int i;
+ for (i = 0; i < N; i++)
+ vi2[i] = vi1[l[i]];
+}
+
+__attribute__((noinline, noclone)) void
+f11 (int x)
+{
+ int i;
+ for (i = 0; i < N; i++)
+ vf2[i] = vf1[l[i] + x];
+}
+
+__attribute__((noinline, noclone)) void
+f12 (int x)
+{
+ int i;
+ for (i = 0; i < N; i++)
+ vi2[i] = vi1[l[i] + x];
+}
+
+__attribute__((noinline, noclone)) void
+f13 (void)
+{
+ int i;
+ for (i = 0; i < N; i++)
+ vd2[i] = vd1[l[i]];
+}
+
+__attribute__((noinline, noclone)) void
+f14 (void)
+{
+ int i;
+ for (i = 0; i < N; i++)
+ vl2[i] = vl1[l[i]];
+}
+
+__attribute__((noinline, noclone)) void
+f15 (int x)
+{
+ int i;
+ for (i = 0; i < N; i++)
+ vd2[i] = vd1[l[i] + x];
+}
+
+__attribute__((noinline, noclone)) void
+f16 (int x)
+{
+ int i;
+ for (i = 0; i < N; i++)
+ vl2[i] = vl1[l[i] + x];
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+
+ for (i = 0; i < N + 16; i++)
+ {
+ asm ("");
+ vf1[i] = 17.0f + i;
+ vd1[i] = 19.0 + i;
+ vi1[i] = 21 + i;
+ vl1[i] = 23L + i;
+ }
+ for (i = 0; i < N; i++)
+ {
+ asm ("");
+ k[i] = (i * 731) & (N - 1);
+ l[i] = (i * 657) & (N - 1);
+ }
+
+ f1 ();
+ f2 ();
+ for (i = 0; i < N; i++)
+ if (vf2[i] != ((i * 731) & (N - 1)) + 17
+ || vi2[i] != ((i * 731) & (N - 1)) + 21)
+ abort ();
+
+ f3 (12);
+ f4 (14);
+ for (i = 0; i < N; i++)
+ if (vf2[i] != ((i * 731) & (N - 1)) + 17 + 12
+ || vi2[i] != ((i * 731) & (N - 1)) + 21 + 14)
+ abort ();
+
+ f5 ();
+ f6 ();
+ for (i = 0; i < N; i++)
+ if (vd2[i] != ((i * 731) & (N - 1)) + 19
+ || vl2[i] != ((i * 731) & (N - 1)) + 23)
+ abort ();
+
+ f7 (6);
+ f8 (3);
+ for (i = 0; i < N; i++)
+ if (vd2[i] != ((i * 731) & (N - 1)) + 19 + 6
+ || vl2[i] != ((i * 731) & (N - 1)) + 23 + 3)
+ abort ();
+
+ f9 ();
+ f10 ();
+ for (i = 0; i < N; i++)
+ if (vf2[i] != ((i * 657) & (N - 1)) + 17
+ || vi2[i] != ((i * 657) & (N - 1)) + 21)
+ abort ();
+
+ f11 (7);
+ f12 (9);
+ for (i = 0; i < N; i++)
+ if (vf2[i] != ((i * 657) & (N - 1)) + 17 + 7
+ || vi2[i] != ((i * 657) & (N - 1)) + 21 + 9)
+ abort ();
+
+ f13 ();
+ f14 ();
+ for (i = 0; i < N; i++)
+ if (vd2[i] != ((i * 657) & (N - 1)) + 19
+ || vl2[i] != ((i * 657) & (N - 1)) + 23)
+ abort ();
+
+ f15 (2);
+ f16 (12);
+ for (i = 0; i < N; i++)
+ if (vd2[i] != ((i * 657) & (N - 1)) + 19 + 2
+ || vl2[i] != ((i * 657) & (N - 1)) + 23 + 12)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */ /* PR59617 */
+/* { dg-options "-O3 -mavx512f -fdump-tree-vect-details" } */
+
+#include "avx512f-gather-1.c"
+
+/* { dg-final { scan-assembler-not "gather\[^\n\]*ymm\[^\n\]*ymm" { xfail { *-*-* } } } } */ /* PR59617 */
+/* { dg-final { scan-assembler-not "gather\[^\n\]*xmm\[^\n\]*ymm" { xfail { *-*-* } } } } */ /* PR59617 */
+/* { dg-final { scan-assembler-not "gather\[^\n\]*ymm\[^\n\]*xmm" { xfail { *-*-* } } } } */ /* PR59617 */
+/* { dg-final { scan-assembler-not "gather\[^\n\]*xmm\[^\n\]*xmm" { xfail { lp64 } } } } */ /* PR59617 */
+/* { dg-final { scan-tree-dump-times "note: vectorized 1 loops in function" 16 "vect" } } */
+/* { dg-final { cleanup-tree-dump "vect" } } */
--- /dev/null
+/* { dg-do run } */
+/* { dg-require-effective-target avx512f } */
+/* { dg-options "-O3 -mavx512f -ffast-math" } */
+
+#include "avx512f-check.h"
+
+#define N 1024
+float f[N];
+double d[N];
+int k[N];
+float *l[N];
+double *n[N];
+int **m[N];
+long q[N];
+long long **o[N];
+long long t[N];
+long long *r[N];
+int *s[N];
+
+__attribute__((noinline, noclone)) float
+f1 (void)
+{
+ int i;
+ float g = 0.0;
+ for (i = 0; i < N / 2; i++)
+ g += f[k[i]];
+ return g;
+}
+
+__attribute__((noinline, noclone)) float
+f2 (float *p)
+{
+ int i;
+ float g = 0.0;
+ for (i = 0; i < N / 2; i++)
+ g += p[k[i]];
+ return g;
+}
+
+__attribute__((noinline, noclone)) float
+f3 (void)
+{
+ int i;
+ float g = 0.0;
+ for (i = 0; i < N / 2; i++)
+ g += *l[i];
+ return g;
+}
+
+__attribute__((noinline, noclone)) int
+f4 (void)
+{
+ int i;
+ int g = 0;
+ for (i = 0; i < N / 2; i++)
+ g += **m[i];
+ return g;
+}
+
+__attribute__((noinline, noclone)) double
+f5 (void)
+{
+ int i;
+ double g = 0.0;
+ for (i = 0; i < N / 2; i++)
+ g += d[k[i]];
+ return g;
+}
+
+__attribute__((noinline, noclone)) double
+f6 (double *p)
+{
+ int i;
+ double g = 0.0;
+ for (i = 0; i < N / 2; i++)
+ g += p[k[i]];
+ return g;
+}
+
+__attribute__((noinline, noclone)) double
+f7 (void)
+{
+ int i;
+ double g = 0.0;
+ for (i = 0; i < N / 2; i++)
+ g += *n[i];
+ return g;
+}
+
+__attribute__((noinline, noclone)) int
+f8 (void)
+{
+ int i;
+ int g = 0;
+ for (i = 0; i < N / 2; i++)
+ g += **o[i];
+ return g;
+}
+
+__attribute__((noinline, noclone)) float
+f9 (void)
+{
+ int i;
+ float g = 0.0;
+ for (i = 0; i < N / 2; i++)
+ g += f[q[i]];
+ return g;
+}
+
+__attribute__((noinline, noclone)) float
+f10 (float *p)
+{
+ int i;
+ float g = 0.0;
+ for (i = 0; i < N / 2; i++)
+ g += p[q[i]];
+ return g;
+}
+
+__attribute__((noinline, noclone)) double
+f11 (void)
+{
+ int i;
+ double g = 0.0;
+ for (i = 0; i < N / 2; i++)
+ g += d[q[i]];
+ return g;
+}
+
+__attribute__((noinline, noclone)) double
+f12 (double *p)
+{
+ int i;
+ double g = 0.0;
+ for (i = 0; i < N / 2; i++)
+ g += p[q[i]];
+ return g;
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+
+ for (i = 0; i < N; i++)
+ {
+ asm ("");
+ f[i] = -256.0f + i;
+ d[i] = -258.0 + i;
+ k[i] = (i * 731) & (N - 1);
+ q[i] = (i * 657) & (N - 1);
+ t[i] = (i * 657) & (N - 1);
+ l[i] = &f[(i * 239) & (N - 1)];
+ n[i] = &d[(i * 271) & (N - 1)];
+ r[i] = &t[(i * 323) & (N - 1)];
+ s[i] = &k[(i * 565) & (N - 1)];
+ m[i] = &s[(i * 13) & (N - 1)];
+ o[i] = &r[(i * 19) & (N - 1)];
+ }
+
+ if (f1 () != 136448.0f || f2 (f) != 136448.0f || f3 () != 130304.0)
+ abort ();
+ if (f4 () != 261376 || f5 () != 135424.0 || f6 (d) != 135424.0)
+ abort ();
+ if (f7 () != 129280.0 || f8 () != 259840L || f9 () != 130816.0f)
+ abort ();
+ if (f10 (f) != 130816.0f || f11 () != 129792.0 || f12 (d) != 129792.0)
+ abort ();
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-require-effective-target avx512f } */
+/* { dg-options "-O3 -mavx512f" } */
+
+#include "avx512f-check.h"
+
+#define N 1024
+int a[N], b[N], c[N], d[N];
+
+__attribute__((noinline, noclone)) void
+foo (float *__restrict p, float *__restrict q, float *__restrict r,
+ int s1, int s2, int s3)
+{
+ int i;
+ for (i = 0; i < N; i++)
+ p[i] = q[a[i] * s1 + b[i] * s2 + s3] * r[c[i] * s1 + d[i] * s2 + s3];
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+ float e[N], f[N], g[N];
+ for (i = 0; i < N; i++)
+ {
+ a[i] = (i * 7) & (N / 8 - 1);
+ b[i] = (i * 13) & (N / 8 - 1);
+ c[i] = (i * 23) & (N / 8 - 1);
+ d[i] = (i * 5) & (N / 8 - 1);
+ e[i] = 16.5 + i;
+ f[i] = 127.5 - i;
+ }
+ foo (g, e, f, 3, 2, 4);
+ for (i = 0; i < N; i++)
+ if (g[i] != (float) ((20.5 + a[i] * 3 + b[i] * 2)
+ * (123.5 - c[i] * 3 - d[i] * 2)))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O3 -mavx512f" } */
+
+#include "avx512f-gather-4.c"
+
+/* { dg-final { scan-assembler "gather\[^\n\]*zmm" { xfail { *-*-* } } } } */ /* PR59617 */
+/* { dg-final { scan-assembler-not "gather\[^\n\]*ymm\[^\n\]*ymm" { xfail { *-*-* } } } } */ /* PR59617 */
+/* { dg-final { scan-assembler-not "gather\[^\n\]*xmm\[^\n\]*ymm" } } */
+/* { dg-final { scan-assembler-not "gather\[^\n\]*ymm\[^\n\]*xmm" } } */
+/* { dg-final { scan-assembler-not "gather\[^\n\]*xmm\[^\n\]*xmm" } } */
--- /dev/null
+/* This file is used to reduce a number of runtime tests for AVX512F
+ instructions. Idea is to create one file per instruction -
+ avx512f-insn-2.c - using defines from this file instead of intrinsic
+ name, vector length etc. Then dg-options are set with appropriate
+ -Dwhatever options in that .c file producing tests for specific
+ length. */
+
+#if defined (AVX512F)
+#include "avx512f-check.h"
+#elif defined (AVX512ER)
+#include "avx512er-check.h"
+#elif defined (AVX512CD)
+#include "avx512cd-check.h"
+#endif
+
+/* Macros expansion. */
+#define CONCAT(a,b,c) a ## b ## c
+#define EVAL(a,b,c) CONCAT(a,b,c)
+
+/* Value to be written into destination.
+ We have one value for all types so it must be small enough
+ to fit into signed char. */
+#define DEFAULT_VALUE 117
+
+#define MAKE_MASK_MERGE(NAME, TYPE) \
+static void \
+__attribute__((noinline, unused)) \
+merge_masking_##NAME (TYPE *arr, unsigned long long mask, int size) \
+{ \
+ int i; \
+ for (i = 0; i < size; i++) \
+ { \
+ arr[i] = (mask & (1LL << i)) ? arr[i] : DEFAULT_VALUE; \
+ } \
+}
+
+MAKE_MASK_MERGE(i_b, char)
+MAKE_MASK_MERGE(i_w, short)
+MAKE_MASK_MERGE(i_d, int)
+MAKE_MASK_MERGE(i_q, long long)
+MAKE_MASK_MERGE(, float)
+MAKE_MASK_MERGE(d, double)
+
+#define MASK_MERGE(TYPE) merge_masking_##TYPE
+
+#define MAKE_MASK_ZERO(NAME, TYPE) \
+static void \
+__attribute__((noinline, unused)) \
+zero_masking_##NAME (TYPE *arr, unsigned long long mask, int size) \
+{ \
+ int i; \
+ for (i = 0; i < size; i++) \
+ { \
+ arr[i] = (mask & (1LL << i)) ? arr[i] : 0; \
+ } \
+}
+
+MAKE_MASK_ZERO(i_b, char)
+MAKE_MASK_ZERO(i_w, short)
+MAKE_MASK_ZERO(i_d, int)
+MAKE_MASK_ZERO(i_q, long long)
+MAKE_MASK_ZERO(, float)
+MAKE_MASK_ZERO(d, double)
+
+#define MASK_ZERO(TYPE) zero_masking_##TYPE
+
+/* Intrinsic being tested. */
+#define INTRINSIC(NAME) EVAL(_mm, AVX512F_LEN, NAME)
+/* Unions used for testing (for example union512d, union256d etc.). */
+#define UNION_TYPE(SIZE, NAME) EVAL(union, SIZE, NAME)
+/* Corresponding union check. */
+#define UNION_CHECK(SIZE, NAME) EVAL(check_union, SIZE, NAME)
+/* Corresponding fp union check. */
+#define UNION_FP_CHECK(SIZE, NAME) EVAL(check_fp_union, SIZE, NAME)
+/* Corresponding rough union check. */
+#define UNION_ROUGH_CHECK(SIZE, NAME) \
+ EVAL(check_rough_union, SIZE, NAME)
+/* Function which tests intrinsic for given length. */
+#define TEST EVAL(test_, AVX512F_LEN,)
+/* Function which calculates result. */
+#define CALC EVAL(calc_, AVX512F_LEN,)
+
+#define AVX512F_LEN 512
+#define AVX512F_LEN_HALF 256
+static void test_512 ();
+
+#if defined (AVX512F)
+void
+avx512f_test (void) { test_512 (); }
+#elif defined (AVX512CD)
+void
+avx512cd_test (void) { test_512 (); }
+#elif defined (AVX512ER)
+void
+avx512er_test (void) { test_512 (); }
+#endif
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpgatherdd\[ \\t\]+\[^\n\]*zmm\[0-9\]\[^\n\]*zmm\[0-9\]{%k\[1-7\]}" 2 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x, idx;
+volatile __mmask16 m16;
+int *base;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_i32gather_epi32 (idx, base, 8);
+ x = _mm512_mask_i32gather_epi32 (x, m16, idx, base, 8);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+#define SCALE 2
+
+static void
+compute_gatherdd (int *res, __mmask16 m16, int *idx,
+ int *src, int scale, int *r)
+{
+ int i;
+
+ for (i = 0; i < 16; i++)
+ {
+ if (m16 & (1 << i))
+ r[i] = *(int *) (((unsigned char *) src) + idx[i] * scale);
+ else
+ r[i] = res[i];
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+ union512i_d idx, res;
+ int src[16];
+ int res_ref[16];
+ __mmask16 m16 = 0xBC5D;
+
+ for (i = 0; i < 16; i++)
+ {
+ src[i] = 1973 * (i + 1) * (i + 2);
+
+ /* About to gather in reverse order,
+ divide by 2 to demonstrate scale */
+ idx.a[i] = (64 - (i + 1) * 4) >> 1;
+ }
+
+ res.x = _mm512_mask_i32gather_epi32 (res.x, m16, idx.x, src, SCALE);
+ compute_gatherdd (res.a, m16, idx.a, src, SCALE, res_ref);
+
+ if (check_union512i_d (res, res_ref))
+ abort ();
+
+ res.x = _mm512_i32gather_epi32 (idx.x, src, SCALE);
+ compute_gatherdd (res.a, 0xFFFF, idx.a, src, SCALE, res_ref);
+
+ if (check_union512i_d (res, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vgatherdpd\[ \\t\]+\[^\n\]*ymm\[0-9\]\[^\n\]*zmm\[0-9\]{%k\[1-7\]}" 2 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __m256i idx;
+volatile __mmask8 m8;
+double *base;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_i32gather_pd (idx, base, 8);
+ x = _mm512_mask_i32gather_pd (x, m8, idx, base, 8);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+#define SCALE 2
+
+static void
+compute_gatherdpd (double *res, __mmask8 m8, int *idx,
+ double *src, int scale, double *r)
+{
+ int i;
+
+ for (i = 0; i < 8; i++)
+ {
+ if (m8 & (1 << i))
+ r[i] = *(double *) (((unsigned char *) src) + idx[i] * scale);
+ else
+ r[i] = res[i];
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+ union512d res;
+ union256i_d idx;
+ double src[8];
+ double res_ref[8];
+ __mmask8 m8 = 0xC5;
+
+ res.x = _mm512_setzero_pd();
+
+ for (i = 0; i < 8; i++)
+ {
+ src[i] = 2.718281828459045 * (i + 1) * (i + 2);
+
+ /* About to gather in reverse order,
+ divide by 2 to demonstrate scale */
+ idx.a[i] = (64 - (i + 1) * 8) >> 1;
+ }
+
+ res.x = _mm512_mask_i32gather_pd (res.x, m8, idx.x, src, SCALE);
+ compute_gatherdpd (res.a, m8, idx.a, src, SCALE, res_ref);
+
+ if (check_union512d (res, res_ref))
+ abort ();
+
+ res.x = _mm512_i32gather_pd (idx.x, src, SCALE);
+ compute_gatherdpd (res.a, 0xFF, idx.a, src, SCALE, res_ref);
+
+ if (check_union512d (res, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vgatherdps\[ \\t\]+\[^\n\]*zmm\[0-9\]\[^\n\]*zmm\[0-9\]{%k\[1-7\]}" 2 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __m512i idx;
+volatile __mmask16 m16;
+float *base;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_i32gather_ps (idx, base, 8);
+ x = _mm512_mask_i32gather_ps (x, m16, idx, base, 8);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+#define SCALE 2
+
+static void
+compute_gatherdps (float *res, __mmask16 m16, int *idx,
+ float *src, int scale, float *r)
+{
+ int i;
+
+ for (i = 0; i < 16; i++)
+ {
+ if (m16 & (1 << i))
+ r[i] = *(float *) (((unsigned char *) src) + idx[i] * scale);
+ else
+ r[i] = res[i];
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+ union512 res;
+ union512i_d idx;
+ float src[16];
+ float res_ref[16];
+ __mmask16 m16 = 0xBC5D;
+
+ res.x = _mm512_setzero_ps();
+
+ for (i = 0; i < 16; i++)
+ {
+ src[i] = 2.718281828459045 * (i + 1) * (i + 2);
+
+ /* About to gather in reverse order,
+ divide by 2 to demonstrate scale */
+ idx.a[i] = (64 - (i + 1) * 4) >> 1;
+ }
+
+ res.x = _mm512_mask_i32gather_ps (res.x, m16, idx.x, src, SCALE);
+ compute_gatherdps (res.a, m16, idx.a, src, SCALE, res_ref);
+
+ if (check_union512 (res, res_ref))
+ abort ();
+
+ res.x = _mm512_i32gather_ps (idx.x, src, SCALE);
+ compute_gatherdps (res.a, 0xFFFF, idx.a, src, SCALE, res_ref);
+
+ if (check_union512 (res, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpgatherdq\[ \\t\]+\[^\n\]*ymm\[0-9\]\[^\n\]*zmm\[0-9\]{%k\[1-7\]}" 2 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __m256i idx;
+volatile __mmask8 m8;
+long long *base;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_i32gather_epi64 (idx, base, 8);
+ x = _mm512_mask_i32gather_epi64 (x, m8, idx, base, 8);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+#define SCALE 2
+
+static void
+compute_gatherdq (long long *res, __mmask8 m8, int *idx,
+ long long *src, int scale, long long *r)
+{
+ int i;
+
+ for (i = 0; i < 8; i++)
+ {
+ if (m8 & (1 << i))
+ r[i] = *(long long *)
+ (((unsigned char *) src) + idx[i] * scale);
+ else
+ r[i] = res[i];
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+ union256i_d idx;
+ union512i_q res;
+ long long src[8];
+ long long res_ref[8];
+ __mmask8 m8 = 0xC5;
+
+ for (i = 0; i < 8; i++)
+ {
+ src[i] = 1983 * (i + 1) * (i + 2);
+
+ /* About to gather in reverse order,
+ divide by 2 to demonstrate scale */
+ idx.a[i] = (64 - (i + 1) * 8) >> 1;
+ }
+
+ res.x = _mm512_mask_i32gather_epi64 (res.x, m8, idx.x, src, SCALE);
+ compute_gatherdq (res.a, m8, idx.a, src, SCALE, res_ref);
+
+ if (check_union512i_q (res, res_ref))
+ abort ();
+
+ res.x = _mm512_i32gather_epi64 (idx.x, src, SCALE);
+ compute_gatherdq (res.a, 0xFF, idx.a, src, SCALE, res_ref);
+
+ if (check_union512i_q (res, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpscatterdd\[ \\t\]+\[^\n\]*zmm\[0-9\]\[^\n\]*zmm\[0-9\]\[^\n\]*{%k\[1-7\]}" 2 } } */
+
+#include <immintrin.h>
+
+volatile __m512i src, idx;
+volatile __mmask16 m16;
+int *addr;
+
+void extern
+avx512f_test (void)
+{
+ _mm512_i32scatter_epi32 (addr, idx, src, 8);
+ _mm512_mask_i32scatter_epi32 (addr, m16, idx, src, 8);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+#define SCALE 2
+
+static void
+compute_scatterdd (__mmask16 m16, int *idx,
+ int *src, int scale, int *r)
+{
+ int i;
+
+ for (i = 0; i < 16; i++)
+ {
+ if (m16 & (1 << i))
+ *(int *) (((unsigned char *) r) + idx[i] * scale) = src[i];
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+ union512i_d src, idx;
+ int res[16] = { 0 };
+ int res_ref[16] = { 0 };
+ __mmask16 m16 = 0xBC5D;
+
+ for (i = 0; i < 16; i++)
+ {
+ src.a[i] = 1973 * (i + 1) * (i + 2);
+
+ /* About to gather in reverse order,
+ divide by 2 to demonstrate scale */
+ idx.a[i] = (64 - (i + 1) * 4) >> 1;
+ }
+
+ _mm512_mask_i32scatter_epi32 (res, m16, idx.x, src.x, SCALE);
+ compute_scatterdd (m16, idx.a, src.a, SCALE, res_ref);
+
+ if (checkVi (res, res_ref, 16))
+ abort ();
+
+ _mm512_i32scatter_epi32 (res, idx.x, src.x, SCALE);
+ compute_scatterdd (0xFFFF, idx.a, src.a, SCALE, res_ref);
+
+ if (checkVi (res, res_ref, 16))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vscatterdpd\[ \\t\]+\[^\n\]*zmm\[0-9\]\[^\n\]*ymm\[0-9\]\[^\n\]*{%k\[1-7\]}" 2 } } */
+
+#include <immintrin.h>
+
+volatile __m512d src;
+volatile __m256i idx;
+volatile __mmask8 m8;
+double *addr;
+
+void extern
+avx512f_test (void)
+{
+ _mm512_i32scatter_pd (addr, idx, src, 8);
+ _mm512_mask_i32scatter_pd (addr, m8, idx, src, 8);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+#define SCALE 2
+
+static void
+compute_scatterdpd (__mmask8 m8, int *idx, double *src,
+ int scale, double *r)
+{
+ int i;
+
+ for (i = 0; i < 8; i++)
+ {
+ if (m8 & (1 << i))
+ *(double *) (((unsigned char *) r) + idx[i] * scale) = src[i];
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+ union512d src;
+ union256i_d idx;
+ double res[8] = { 0.0 };
+ double res_ref[8] = { 0.0 };
+ __mmask8 m8 = 0xC5;
+
+ for (i = 0; i < 8; i++)
+ {
+ src.a[i] = 2.718281828459045 * (i + 1) * (i + 2);
+
+ /* About to gather in reverse order,
+ divide by 2 to demonstrate scale */
+ idx.a[i] = (64 - (i + 1) * 8) >> 1;
+ }
+
+ _mm512_mask_i32scatter_pd (res, m8, idx.x, src.x, SCALE);
+ compute_scatterdpd (m8, idx.a, src.a, SCALE, res_ref);
+
+ if (checkVd (res, res_ref, 8))
+ abort ();
+
+ _mm512_i32scatter_pd (res, idx.x, src.x, SCALE);
+ compute_scatterdpd (0xFF, idx.a, src.a, SCALE, res_ref);
+
+ if (checkVd (res, res_ref, 8))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vscatterdps\[ \\t\]+\[^\n\]*zmm\[0-9\]\[^\n\]*zmm\[0-9\]\[^\n\]*{%k\[1-7\]}" 2 } } */
+
+#include <immintrin.h>
+
+volatile __m512 src;
+volatile __m512i idx;
+volatile __mmask16 m16;
+float *addr;
+
+void extern
+avx512f_test (void)
+{
+ _mm512_i32scatter_ps (addr, idx, src, 8);
+ _mm512_mask_i32scatter_ps (addr, m16, idx, src, 8);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+#define SCALE 2
+
+static void
+compute_scatterdps (__mmask16 m16, int *idx,
+ float *src, int scale, float *r)
+{
+ int i;
+
+ for (i = 0; i < 16; i++)
+ {
+ if (m16 & (1 << i))
+ *(float *) (((unsigned char *) r) + idx[i] * scale) = src[i];
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+ union512 src;
+ union512i_d idx;
+ float res[16] = { 0.0 };
+ float res_ref[16] = { 0.0 };
+ __mmask16 m16 = 0xBC5D;
+
+ for (i = 0; i < 16; i++)
+ {
+ src.a[i] = 2.718281828459045 * (i + 1) * (i + 2);
+
+ /* About to gather in reverse order,
+ divide by 2 to demonstrate scale */
+ idx.a[i] = (64 - (i + 1) * 4) >> 1;
+ }
+
+ _mm512_mask_i32scatter_ps (res, m16, idx.x, src.x, SCALE);
+ compute_scatterdps (m16, idx.a, src.a, SCALE, res_ref);
+
+ if (checkVf (res, res_ref, 16))
+ abort ();
+
+ _mm512_i32scatter_ps (res, idx.x, src.x, SCALE);
+ compute_scatterdps (0xFFFF, idx.a, src.a, SCALE, res_ref);
+
+ if (checkVf (res, res_ref, 16))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpscatterdq\[ \\t\]+\[^\n\]*zmm\[0-9\]\[^\n\]*ymm\[0-9\]\[^\n\]*{%k\[1-7\]}" 2 } } */
+
+#include <immintrin.h>
+
+volatile __m512i src;
+volatile __m256i idx;
+volatile __mmask8 m8;
+long long *addr;
+
+void extern
+avx512f_test (void)
+{
+ _mm512_i32scatter_epi64 (addr, idx, src, 8);
+ _mm512_mask_i32scatter_epi64 (addr, m8, idx, src, 8);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+#define SCALE 2
+
+static void
+compute_scatterdq (__mmask8 m8, int *idx, long long *src,
+ int scale, long long *r)
+{
+ int i;
+
+ for (i = 0; i < 8; i++)
+ {
+ if (m8 & (1 << i))
+ *(long long *) (((unsigned char *) r) + idx[i] * scale) =
+ src[i];
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+ union256i_d idx;
+ union512i_q src;
+ long long res[8] = { 0 };
+ long long res_ref[8] = { 0 };
+ __mmask8 m8 = 0xC5;
+
+ for (i = 0; i < 8; i++)
+ {
+ src.a[i] = 1983 * (i + 1) * (i + 2);
+
+ /* About to gather in reverse order,
+ divide by 2 to demonstrate scale */
+ idx.a[i] = (64 - (i + 1) * 8) >> 1;
+ }
+
+ _mm512_mask_i32scatter_epi64 (res, m8, idx.x, src.x, SCALE);
+ compute_scatterdq (m8, idx.a, src.a, SCALE, res_ref);
+
+ if (checkVl (res, res_ref, 8))
+ abort ();
+
+ _mm512_i32scatter_epi64 (res, idx.x, src.x, SCALE);
+ compute_scatterdq (0xFF, idx.a, src.a, SCALE, res_ref);
+
+ if (checkVl (res, res_ref, 8))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpgatherqd\[ \\t\]+\[^\n\]*zmm\[0-9\]\[^\n\]*ymm\[0-9\]{%k\[1-7\]}" 2 } } */
+
+#include <immintrin.h>
+
+volatile __m256i x;
+volatile __m512i idx;
+volatile __mmask8 m8;
+int *base;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_i64gather_epi32 (idx, base, 8);
+ x = _mm512_mask_i64gather_epi32 (x, m8, idx, base, 8);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+#define SCALE 2
+
+static void
+compute_gatherqd (int *res, __mmask8 m8, long long *idx,
+ int *src, int scale, int *r)
+{
+ int i;
+
+ for (i = 0; i < 8; i++)
+ {
+ if (m8 & (1 << i))
+ r[i] = *(int *) (((unsigned char *) src) + idx[i] * scale);
+ else
+ r[i] = res[i];
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+ union256i_d res;
+ union512i_q idx;
+ int src[8];
+ int res_ref[8];
+ __mmask8 m8 = 0xC5;
+
+ for (i = 0; i < 8; i++)
+ {
+ src[i] = 1973 * (i + 1) * (i + 2);
+
+ /* About to gather in reverse order,
+ divide by 2 to demonstrate scale */
+ idx.a[i] = (32 - (i + 1) * 4) >> 1;
+ }
+
+ res.x = _mm512_mask_i64gather_epi32 (res.x, m8, idx.x, src, SCALE);
+ compute_gatherqd (res.a, m8, idx.a, src, SCALE, res_ref);
+
+ if (check_union256i_d (res, res_ref))
+ abort ();
+
+ res.x = _mm512_i64gather_epi32 (idx.x, src, SCALE);
+ compute_gatherqd (res.a, 0xFF, idx.a, src, SCALE, res_ref);
+
+ if (check_union256i_d (res, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vgatherqpd\[ \\t\]+\[^\n\]*zmm\[0-9\]\[^\n\]*zmm\[0-9\]{%k\[1-7\]}" 2 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __m512i idx;
+volatile __mmask8 m8;
+double *base;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_i64gather_pd (idx, base, 8);
+ x = _mm512_mask_i64gather_pd (x, m8, idx, base, 8);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+#define SCALE 2
+
+static void
+compute_gatherqpd (double *res, __mmask8 m8, long long *idx,
+ double *src, int scale, double *r)
+{
+ int i;
+
+ for (i = 0; i < 8; i++)
+ {
+ if (m8 & (1 << i))
+ r[i] = *(double *) (((unsigned char *) src) + idx[i] * scale);
+ else
+ r[i] = res[i];
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+ union512d res;
+ union512i_q idx;
+ double src[8];
+ double res_ref[8];
+ __mmask8 m8 = 0xC5;
+
+ res.x = _mm512_setzero_pd();
+
+ for (i = 0; i < 8; i++)
+ {
+ src[i] = 2.718281828459045 * (i + 1) * (i + 2);
+
+ /* About to gather in reverse order,
+ divide by 2 to demonstrate scale */
+ idx.a[i] = (64 - (i + 1) * 8) >> 1;
+ }
+
+ res.x = _mm512_mask_i64gather_pd (res.x, m8, idx.x, src, SCALE);
+ compute_gatherqpd (res.a, m8, idx.a, src, SCALE, res_ref);
+
+ if (check_union512d (res, res_ref))
+ abort ();
+
+ res.x = _mm512_i64gather_pd (idx.x, src, SCALE);
+ compute_gatherqpd (res.a, 0xFF, idx.a, src, SCALE, res_ref);
+
+ if (check_union512d (res, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vgatherqps\[ \\t\]+\[^\n\]*zmm\[0-9\]\[^\n\]*ymm\[0-9\]{%k\[1-7\]}" 2 } } */
+
+#include <immintrin.h>
+
+volatile __m256 x;
+volatile __m512i idx;
+volatile __mmask8 m8;
+float *base;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_i64gather_ps (idx, base, 8);
+ x = _mm512_mask_i64gather_ps (x, m8, idx, base, 8);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+#define SCALE 2
+
+static void
+compute_gatherqps (float *res, __mmask8 m8, long long *idx,
+ float *src, int scale, float *r)
+{
+ int i;
+
+ for (i = 0; i < 8; i++)
+ {
+ if (m8 & (1 << i))
+ r[i] = *(float *) (((unsigned char *) src) + idx[i] * scale);
+ else
+ r[i] = res[i];
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+ union256 res;
+ union512i_q idx;
+ float src[8];
+ float res_ref[8];
+ __mmask8 m8 = 0xC5;
+
+ res.x = _mm256_setzero_ps();
+
+ for (i = 0; i < 8; i++)
+ {
+ src[i] = 2.718281828459045 * (i + 1) * (i + 2);
+
+ /* About to gather in reverse order,
+ divide by 2 to demonstrate scale */
+ idx.a[i] = (32 - (i + 1) * 4) >> 1;
+ }
+
+ res.x = _mm512_mask_i64gather_ps (res.x, m8, idx.x, src, SCALE);
+ compute_gatherqps (res.a, m8, idx.a, src, SCALE, res_ref);
+
+ if (check_union256 (res, res_ref))
+ abort ();
+
+ res.x = _mm512_i64gather_ps (idx.x, src, SCALE);
+ compute_gatherqps (res.a, 0xFF, idx.a, src, SCALE, res_ref);
+
+ if (check_union256 (res, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpgatherqq\[ \\t\]+\[^\n\]*zmm\[0-9\]\[^\n\]*zmm\[0-9\]{%k\[1-7\]}" 2 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x, idx;
+volatile __mmask8 m8;
+long long *base;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_i64gather_epi64 (idx, base, 8);
+ x = _mm512_mask_i64gather_epi64 (x, m8, idx, base, 8);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+#define SCALE 2
+
+static void
+compute_gatherqq (long long *res, __mmask8 m8, long long *idx,
+ long long *src, int scale, long long *r)
+{
+ int i;
+
+ for (i = 0; i < 8; i++)
+ {
+ if (m8 & (1 << i))
+ r[i] = *(long long *)
+ (((unsigned char *) src) + idx[i] * scale);
+ else
+ r[i] = res[i];
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+ union512i_q idx, res;
+ long long src[8];
+ long long res_ref[8];
+ __mmask8 m8 = 0xC5;
+
+ for (i = 0; i < 8; i++)
+ {
+ src[i] = 1983 * (i + 1) * (i + 2);
+
+ /* About to gather in reverse order,
+ divide by 2 to demonstrate scale */
+ idx.a[i] = (64 - (i + 1) * 8) >> 1;
+ }
+
+ res.x = _mm512_mask_i64gather_epi64 (res.x, m8, idx.x, src, SCALE);
+ compute_gatherqq (res.a, m8, idx.a, src, SCALE, res_ref);
+
+ if (check_union512i_q (res, res_ref))
+ abort ();
+
+ res.x = _mm512_i64gather_epi64 (idx.x, src, SCALE);
+ compute_gatherqq (res.a, 0xFF, idx.a, src, SCALE, res_ref);
+
+ if (check_union512i_q (res, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpscatterqd\[ \\t\]+\[^\n\]*ymm\[0-9\]\[^\n\]*zmm\[0-9\]\[^\n\]*{%k\[1-7\]}" 2 } } */
+
+#include <immintrin.h>
+
+volatile __m256i src;
+volatile __m512i idx;
+volatile __mmask8 m8;
+int *addr;
+
+void extern
+avx512f_test (void)
+{
+ _mm512_i64scatter_epi32 (addr, idx, src, 8);
+ _mm512_mask_i64scatter_epi32 (addr, m8, idx, src, 8);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+#define SCALE 2
+
+static void
+compute_scatterqd (__mmask8 m8, long long *idx,
+ int *src, int scale, int *r)
+{
+ int i;
+
+ for (i = 0; i < 8; i++)
+ {
+ if (m8 & (1 << i))
+ *(int *) (((unsigned char *) r) + idx[i] * scale) = src[i];
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+ union256i_d src;
+ union512i_q idx;
+ int res[8] = { 0 };
+ int res_ref[8] = { 0 };
+ __mmask8 m8 = 0xC5;
+
+ for (i = 0; i < 8; i++)
+ {
+ src.a[i] = 1973 * (i + 1) * (i + 2);
+
+ /* About to gather in reverse order,
+ divide by 2 to demonstrate scale */
+ idx.a[i] = (32 - (i + 1) * 4) >> 1;
+ }
+
+ _mm512_mask_i64scatter_epi32 (res, m8, idx.x, src.x, SCALE);
+ compute_scatterqd (m8, idx.a, src.a, SCALE, res_ref);
+
+ if (checkVi (res, res_ref, 8))
+ abort ();
+
+ _mm512_i64scatter_epi32 (res, idx.x, src.x, SCALE);
+ compute_scatterqd (0xFF, idx.a, src.a, SCALE, res_ref);
+
+ if (checkVi (res, res_ref, 8))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vscatterqpd\[ \\t\]+\[^\n\]*zmm\[0-9\]\[^\n\]*zmm\[0-9\]\[^\n\]*{%k\[1-7\]}" 2 } } */
+
+#include <immintrin.h>
+
+volatile __m512d src;
+volatile __m512i idx;
+volatile __mmask8 m8;
+double *addr;
+
+void extern
+avx512f_test (void)
+{
+ _mm512_i64scatter_pd (addr, idx, src, 8);
+ _mm512_mask_i64scatter_pd (addr, m8, idx, src, 8);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+#define SCALE 2
+
+static void
+compute_scatterqpd (__mmask8 m8, long long *idx, double *src,
+ int scale, double *r)
+{
+ int i;
+
+ for (i = 0; i < 8; i++)
+ {
+ if (m8 & (1 << i))
+ *(double *) (((unsigned char *) r) + idx[i] * scale) = src[i];
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+ union512d src;
+ union512i_q idx;
+ double res[8] = { 0.0 };
+ double res_ref[8] = { 0.0 };
+ __mmask8 m8 = 0xC5;
+
+ for (i = 0; i < 8; i++)
+ {
+ src.a[i] = 2.718281828459045 * (i + 1) * (i + 2);
+
+ /* About to gather in reverse order,
+ divide by 2 to demonstrate scale */
+ idx.a[i] = (64 - (i + 1) * 8) >> 1;
+ }
+
+ _mm512_mask_i64scatter_pd (res, m8, idx.x, src.x, SCALE);
+ compute_scatterqpd (m8, idx.a, src.a, SCALE, res_ref);
+
+ if (checkVd (res, res_ref, 8))
+ abort ();
+
+ _mm512_i64scatter_pd (res, idx.x, src.x, SCALE);
+ compute_scatterqpd (0xFF, idx.a, src.a, SCALE, res_ref);
+
+ if (checkVd (res, res_ref, 8))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vscatterqps\[ \\t\]+\[^\n\]*ymm\[0-9\]\[^\n\]*zmm\[0-9\]\[^\n\]*{%k\[1-7\]}" 2 } } */
+
+#include <immintrin.h>
+
+volatile __m256 src;
+volatile __m512i idx;
+volatile __mmask8 m8;
+float *addr;
+
+void extern
+avx512f_test (void)
+{
+ _mm512_i64scatter_ps (addr, idx, src, 8);
+ _mm512_mask_i64scatter_ps (addr, m8, idx, src, 8);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+#define SCALE 2
+
+static void
+compute_scatterqps (__mmask8 m8, long long *idx,
+ float *src, int scale, float *r)
+{
+ int i;
+
+ for (i = 0; i < 8; i++)
+ {
+ if (m8 & (1 << i))
+ *(float *) (((unsigned char *) r) + idx[i] * scale) = src[i];
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+ union256 src;
+ union512i_q idx;
+ float res[8] = { 0.0 };
+ float res_ref[8] = { 0.0 };
+ __mmask8 m8 = 0xC5;
+
+ for (i = 0; i < 8; i++)
+ {
+ src.a[i] = 2.718281828459045 * (i + 1) * (i + 2);
+
+ /* About to gather in reverse order,
+ divide by 2 to demonstrate scale */
+ idx.a[i] = (32 - (i + 1) * 4) >> 1;
+ }
+
+ _mm512_mask_i64scatter_ps (res, m8, idx.x, src.x, SCALE);
+ compute_scatterqps (m8, idx.a, src.a, SCALE, res_ref);
+
+ if (checkVf (res, res_ref, 8))
+ abort ();
+
+ _mm512_i64scatter_ps (res, idx.x, src.x, SCALE);
+ compute_scatterqps (0xFF, idx.a, src.a, SCALE, res_ref);
+
+ if (checkVf (res, res_ref, 8))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpscatterqq\[ \\t\]+\[^\n\]*zmm\[0-9\]\[^\n\]*zmm\[0-9\]\[^\n\]*{%k\[1-7\]}" 2 } } */
+
+#include <immintrin.h>
+
+volatile __m512i src, idx;
+volatile __mmask8 m8;
+long long *addr;
+
+void extern
+avx512f_test (void)
+{
+ _mm512_i64scatter_epi64 (addr, idx, src, 8);
+ _mm512_mask_i64scatter_epi64 (addr, m8, idx, src, 8);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+#define SCALE 2
+
+static void
+compute_scatterqq (__mmask8 m8, long long *idx, long long *src,
+ int scale, long long *r)
+{
+ int i;
+
+ for (i = 0; i < 8; i++)
+ {
+ if (m8 & (1 << i))
+ *(long long *) (((unsigned char *) r) + idx[i] * scale) =
+ src[i];
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+ union512i_q src, idx;
+ long long res[8] = { 0 };
+ long long res_ref[8] = { 0 };
+ __mmask8 m8 = 0xC5;
+
+ for (i = 0; i < 8; i++)
+ {
+ src.a[i] = 1983 * (i + 1) * (i + 2);
+
+ /* About to gather in reverse order,
+ divide by 2 to demonstrate scale */
+ idx.a[i] = (64 - (i + 1) * 8) >> 1;
+ }
+
+ _mm512_mask_i64scatter_epi64 (res, m8, idx.x, src.x, SCALE);
+ compute_scatterqq (m8, idx.a, src.a, SCALE, res_ref);
+
+ if (checkVl (res, res_ref, 8))
+ abort ();
+
+ _mm512_i64scatter_epi64 (res, idx.x, src.x, SCALE);
+ compute_scatterqq (0xFF, idx.a, src.a, SCALE, res_ref);
+
+ if (checkVl (res, res_ref, 8))
+ abort ();
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static void
+init_vpadd_mask (int* dst, int *src1, int *src2, int seed)
+{
+ int i;
+
+ for (i = 0; i < 16; i++)
+ {
+ dst[i] = -1;
+ src1[i] = seed * 2 * i + 1;
+ src2[i] = seed * 2 * i;
+ }
+}
+
+static inline void
+calc_vpadd_mask_zeroed (int *dst, __mmask16 m, int *src1, int *src2)
+{
+ int i;
+
+ for (i = 0; i < 16; i++)
+ {
+ if (m & (1 << i))
+ dst[i] = src1[i] + src2[i];
+ else
+ dst[i] = 0;
+ }
+}
+
+void static
+avx512f_test (void)
+{
+ /* Checking mask arithmetic instruction */
+
+ __mmask16 msk_dst, msk_src1, msk_src2, msk_dst_ref;
+
+ msk_src1 = 0x0FFB;
+ msk_src2 = 0x0F0F;
+
+ asm ("kandw\t%2, %1, %0"
+ : "=Yk" (msk_dst)
+ : "Yk" (msk_src1), "Yk" (msk_src2));
+
+ msk_dst_ref = _mm512_kand (msk_src1, msk_src2);
+ if (msk_dst != msk_dst_ref)
+ abort ();
+
+
+ /* Checking zero-masked vector instruction */
+ union512i_d dst, src1, src2;
+ int dst_ref[16];
+
+ init_vpadd_mask (dst.a, src1.a, src2.a, 1);
+ init_vpadd_mask (dst_ref, src1.a, src2.a, 1);
+
+ asm ("vpaddd\t%2, %1, %0 %{%3%}%{z%}"
+ : "=x" (dst.x)
+ : "x" (src1.x), "x" (src2.x), "k" (msk_dst));
+
+ calc_vpadd_mask_zeroed (dst_ref, msk_dst, src1.a, src2.a);
+
+ if (check_union512i_d (dst, dst_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "kandnw\[ \\t\]+\[^\n\]*%k\[1-7\]" 1 } } */
+
+#include <immintrin.h>
+
+void
+avx512f_test ()
+{
+ __mmask16 k1, k2, k3;
+ volatile __m512 x;
+
+ __asm__( "kmovw %1, %0" : "=k" (k1) : "r" (1) );
+ __asm__( "kmovw %1, %0" : "=k" (k2) : "r" (2) );
+
+ k3 = _mm512_kandn (k1, k2);
+ x = _mm512_mask_add_ps (x, k3, x, x);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "kandw\[ \\t\]+\[^\n\]*%k\[1-7\]" 1 } } */
+
+#include <immintrin.h>
+
+void
+avx512f_test ()
+{
+ __mmask16 k1, k2, k3;
+ volatile __m512 x;
+
+ __asm__( "kmovw %1, %0" : "=k" (k1) : "r" (1) );
+ __asm__( "kmovw %1, %0" : "=k" (k2) : "r" (2) );
+
+ k3 = _mm512_kand (k1, k2);
+ x = _mm512_mask_add_ps (x, k3, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+void
+avx512f_test (void)
+{
+ __mmask16 dst, src1, src2, dst_ref;
+ volatile __m512 x;
+
+ __asm__( "kmovw %1, %0" : "=k" (src1) : "r" (0x0FFF) );
+ __asm__( "kmovw %1, %0" : "=k" (src2) : "r" (0x0F0F) );
+
+ dst = _mm512_kand (src1, src2);
+ x = _mm512_mask_add_ps (x, dst, x, x);
+ dst_ref = src1 & src2;
+ if (dst != dst_ref)
+ abort ();
+
+ dst = _mm512_kandn (src1, src2);
+ x = _mm512_mask_add_ps (x, dst, x, x);
+ dst_ref = ~src1 & src2;
+ if (dst != dst_ref)
+ abort ();
+
+ dst = _mm512_kor (src1, src2);
+ x = _mm512_mask_add_ps (x, dst, x, x);
+ dst_ref = src1 | src2;
+ if (dst != dst_ref)
+ abort ();
+
+ dst = _mm512_kxnor (src1, src2);
+ x = _mm512_mask_add_ps (x, dst, x, x);
+ dst_ref = ~(src1 ^ src2);
+ if (dst != dst_ref)
+ abort ();
+
+ dst = _mm512_kxor (src1, src2);
+ x = _mm512_mask_add_ps (x, dst, x, x);
+ dst_ref = src1 ^ src2;
+ if (dst != dst_ref)
+ abort ();
+
+ dst = _mm512_knot (src1);
+ x = _mm512_mask_add_ps (x, dst, x, x);
+ dst_ref = ~src1;
+ if (dst != dst_ref)
+ abort ();
+
+ dst = _mm512_kunpackb (src1, src2);
+ x = _mm512_mask_add_ps (x, dst, x, x);
+ dst_ref = 0xFF0F;
+
+ if (dst != dst_ref)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "knotw\[ \\t\]+\[^\n\]*%k\[1-7\]" 1 } } */
+
+#include <immintrin.h>
+
+void
+avx512f_test ()
+{
+ __mmask16 k1, k2;
+ volatile __m512 x;
+
+ __asm__( "kmovw %1, %0" : "=k" (k1) : "r" (45) );
+
+ k2 = _mm512_knot (k1);
+
+ x = _mm512_mask_add_ps (x, k1, x, x);
+ x = _mm512_mask_add_ps (x, k2, x, x);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O0 -mavx512f" } */
+/* { dg-final { scan-assembler-times "kortestw\[ \\t\]+\[^\n\]*%k\[0-7\]" 4 } } */
+
+#include <immintrin.h>
+
+void
+avx512f_test () {
+ volatile __mmask16 k1;
+ __mmask16 k2;
+ volatile __mmask8 k3;
+ __mmask8 k4;
+
+ volatile short r;
+
+ /* Check that appropriate insn sequence is generated at -O0. */
+ r = _mm512_kortestc (k1, k2);
+ r = _mm512_kortestz (k1, k2);
+
+ r = _mm512_kortestc (k3, k4);
+ r = _mm512_kortestz (k3, k4);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+void
+avx512f_test () {
+ volatile __mmask16 k1;
+ __mmask16 k2;
+ volatile short r = 0;
+
+ /* Test kortestc. */
+ __asm__( "kmovw %1, %0" : "=k" (k1) : "r" (0) );
+ __asm__( "kmovw %1, %0" : "=k" (k2) : "r" (45) );
+
+ r += _mm512_kortestc (k1, k2);
+
+ __asm__( "kmovw %1, %0" : "=k" (k1) : "r" (0) );
+ __asm__( "kmovw %1, %0" : "=k" (k2) : "r" (0) );
+
+ r += _mm512_kortestc (k1, k2);
+ if (r)
+ abort ();
+
+ __asm__( "kmovw %1, %0" : "=k" (k1) : "r" (-1) );
+ __asm__( "kmovw %1, %0" : "=k" (k2) : "r" (0) );
+
+ r += _mm512_kortestc (k1, k2);
+ if (!r)
+ abort ();
+
+ r = 0;
+ /* Test kortestz. */
+ __asm__( "kmovw %1, %0" : "=k" (k1) : "r" (0) );
+ __asm__( "kmovw %1, %0" : "=k" (k2) : "r" (45) );
+
+ r += _mm512_kortestz (k1, k2);
+
+ __asm__( "kmovw %1, %0" : "=k" (k1) : "r" (-1) );
+ __asm__( "kmovw %1, %0" : "=k" (k2) : "r" (0) );
+
+ r += _mm512_kortestz (k1, k2);
+ if (r)
+ abort ();
+
+ __asm__( "kmovw %1, %0" : "=k" (k1) : "r" (0) );
+ __asm__( "kmovw %1, %0" : "=k" (k2) : "r" (0) );
+
+ r += _mm512_kortestz (k1, k2);
+ if (!r)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "korw\[ \\t\]+\[^\n\]*%k\[1-7\]" 1 } } */
+
+#include <immintrin.h>
+
+void
+avx512f_test ()
+{
+ __mmask16 k1, k2, k3;
+ volatile __m512 x;
+
+ __asm__( "kmovw %1, %0" : "=k" (k1) : "r" (1) );
+ __asm__( "kmovw %1, %0" : "=k" (k2) : "r" (2) );
+
+ k3 = _mm512_kor (k1, k2);
+ x = _mm512_mask_add_ps (x, k3, x, x);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "kunpckbw\[ \\t\]+\[^\n\]*%k\[1-7\]" 1 } } */
+
+#include <immintrin.h>
+
+void
+avx512f_test () {
+ __mmask16 k1, k2, k3;
+ volatile __m512 x;
+
+ __asm__( "kmovw %1, %0" : "=k" (k1) : "r" (1) );
+ __asm__( "kmovw %1, %0" : "=k" (k2) : "r" (2) );
+
+ k3 = _mm512_kunpackb (k1, k2);
+ x = _mm512_mask_add_ps (x, k3, x, x);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "kxnorw\[ \\t\]+\[^\n\]*%k\[1-7\]" 1 } } */
+
+#include <immintrin.h>
+
+void
+avx512f_test ()
+{
+ __mmask16 k1, k2, k3;
+ volatile __m512 x;
+
+ __asm__( "kmovw %1, %0" : "=k" (k1) : "r" (1) );
+ __asm__( "kmovw %1, %0" : "=k" (k2) : "r" (2) );
+
+ k3 = _mm512_kxnor (k1, k2);
+ x = _mm512_mask_add_ps (x, k3, x, x);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "kxorw\[ \\t\]+\[^\n\]*%k\[1-7\]" 1 } } */
+
+#include <immintrin.h>
+
+void
+avx512f_test ()
+{
+ __mmask16 k1, k2, k3;
+ volatile __m512 x;
+
+ __asm__( "kmovw %1, %0" : "=k" (k1) : "r" (1) );
+ __asm__( "kmovw %1, %0" : "=k" (k2) : "r" (2) );
+
+ k3 = _mm512_kxor (k1, k2);
+ x = _mm512_mask_add_ps (x, k3, x, x);
+}
--- /dev/null
+/* Type of mask. */
+#if SIZE <= 8
+#define MASK_TYPE __mmask8
+#define MASK_VALUE 0xB9
+#elif SIZE <= 16
+#define MASK_TYPE __mmask16
+#define MASK_VALUE 0xA6BA
+#endif
--- /dev/null
+/* Check if the OS supports executing AVX512F instructions. */
+
+static int
+avx512f_os_support (void)
+{
+ unsigned int eax, edx;
+
+ __asm__ ("xgetbv" : "=a" (eax), "=d" (edx) : "c" (0));
+ return (eax & 230) == 230;
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O0 -mavx512f" } */
+
+#include <x86intrin.h>
+
+int
+test_rounding (__m128d x, int r)
+{
+ return _mm_cvt_roundsd_i32 (x, r); /* { dg-error "incorrect rounding operand." } */
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static __m512
+__attribute__ ((noinline))
+foo (float *v)
+{
+ return _mm512_set_ps (v[15], v[14], v[13], v[12],
+ v[11], v[10], v[9], v[8],
+ v[7], v[6], v[5], v[4],
+ v[3], v[2], v[1], v[0]);
+}
+
+static __m512
+__attribute__ ((noinline))
+foo_r (float *v)
+{
+ return _mm512_setr_ps (v[0], v[1], v[2], v[3],
+ v[4], v[5], v[6], v[7],
+ v[8], v[9], v[10], v[11],
+ v[12], v[13], v[14], v[15]);
+}
+
+static void
+avx512f_test (void)
+{
+ float v[16] = { -3.3, 2.6, 1.48, 9.104, -23.9, 17, -13.48, 4,
+ 69.78, 0.33, 81, 0.4, -8.9, -173.37, 0.8, 68 };
+ union512 res;
+
+ res.x = foo (v);
+
+ if (check_union512 (res, v))
+ abort ();
+
+ res.x = _mm512_setzero_ps ();
+
+ res.x = foo_r (v);
+
+ if (check_union512 (res, v))
+ abort ();
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static __m512
+__attribute__ ((noinline))
+foo (float x1, float x2, float x3, float x4,
+ float x5, float x6, float x7, float x8,
+ float x9, float x10, float x11, float x12,
+ float x13, float x14, float x15, float x16)
+{
+ return _mm512_set_ps (x1, x2, x3, x4, x5, x6, x7, x8,
+ x9, x10, x11, x12, x13, x14, x15, x16);
+}
+
+static __m512
+__attribute__ ((noinline))
+foo_r (float x1, float x2, float x3, float x4,
+ float x5, float x6, float x7, float x8,
+ float x9, float x10, float x11, float x12,
+ float x13, float x14, float x15, float x16)
+{
+ return _mm512_setr_ps (x16, x15, x14, x13, x12, x11, x10, x9,
+ x8, x7, x6, x5, x4, x3, x2, x1);
+}
+
+static void
+avx512f_test (void)
+{
+ float v[16] = { -3.3, 2.6, 1.48, 9.104, -23.9, 17, -13.48, 4,
+ 69.78, 0.33, 81, 0.4, -8.9, -173.37, 0.8, 68 };
+ union512 res;
+
+ res.x = foo (v[15], v[14], v[13], v[12], v[11], v[10], v[9], v[8],
+ v[7], v[6], v[5], v[4], v[3], v[2], v[1], v[0]);
+
+ if (check_union512 (res, v))
+ abort ();
+
+ res.x = _mm512_setzero_ps ();
+
+ res.x = foo_r (v[15], v[14], v[13], v[12], v[11], v[10], v[9], v[8],
+ v[7], v[6], v[5], v[4], v[3], v[2], v[1], v[0]);
+
+ if (check_union512 (res, v))
+ abort ();
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static __m512
+__attribute__ ((noinline))
+foo (float x)
+{
+ return _mm512_set_ps (x, x, x, x, x, x, x, x,
+ x, x, x, x, x, x, x, x);
+}
+
+static __m512
+__attribute__ ((noinline))
+foo_r (float x)
+{
+ return _mm512_setr_ps (x, x, x, x, x, x, x, x,
+ x, x, x, x, x, x, x, x);
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+ float e = 34.5;
+ float v[16];
+ union512 res;
+
+ for (i = 0; i < 16; i++)
+ v[i] = e;
+
+ res.x = foo (e);
+
+ if (check_union512 (res, v))
+ abort ();
+
+ res.x = _mm512_setzero_ps ();
+
+ res.x = foo_r (e);
+
+ if (check_union512 (res, v))
+ abort ();
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static __m512
+__attribute__ ((noinline))
+foo (float x, int i)
+{
+ switch (i)
+ {
+ case 15:
+ return _mm512_set_ps (x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 14:
+ return _mm512_set_ps (0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 13:
+ return _mm512_set_ps (0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 12:
+ return _mm512_set_ps (0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 11:
+ return _mm512_set_ps (0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 10:
+ return _mm512_set_ps (0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 9:
+ return _mm512_set_ps (0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 8:
+ return _mm512_set_ps (0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 7:
+ return _mm512_set_ps (0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0);
+ case 6:
+ return _mm512_set_ps (0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0);
+ case 5:
+ return _mm512_set_ps (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0);
+ case 4:
+ return _mm512_set_ps (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0);
+ case 3:
+ return _mm512_set_ps (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0);
+ case 2:
+ return _mm512_set_ps (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0);
+ case 1:
+ return _mm512_set_ps (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0);
+ case 0:
+ return _mm512_set_ps (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x);
+ default:
+ abort ();
+ }
+}
+
+static __m512
+__attribute__ ((noinline))
+foo_r (float x, int i)
+{
+ switch (i)
+ {
+ case 0:
+ return _mm512_setr_ps (x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 1:
+ return _mm512_setr_ps (0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 2:
+ return _mm512_setr_ps (0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 3:
+ return _mm512_setr_ps (0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 4:
+ return _mm512_setr_ps (0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 5:
+ return _mm512_setr_ps (0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 6:
+ return _mm512_setr_ps (0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 7:
+ return _mm512_setr_ps (0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 8:
+ return _mm512_setr_ps (0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0);
+ case 9:
+ return _mm512_setr_ps (0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0);
+ case 10:
+ return _mm512_setr_ps (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0);
+ case 11:
+ return _mm512_setr_ps (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0);
+ case 12:
+ return _mm512_setr_ps (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0);
+ case 13:
+ return _mm512_setr_ps (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0);
+ case 14:
+ return _mm512_setr_ps (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0);
+ case 15:
+ return _mm512_setr_ps (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x);
+ default:
+ abort ();
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ float e = -3.234;
+ float v[16];
+ union512 res;
+ int i, j;
+
+ for (i = 0; i < 16; i++)
+ {
+ for (j = 0; j < 16; j++)
+ v[j] = 0;
+ v[i] = e;
+
+ res.x = foo (e, i);
+
+ if (check_union512 (res, v))
+ abort ();
+
+ res.x = _mm512_setzero_ps ();
+
+ res.x = foo_r (e, i);
+
+ if (check_union512 (res, v))
+ abort ();
+ }
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static __m512
+__attribute__ ((noinline))
+foo (float x, int i)
+{
+ switch (i)
+ {
+ case 15:
+ return _mm512_set_ps (x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 14:
+ return _mm512_set_ps (1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 13:
+ return _mm512_set_ps (1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 12:
+ return _mm512_set_ps (1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 11:
+ return _mm512_set_ps (1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 10:
+ return _mm512_set_ps (1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 9:
+ return _mm512_set_ps (1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 8:
+ return _mm512_set_ps (1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 7:
+ return _mm512_set_ps (1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1);
+ case 6:
+ return _mm512_set_ps (1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1);
+ case 5:
+ return _mm512_set_ps (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1);
+ case 4:
+ return _mm512_set_ps (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1);
+ case 3:
+ return _mm512_set_ps (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1);
+ case 2:
+ return _mm512_set_ps (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1);
+ case 1:
+ return _mm512_set_ps (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1);
+ case 0:
+ return _mm512_set_ps (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x);
+ default:
+ abort ();
+ }
+}
+
+static __m512
+__attribute__ ((noinline))
+foo_r (float x, int i)
+{
+ switch (i)
+ {
+ case 0:
+ return _mm512_setr_ps (x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 1:
+ return _mm512_setr_ps (1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 2:
+ return _mm512_setr_ps (1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 3:
+ return _mm512_setr_ps (1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 4:
+ return _mm512_setr_ps (1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 5:
+ return _mm512_setr_ps (1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 6:
+ return _mm512_setr_ps (1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 7:
+ return _mm512_setr_ps (1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 8:
+ return _mm512_setr_ps (1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1);
+ case 9:
+ return _mm512_setr_ps (1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1);
+ case 10:
+ return _mm512_setr_ps (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1);
+ case 11:
+ return _mm512_setr_ps (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1);
+ case 12:
+ return _mm512_setr_ps (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1);
+ case 13:
+ return _mm512_setr_ps (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1);
+ case 14:
+ return _mm512_setr_ps (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1);
+ case 15:
+ return _mm512_setr_ps (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x);
+ default:
+ abort ();
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ float e = -3.234;
+ float v[16];
+ union512 res;
+ int i, j;
+
+ for (i = 0; i < 16; i++)
+ {
+ for (j = 0; j < 16; j++)
+ v[j] = 1;
+ v[i] = e;
+
+ res.x = foo (e, i);
+
+ if (check_union512 (res, v))
+ abort ();
+
+ res.x = _mm512_setzero_ps ();
+
+ res.x = foo_r (e, i);
+
+ if (check_union512 (res, v))
+ abort ();
+ }
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static __m512i
+__attribute__ ((noinline))
+foo (int *v)
+{
+ return _mm512_set_epi32 (v[15], v[14], v[13], v[12],
+ v[11], v[10], v[9], v[8],
+ v[7], v[6], v[5], v[4],
+ v[3], v[2], v[1], v[0]);
+}
+
+static __m512i
+__attribute__ ((noinline))
+foo_r (int *v)
+{
+ return _mm512_setr_epi32 (v[0], v[1], v[2], v[3],
+ v[4], v[5], v[6], v[7],
+ v[8], v[9], v[10], v[11],
+ v[12], v[13], v[14], v[15]);
+}
+
+static void
+avx512f_test (void)
+{
+ int v[16] = { 19832468, 2134, 6576856, 6678,
+ 8723467, 54646, 234566, 12314,
+ 786784, 77575, 645245, 234555,
+ 9487733, 411244, 12344, 86533 };
+ union512i_d res;
+
+ res.x = foo (v);
+
+ if (check_union512i_d (res, v))
+ abort ();
+
+ res.x = _mm512_setzero_si512 ();
+
+ res.x = foo_r (v);
+
+ if (check_union512i_d (res, v))
+ abort ();
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static __m512i
+__attribute__ ((noinline))
+foo (int x1, int x2, int x3, int x4,
+ int x5, int x6, int x7, int x8,
+ int x9, int x10, int x11, int x12,
+ int x13, int x14, int x15, int x16)
+{
+ return _mm512_set_epi32 (x1, x2, x3, x4, x5, x6, x7, x8,
+ x9, x10, x11, x12, x13, x14, x15, x16);
+}
+
+static __m512i
+__attribute__ ((noinline))
+foo_r (int x1, int x2, int x3, int x4,
+ int x5, int x6, int x7, int x8,
+ int x9, int x10, int x11, int x12,
+ int x13, int x14, int x15, int x16)
+{
+ return _mm512_setr_epi32 (x16, x15, x14, x13, x12, x11, x10, x9,
+ x8, x7, x6, x5, x4, x3, x2, x1);
+}
+
+static void
+avx512f_test (void)
+{
+ int v[16] = { -3, -453, 2, -231, 1, -111, 9, -145,
+ 23, 671, -173, 166, -13, 714, 69, 123 };
+ union512i_d res;
+
+ res.x = foo (v[15], v[14], v[13], v[12], v[11], v[10], v[9], v[8],
+ v[7], v[6], v[5], v[4], v[3], v[2], v[1], v[0]);
+
+ if (check_union512i_d (res, v))
+ abort ();
+
+ res.x = _mm512_setzero_si512 ();
+
+ res.x = foo_r (v[15], v[14], v[13], v[12], v[11], v[10], v[9], v[8],
+ v[7], v[6], v[5], v[4], v[3], v[2], v[1], v[0]);
+
+ if (check_union512i_d (res, v))
+ abort ();
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static __m512i
+__attribute__ ((noinline))
+foo (int x)
+{
+ return _mm512_set_epi32 (x, x, x, x, x, x, x, x,
+ x, x, x, x, x, x, x, x);
+}
+
+static __m512i
+__attribute__ ((noinline))
+foo_r (int x)
+{
+ return _mm512_setr_epi32 (x, x, x, x, x, x, x, x,
+ x, x, x, x, x, x, x, x);
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+ int e = 0xabadbeef;
+ int v[16];
+ union512i_d res;
+
+ for (i = 0; i < 16; i++)
+ v[i] = e;
+
+ res.x = foo (e);
+
+ if (check_union512i_d (res, v))
+ abort ();
+
+ res.x = _mm512_setzero_si512 ();
+
+ res.x = foo_r (e);
+
+ if (check_union512i_d (res, v))
+ abort ();
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static __m512i
+__attribute__ ((noinline))
+foo (int x, int i)
+{
+ switch (i)
+ {
+ case 15:
+ return _mm512_set_epi32 (x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 14:
+ return _mm512_set_epi32 (0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 13:
+ return _mm512_set_epi32 (0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 12:
+ return _mm512_set_epi32 (0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 11:
+ return _mm512_set_epi32 (0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 10:
+ return _mm512_set_epi32 (0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 9:
+ return _mm512_set_epi32 (0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 8:
+ return _mm512_set_epi32 (0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 7:
+ return _mm512_set_epi32 (0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0);
+ case 6:
+ return _mm512_set_epi32 (0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0);
+ case 5:
+ return _mm512_set_epi32 (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0);
+ case 4:
+ return _mm512_set_epi32 (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0);
+ case 3:
+ return _mm512_set_epi32 (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0);
+ case 2:
+ return _mm512_set_epi32 (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0);
+ case 1:
+ return _mm512_set_epi32 (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0);
+ case 0:
+ return _mm512_set_epi32 (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x);
+ default:
+ abort ();
+ }
+}
+
+static __m512i
+__attribute__ ((noinline))
+foo_r (int x, int i)
+{
+ switch (i)
+ {
+ case 0:
+ return _mm512_setr_epi32 (x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 1:
+ return _mm512_setr_epi32 (0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 2:
+ return _mm512_setr_epi32 (0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 3:
+ return _mm512_setr_epi32 (0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 4:
+ return _mm512_setr_epi32 (0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 5:
+ return _mm512_setr_epi32 (0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 6:
+ return _mm512_setr_epi32 (0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 7:
+ return _mm512_setr_epi32 (0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0, 0);
+ case 8:
+ return _mm512_setr_epi32 (0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0, 0);
+ case 9:
+ return _mm512_setr_epi32 (0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0, 0);
+ case 10:
+ return _mm512_setr_epi32 (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0, 0);
+ case 11:
+ return _mm512_setr_epi32 (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0, 0);
+ case 12:
+ return _mm512_setr_epi32 (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0, 0);
+ case 13:
+ return _mm512_setr_epi32 (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0, 0);
+ case 14:
+ return _mm512_setr_epi32 (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x, 0);
+ case 15:
+ return _mm512_setr_epi32 (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, x);
+ default:
+ abort ();
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ int e = 0xabadbeef;
+ int v[16];
+ union512i_d res;
+ int i, j;
+
+ for (i = 0; i < 16; i++)
+ {
+ for (j = 0; j < 16; j++)
+ v[j] = 0;
+ v[i] = e;
+
+ res.x = foo (e, i);
+
+ if (check_union512i_d (res, v))
+ abort ();
+
+ res.x = _mm512_setzero_si512 ();
+
+ res.x = foo_r (e, i);
+
+ if (check_union512i_d (res, v))
+ abort ();
+ }
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static __m512i
+__attribute__ ((noinline))
+foo (int x, int i)
+{
+ switch (i)
+ {
+ case 15:
+ return _mm512_set_epi32 (x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 14:
+ return _mm512_set_epi32 (1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 13:
+ return _mm512_set_epi32 (1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 12:
+ return _mm512_set_epi32 (1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 11:
+ return _mm512_set_epi32 (1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 10:
+ return _mm512_set_epi32 (1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 9:
+ return _mm512_set_epi32 (1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 8:
+ return _mm512_set_epi32 (1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 7:
+ return _mm512_set_epi32 (1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1);
+ case 6:
+ return _mm512_set_epi32 (1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1);
+ case 5:
+ return _mm512_set_epi32 (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1);
+ case 4:
+ return _mm512_set_epi32 (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1);
+ case 3:
+ return _mm512_set_epi32 (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1);
+ case 2:
+ return _mm512_set_epi32 (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1);
+ case 1:
+ return _mm512_set_epi32 (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1);
+ case 0:
+ return _mm512_set_epi32 (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x);
+ default:
+ abort ();
+ }
+}
+
+static __m512i
+__attribute__ ((noinline))
+foo_r (int x, int i)
+{
+ switch (i)
+ {
+ case 0:
+ return _mm512_setr_epi32 (x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 1:
+ return _mm512_setr_epi32 (1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 2:
+ return _mm512_setr_epi32 (1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 3:
+ return _mm512_setr_epi32 (1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 4:
+ return _mm512_setr_epi32 (1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 5:
+ return _mm512_setr_epi32 (1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 6:
+ return _mm512_setr_epi32 (1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 7:
+ return _mm512_setr_epi32 (1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1, 1);
+ case 8:
+ return _mm512_setr_epi32 (1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1, 1);
+ case 9:
+ return _mm512_setr_epi32 (1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1, 1);
+ case 10:
+ return _mm512_setr_epi32 (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1, 1);
+ case 11:
+ return _mm512_setr_epi32 (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1, 1);
+ case 12:
+ return _mm512_setr_epi32 (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1, 1);
+ case 13:
+ return _mm512_setr_epi32 (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1, 1);
+ case 14:
+ return _mm512_setr_epi32 (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x, 1);
+ case 15:
+ return _mm512_setr_epi32 (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, x);
+ default:
+ abort ();
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ int e = 0xabadbeef;
+ int v[16];
+ union512i_d res;
+ int i, j;
+
+ for (i = 0; i < 16; i++)
+ {
+ for (j = 0; j < 16; j++)
+ v[j] = 1;
+ v[i] = e;
+
+ res.x = foo (e, i);
+
+ if (check_union512i_d (res, v))
+ abort ();
+
+ res.x = _mm512_setzero_si512 ();
+
+ res.x = foo_r (e, i);
+
+ if (check_union512i_d (res, v))
+ abort ();
+ }
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static __m512d
+__attribute__ ((noinline))
+foo (double *v)
+{
+ return _mm512_set_pd (v[7], v[6], v[5], v[4], v[3], v[2], v[1], v[0]);
+}
+
+static __m512d
+__attribute__ ((noinline))
+foo_r (double *v)
+{
+ return _mm512_setr_pd (v[0], v[1], v[2], v[3], v[4], v[5], v[6], v[7]);
+}
+
+static void
+avx512f_test (void)
+{
+ double v[8] = { -3.3, 2.6, 1.48, 9.104, -23.9, -173.37, -13.48, 69.78 };
+ union512d res;
+
+ res.x = foo (v);
+
+ if (check_union512d (res, v))
+ abort ();
+
+ res.x = _mm512_setzero_pd ();
+
+ res.x = foo_r (v);
+
+ if (check_union512d (res, v))
+ abort ();
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static __m512d
+__attribute__ ((noinline))
+foo (double x1, double x2, double x3, double x4,
+ double x5, double x6, double x7, double x8)
+{
+ return _mm512_set_pd (x1, x2, x3, x4, x5, x6, x7, x8);
+}
+
+static __m512d
+__attribute__ ((noinline))
+foo_r (double x1, double x2, double x3, double x4,
+ double x5, double x6, double x7, double x8)
+{
+ return _mm512_setr_pd (x8, x7, x6, x5, x4, x3, x2, x1);
+}
+
+static void
+avx512f_test (void)
+{
+ double v[8] = { -3.3, 2.6, 1.48, 9.104, -23.9, -173.37, -13.48, 69.78 };
+ union512d res;
+
+ res.x = foo (v[7], v[6], v[5], v[4], v[3], v[2], v[1], v[0]);
+
+ if (check_union512d (res, v))
+ abort ();
+
+ res.x = _mm512_setzero_pd ();
+
+ res.x = foo_r (v[7], v[6], v[5], v[4], v[3], v[2], v[1], v[0]);
+
+ if (check_union512d (res, v))
+ abort ();
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static __m512d
+__attribute__ ((noinline))
+foo (double x)
+{
+ return _mm512_set_pd (x, x, x, x, x, x, x, x);
+}
+
+static __m512d
+__attribute__ ((noinline))
+foo_r (double x)
+{
+ return _mm512_setr_pd (x, x, x, x, x, x, x, x);
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+ double e = 34.5;
+ double v[8];
+ union512d res;
+
+ for (i = 0; i < 8; i++)
+ v[i] = e;
+
+ res.x = foo (e);
+
+ if (check_union512d (res, v))
+ abort ();
+
+ res.x = _mm512_setzero_pd ();
+
+ res.x = foo_r (e);
+
+ if (check_union512d (res, v))
+ abort ();
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static __m512d
+__attribute__ ((noinline))
+foo (double x, int i)
+{
+ switch (i)
+ {
+ case 7:
+ return _mm512_set_pd (x, 0, 0, 0, 0, 0, 0, 0);
+ case 6:
+ return _mm512_set_pd (0, x, 0, 0, 0, 0, 0, 0);
+ case 5:
+ return _mm512_set_pd (0, 0, x, 0, 0, 0, 0, 0);
+ case 4:
+ return _mm512_set_pd (0, 0, 0, x, 0, 0, 0, 0);
+ case 3:
+ return _mm512_set_pd (0, 0, 0, 0, x, 0, 0, 0);
+ case 2:
+ return _mm512_set_pd (0, 0, 0, 0, 0, x, 0, 0);
+ case 1:
+ return _mm512_set_pd (0, 0, 0, 0, 0, 0, x, 0);
+ case 0:
+ return _mm512_set_pd (0, 0, 0, 0, 0, 0, 0, x);
+ default:
+ abort ();
+ }
+}
+
+static __m512d
+__attribute__ ((noinline))
+foo_r (double x, int i)
+{
+ switch (i)
+ {
+ case 0:
+ return _mm512_setr_pd (x, 0, 0, 0, 0, 0, 0, 0);
+ case 1:
+ return _mm512_setr_pd (0, x, 0, 0, 0, 0, 0, 0);
+ case 2:
+ return _mm512_setr_pd (0, 0, x, 0, 0, 0, 0, 0);
+ case 3:
+ return _mm512_setr_pd (0, 0, 0, x, 0, 0, 0, 0);
+ case 4:
+ return _mm512_setr_pd (0, 0, 0, 0, x, 0, 0, 0);
+ case 5:
+ return _mm512_setr_pd (0, 0, 0, 0, 0, x, 0, 0);
+ case 6:
+ return _mm512_setr_pd (0, 0, 0, 0, 0, 0, x, 0);
+ case 7:
+ return _mm512_setr_pd (0, 0, 0, 0, 0, 0, 0, x);
+ default:
+ abort ();
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ double e = -3.234;
+ double v[8];
+ union512d res;
+ int i, j;
+
+ for (i = 0; i < 8; i++)
+ {
+ for (j = 0; j < 8; j++)
+ v[j] = 0;
+ v[i] = e;
+
+ res.x = foo (e, i);
+
+ if (check_union512d (res, v))
+ abort ();
+
+ res.x = _mm512_setzero_pd ();
+
+ res.x = foo_r (e, i);
+
+ if (check_union512d (res, v))
+ abort ();
+ }
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static __m512d
+__attribute__ ((noinline))
+foo (double x, int i)
+{
+ switch (i)
+ {
+ case 7:
+ return _mm512_set_pd (x, 1, 1, 1, 1, 1, 1, 1);
+ case 6:
+ return _mm512_set_pd (1, x, 1, 1, 1, 1, 1, 1);
+ case 5:
+ return _mm512_set_pd (1, 1, x, 1, 1, 1, 1, 1);
+ case 4:
+ return _mm512_set_pd (1, 1, 1, x, 1, 1, 1, 1);
+ case 3:
+ return _mm512_set_pd (1, 1, 1, 1, x, 1, 1, 1);
+ case 2:
+ return _mm512_set_pd (1, 1, 1, 1, 1, x, 1, 1);
+ case 1:
+ return _mm512_set_pd (1, 1, 1, 1, 1, 1, x, 1);
+ case 0:
+ return _mm512_set_pd (1, 1, 1, 1, 1, 1, 1, x);
+ default:
+ abort ();
+ }
+}
+
+static __m512d
+__attribute__ ((noinline))
+foo_r (double x, int i)
+{
+ switch (i)
+ {
+ case 0:
+ return _mm512_setr_pd (x, 1, 1, 1, 1, 1, 1, 1);
+ case 1:
+ return _mm512_setr_pd (1, x, 1, 1, 1, 1, 1, 1);
+ case 2:
+ return _mm512_setr_pd (1, 1, x, 1, 1, 1, 1, 1);
+ case 3:
+ return _mm512_setr_pd (1, 1, 1, x, 1, 1, 1, 1);
+ case 4:
+ return _mm512_setr_pd (1, 1, 1, 1, x, 1, 1, 1);
+ case 5:
+ return _mm512_setr_pd (1, 1, 1, 1, 1, x, 1, 1);
+ case 6:
+ return _mm512_setr_pd (1, 1, 1, 1, 1, 1, x, 1);
+ case 7:
+ return _mm512_setr_pd (1, 1, 1, 1, 1, 1, 1, x);
+ default:
+ abort ();
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ double e = -3.234;
+ double v[8];
+ union512d res;
+ int i, j;
+
+ for (i = 0; i < 8; i++)
+ {
+ for (j = 0; j < 8; j++)
+ v[j] = 1;
+ v[i] = e;
+
+ res.x = foo (e, i);
+
+ if (check_union512d (res, v))
+ abort ();
+
+ res.x = _mm512_setzero_pd ();
+
+ res.x = foo_r (e, i);
+
+ if (check_union512d (res, v))
+ abort ();
+ }
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static __m512i
+__attribute__ ((noinline))
+foo (long long *v)
+{
+ return _mm512_set_epi64 (v[7], v[6], v[5], v[4], v[3], v[2], v[1], v[0]);
+}
+
+static __m512i
+__attribute__ ((noinline))
+foo_r (long long *v)
+{
+ return _mm512_setr_epi64 (v[0], v[1], v[2], v[3], v[4], v[5], v[6], v[7]);
+}
+
+static void
+avx512f_test (void)
+{
+ long long v[8] = { 0x12e9e94645ad8LL, 0x851c0b39446LL, 2134, 6678,
+ 0x786784645245LL, 0x9487731234LL, 41124, 86530 };
+ union512i_q res;
+
+ res.x = foo (v);
+
+ if (check_union512i_q (res, v))
+ abort ();
+
+ res.x = _mm512_setzero_si512 ();
+
+ res.x = foo_r (v);
+
+ if (check_union512i_q (res, v))
+ abort ();
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static __m512i
+__attribute__ ((noinline))
+foo (long long x1, long long x2, long long x3, long long x4,
+ long long x5, long long x6, long long x7, long long x8)
+{
+ return _mm512_set_epi64 (x1, x2, x3, x4, x5, x6, x7, x8);
+}
+
+static __m512i
+__attribute__ ((noinline))
+foo_r (long long x1, long long x2, long long x3, long long x4,
+ long long x5, long long x6, long long x7, long long x8)
+{
+ return _mm512_setr_epi64 (x8, x7, x6, x5, x4, x3, x2, x1);
+}
+
+static void
+avx512f_test (void)
+{
+ long long v[8] = { 0x12e9e94645ad8LL, 0x851c0b39446LL, 2134, 6678,
+ 0x786784645245LL, 0x9487731234LL, 41124, 86530 };
+ union512i_q res;
+
+ res.x = foo (v[7], v[6], v[5], v[4], v[3], v[2], v[1], v[0]);
+
+ if (check_union512i_q (res, v))
+ abort ();
+
+ res.x = _mm512_setzero_si512 ();
+
+ res.x = foo_r (v[7], v[6], v[5], v[4], v[3], v[2], v[1], v[0]);
+
+ if (check_union512i_q (res, v))
+ abort ();
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static __m512i
+__attribute__ ((noinline))
+foo (long long x)
+{
+ return _mm512_set_epi64 (x, x, x, x, x, x, x, x);
+}
+
+static __m512i
+__attribute__ ((noinline))
+foo_r (long long x)
+{
+ return _mm512_setr_epi64 (x, x, x, x, x, x, x, x);
+}
+
+static void
+avx512f_test (void)
+{
+ int i;
+ long long e = 0xfed178ab134badf1LL;
+ long long v[8];
+ union512i_q res;
+
+ for (i = 0; i < 8; i++)
+ v[i] = e;
+
+ res.x = foo (e);
+
+ if (check_union512i_q (res, v))
+ abort ();
+
+ res.x = _mm512_setzero_si512 ();
+
+ res.x = foo_r (e);
+
+ if (check_union512i_q (res, v))
+ abort ();
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static __m512i
+__attribute__ ((noinline))
+foo (long long x, int i)
+{
+ switch (i)
+ {
+ case 7:
+ return _mm512_set_epi64 (x, 0, 0, 0, 0, 0, 0, 0);
+ case 6:
+ return _mm512_set_epi64 (0, x, 0, 0, 0, 0, 0, 0);
+ case 5:
+ return _mm512_set_epi64 (0, 0, x, 0, 0, 0, 0, 0);
+ case 4:
+ return _mm512_set_epi64 (0, 0, 0, x, 0, 0, 0, 0);
+ case 3:
+ return _mm512_set_epi64 (0, 0, 0, 0, x, 0, 0, 0);
+ case 2:
+ return _mm512_set_epi64 (0, 0, 0, 0, 0, x, 0, 0);
+ case 1:
+ return _mm512_set_epi64 (0, 0, 0, 0, 0, 0, x, 0);
+ case 0:
+ return _mm512_set_epi64 (0, 0, 0, 0, 0, 0, 0, x);
+ default:
+ abort ();
+ }
+}
+
+static __m512i
+__attribute__ ((noinline))
+foo_r (long long x, int i)
+{
+ switch (i)
+ {
+ case 0:
+ return _mm512_setr_epi64 (x, 0, 0, 0, 0, 0, 0, 0);
+ case 1:
+ return _mm512_setr_epi64 (0, x, 0, 0, 0, 0, 0, 0);
+ case 2:
+ return _mm512_setr_epi64 (0, 0, x, 0, 0, 0, 0, 0);
+ case 3:
+ return _mm512_setr_epi64 (0, 0, 0, x, 0, 0, 0, 0);
+ case 4:
+ return _mm512_setr_epi64 (0, 0, 0, 0, x, 0, 0, 0);
+ case 5:
+ return _mm512_setr_epi64 (0, 0, 0, 0, 0, x, 0, 0);
+ case 6:
+ return _mm512_setr_epi64 (0, 0, 0, 0, 0, 0, x, 0);
+ case 7:
+ return _mm512_setr_epi64 (0, 0, 0, 0, 0, 0, 0, x);
+ default:
+ abort ();
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ long long e = 0xabadbeef01234567LL;
+ long long v[8];
+ union512i_q res;
+ int i, j;
+
+ for (i = 0; i < 8; i++)
+ {
+ for (j = 0; j < 8; j++)
+ v[j] = 0;
+ v[i] = e;
+
+ res.x = foo (e, i);
+
+ if (check_union512i_q (res, v))
+ abort ();
+
+ res.x = _mm512_setzero_si512 ();
+
+ res.x = foo_r (e, i);
+
+ if (check_union512i_q (res, v))
+ abort ();
+ }
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static __m512i
+__attribute__ ((noinline))
+foo (long long x, int i)
+{
+ switch (i)
+ {
+ case 7:
+ return _mm512_set_epi64 (x, 1, 1, 1, 1, 1, 1, 1);
+ case 6:
+ return _mm512_set_epi64 (1, x, 1, 1, 1, 1, 1, 1);
+ case 5:
+ return _mm512_set_epi64 (1, 1, x, 1, 1, 1, 1, 1);
+ case 4:
+ return _mm512_set_epi64 (1, 1, 1, x, 1, 1, 1, 1);
+ case 3:
+ return _mm512_set_epi64 (1, 1, 1, 1, x, 1, 1, 1);
+ case 2:
+ return _mm512_set_epi64 (1, 1, 1, 1, 1, x, 1, 1);
+ case 1:
+ return _mm512_set_epi64 (1, 1, 1, 1, 1, 1, x, 1);
+ case 0:
+ return _mm512_set_epi64 (1, 1, 1, 1, 1, 1, 1, x);
+ default:
+ abort ();
+ }
+}
+
+static __m512i
+__attribute__ ((noinline))
+foo_r (long long x, int i)
+{
+ switch (i)
+ {
+ case 0:
+ return _mm512_setr_epi64 (x, 1, 1, 1, 1, 1, 1, 1);
+ case 1:
+ return _mm512_setr_epi64 (1, x, 1, 1, 1, 1, 1, 1);
+ case 2:
+ return _mm512_setr_epi64 (1, 1, x, 1, 1, 1, 1, 1);
+ case 3:
+ return _mm512_setr_epi64 (1, 1, 1, x, 1, 1, 1, 1);
+ case 4:
+ return _mm512_setr_epi64 (1, 1, 1, 1, x, 1, 1, 1);
+ case 5:
+ return _mm512_setr_epi64 (1, 1, 1, 1, 1, x, 1, 1);
+ case 6:
+ return _mm512_setr_epi64 (1, 1, 1, 1, 1, 1, x, 1);
+ case 7:
+ return _mm512_setr_epi64 (1, 1, 1, 1, 1, 1, 1, x);
+ default:
+ abort ();
+ }
+}
+
+static void
+avx512f_test (void)
+{
+ long long e = 0xabadbeef01234567LL;
+ long long v[8];
+ union512i_q res;
+ int i, j;
+
+ for (i = 0; i < 8; i++)
+ {
+ for (j = 0; j < 8; j++)
+ v[j] = 1;
+ v[i] = e;
+
+ res.x = foo (e, i);
+
+ if (check_union512i_q (res, v))
+ abort ();
+
+ res.x = _mm512_setzero_si512 ();
+
+ res.x = foo_r (e, i);
+
+ if (check_union512i_q (res, v))
+ abort ();
+ }
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+void static
+avx512f_test (void)
+{
+ int i;
+ union512d res;
+ double res_ref[8];
+
+ res.x = _mm512_setzero_pd ();
+
+ for (i = 0; i < 8; i++)
+ res_ref[i] = 0.0;
+
+ if (check_union512d (res, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+void static
+avx512f_test (void)
+{
+ int i;
+ union512 res;
+ float res_ref[16];
+
+ res.x = _mm512_setzero_ps ();
+
+ for (i = 0; i < 16; i++)
+ res_ref[i] = 0.0;
+
+ if (check_union512 (res, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+void static
+avx512f_test (void)
+{
+ int i;
+ union512i_q res;
+ long long res_ref[8];
+
+ res.x = _mm512_setzero_si512 ();
+
+ for (i = 0; i < 8; i++)
+ res_ref[i] = 0;
+
+ if (check_union512i_q (res, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vaddpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vaddpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vaddpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vaddpd\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vaddpd\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vaddpd\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_add_pd (x, x);
+ x = _mm512_mask_add_pd (x, m, x, x);
+ x = _mm512_maskz_add_pd (m, x, x);
+ x = _mm512_add_round_pd (x, x, _MM_FROUND_TO_NEAREST_INT);
+ x = _mm512_mask_add_round_pd (x, m, x, x, _MM_FROUND_TO_NEG_INF);
+ x = _mm512_maskz_add_round_pd (m, x, x, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (double *r, double *s1, double *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[i] + s2[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, d) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 1.5 + 34.67 * i * sign;
+ src2.a[i] = -22.17 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_add_pd) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_add_pd) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_add_pd) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vaddps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vaddps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vaddps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vaddps\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vaddps\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vaddps\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_add_ps (x, x);
+ x = _mm512_mask_add_ps (x, m, x, x);
+ x = _mm512_maskz_add_ps (m, x, x);
+ x = _mm512_add_round_ps (x, x, _MM_FROUND_TO_NEAREST_INT);
+ x = _mm512_mask_add_round_ps (x, m, x, x, _MM_FROUND_TO_POS_INF);
+ x = _mm512_maskz_add_round_ps (m, x, x, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (float *r, float *s1, float *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[i] + s2[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN,) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 1.5 + 34.67 * i * sign;
+ src2.a[i] = -22.17 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_add_ps) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_add_ps) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_add_ps) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN,) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "valignd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "valignd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "valignd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i z;
+volatile __mmask16 m1;
+
+void extern
+avx512f_test (void)
+{
+ z = _mm512_alignr_epi32 (z, z, 3);
+ z = _mm512_mask_alignr_epi32 (z, m1, z, z, 3);
+ z = _mm512_maskz_alignr_epi32 (m1, z, z, 3);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+#define N (SIZE / 2)
+
+static void
+CALC (int *s1, int *s2, int *r)
+{
+ int i;
+ int s[2 * SIZE];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s[i] = s2[i];
+ s[i + SIZE] = s1[i];
+ }
+
+ for (i = 0; i < SIZE; i++)
+ r[i] = s[i + N];
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3, s1, s2;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 2 * i;
+ s2.a[i] = i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_alignr_epi32) (s1.x, s2.x, N);
+ res2.x = INTRINSIC (_mask_alignr_epi32) (res2.x, mask, s1.x, s2.x, N);
+ res3.x = INTRINSIC (_maskz_alignr_epi32) (mask, s1.x, s2.x, N);
+
+ CALC (s1.a, s2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "valignq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "valignq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "valignq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i z;
+volatile __mmask8 m1;
+
+void extern
+avx512f_test (void)
+{
+ z = _mm512_alignr_epi64 (z, z, 3);
+ z = _mm512_mask_alignr_epi64 (z, m1, z, z, 3);
+ z = _mm512_maskz_alignr_epi64 (m1, z, z, 3);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+#define N (SIZE / 2)
+
+static void
+CALC (long long *s1, long long *s2, long long *r)
+{
+ int i;
+ long long s[2 * SIZE];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s[i] = s2[i];
+ s[i + SIZE] = s1[i];
+ }
+
+ for (i = 0; i < SIZE; i++)
+ r[i] = s[i + N];
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3, s1, s2;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 2 * i;
+ s2.a[i] = i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_alignr_epi64) (s1.x, s2.x, N);
+ res2.x = INTRINSIC (_mask_alignr_epi64) (res2.x, mask, s1.x, s2.x, N);
+ res3.x = INTRINSIC (_maskz_alignr_epi64) (mask, s1.x, s2.x, N);
+
+ CALC (s1.a, s2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler "(vblendmpd|vmovapd)\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mask_blend_pd (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (double *r, double *s1, double *s2, MASK_TYPE mask)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = (mask & (1LL << i)) ? s2[i] : s1[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, d) res1, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 1.5 + 34.67 * i * sign;
+ src2.a[i] = -22.17 * i * sign;
+ sign = sign * -1;
+ }
+
+ res1.x = INTRINSIC (_mask_blend_pd) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a, mask);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler "(vblendmps|vmovaps)\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mask_blend_ps (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (float *r, float *s1, float *s2, MASK_TYPE mask)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = (mask & (1 << i)) ? s2[i] : s1[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN,) res1, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 1.5 + 34.67 * i * sign;
+ src2.a[i] = -22.17 * i * sign;
+ sign = sign * -1;
+ }
+
+ res1.x = INTRINSIC (_mask_blend_ps) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a, mask);
+
+ if (UNION_CHECK (AVX512F_LEN,) (res1, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vbroadcastf32x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]|vshuff32x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vbroadcastf32x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]|vshuff32x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vbroadcastf32x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}|vshuff32x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __m128 y;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_broadcast_f32x4 (y);
+ x = _mm512_mask_broadcast_f32x4 (x, m, y);
+ x = _mm512_maskz_broadcast_f32x4 (m, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (float *r, float *s)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s[i % 4];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN,) res1, res2, res3;
+ UNION_TYPE (128,) src;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < 4; i++)
+ {
+ src.a[i] = 34.67 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_broadcast_f32x4) (src.x);
+ res2.x = INTRINSIC (_mask_broadcast_f32x4) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_broadcast_f32x4) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (AVX512F_LEN,) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE ()(res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO ()(res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vbroadcastf64x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]|vshuff64x2\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vbroadcastf64x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]|vshuff64x2\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vbroadcastf64x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}|vshuff64x2\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __m256d y;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_broadcast_f64x4 (y);
+ x = _mm512_mask_broadcast_f64x4 (x, m, y);
+ x = _mm512_maskz_broadcast_f64x4 (m, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (double *r, double *s)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s[i % 4];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, d) res1, res2, res3;
+ UNION_TYPE (256, d) src;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < 2; i++)
+ {
+ src.a[i] = 34.67 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_broadcast_f64x4) (src.x);
+ res2.x = INTRINSIC (_mask_broadcast_f64x4) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_broadcast_f64x4) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vbroadcasti32x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]|vshufi32x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vbroadcasti32x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]|vshufi32x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vbroadcasti32x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}|vshufi32x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __m128i y;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_broadcast_i32x4 (y);
+ x = _mm512_mask_broadcast_i32x4 (x, m, y);
+ x = _mm512_maskz_broadcast_i32x4 (m, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (int *r, int *s)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s[i % 4];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3;
+ UNION_TYPE (128, i_d) src;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < 4; i++)
+ {
+ src.a[i] = 34 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_broadcast_i32x4) (src.x);
+ res2.x = INTRINSIC (_mask_broadcast_i32x4) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_broadcast_i32x4) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vbroadcasti64x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]|vshufi64x2\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vbroadcasti64x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]|vshufi64x2\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vbroadcasti64x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}|vshufi64x2\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __m256i y;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_broadcast_i64x4 (y);
+ x = _mm512_mask_broadcast_i64x4 (x, m, y);
+ x = _mm512_maskz_broadcast_i64x4 (m, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (long long *r, long long *s)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s[i % 4];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3;
+ UNION_TYPE (256, i_q) src;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < 2; i++)
+ {
+ src.a[i] = 34 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_broadcast_i64x4) (src.x);
+ res2.x = INTRINSIC (_mask_broadcast_i64x4) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_broadcast_i64x4) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vbroadcastsd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vbroadcastsd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vbroadcastsd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __m128d y;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_broadcastsd_pd (y);
+ x = _mm512_mask_broadcastsd_pd (x, m, y);
+ x = _mm512_maskz_broadcastsd_pd (m, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (double *r, double *s)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s[0];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, d) res1, res2, res3;
+ UNION_TYPE (128, d) src;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < 2; i++)
+ {
+ src.a[i] = 1.5 + 34.67 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_broadcastsd_pd) (src.x);
+ res2.x = INTRINSIC (_mask_broadcastsd_pd) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_broadcastsd_pd) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vbroadcastss\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vbroadcastss\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vbroadcastss\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __m128 y;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_broadcastss_ps (y);
+ x = _mm512_mask_broadcastss_ps (x, m, y);
+ x = _mm512_maskz_broadcastss_ps (m, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (float *r, float *s)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s[0];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN,) res1, res2, res3;
+ UNION_TYPE (128,) src;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < 4; i++)
+ {
+ src.a[i] = 1.5 + 34.67 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_broadcastss_ps) (src.x);
+ res2.x = INTRINSIC (_mask_broadcastss_ps) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_broadcastss_ps) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (AVX512F_LEN,) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE ()(res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO ()(res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler "vcmppd\[ \\t\]+\[^\n\]*\[^\}\]%zmm\[0-9\]\[^\n^k\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vcmppd\[ \\t\]+\[^\n\]*\[^\}\]%zmm\[0-9\]\[^\n^k\]*%k\[1-7\]\{" } } */
+/* { dg-final { scan-assembler "vcmppd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[^\n^k\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vcmppd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[^\n^k\]*%k\[1-7\]\{" } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ m = _mm512_cmp_pd_mask (x, x, _CMP_FALSE_OQ);
+ m = _mm512_mask_cmp_pd_mask (m, x, x, _CMP_FALSE_OQ);
+ m = _mm512_cmp_round_pd_mask (x, x, _CMP_FALSE_OQ, _MM_FROUND_NO_EXC);
+ m = _mm512_mask_cmp_round_pd_mask (m, x, x, _CMP_FALSE_OQ, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#include <math.h>
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+#if AVX512F_LEN == 512
+#define CMP(imm, rel) \
+ dst_ref = 0; \
+ for (i = 0; i < 8; i++) \
+ { \
+ dst_ref = (((int) rel) << i) | dst_ref; \
+ } \
+ source1.x = _mm512_loadu_pd(s1); \
+ source2.x = _mm512_loadu_pd(s2); \
+ dst1 = _mm512_cmp_pd_mask(source1.x, source2.x, imm);\
+ dst2 = _mm512_mask_cmp_pd_mask(mask, source1.x, source2.x, imm);\
+ if (dst_ref != dst1) abort(); \
+ if ((dst_ref & mask) != dst2) abort();
+#endif
+
+static void
+TEST ()
+{
+ UNION_TYPE (AVX512F_LEN, d) source1, source2;
+ MASK_TYPE dst1, dst2, dst_ref;
+ MASK_TYPE mask = MASK_VALUE;
+ int i;
+ double s1[8]={2134.3343, 6678.346, 453.345635, 54646.464,
+ 231.23311, 5674.455, 111.111111, 23241.152};
+ double s2[8]={41124.234, 6678.346, 8653.65635, 856.43576,
+ 231.23311, 4646.123, 111.111111, 124.12455};
+
+ CMP(_CMP_EQ_OQ, !isunordered(s1[i], s2[i]) && s1[i] == s2[i]);
+ CMP(_CMP_LT_OS, !isunordered(s1[i], s2[i]) && s1[i] < s2[i]);
+ CMP(_CMP_LE_OS, !isunordered(s1[i], s2[i]) && s1[i] <= s2[i]);
+ CMP(_CMP_UNORD_Q, isunordered(s1[i], s2[i]));
+ CMP(_CMP_NEQ_UQ, isunordered(s1[i], s2[i]) || s1[i] != s2[i]);
+ CMP(_CMP_NLT_US, isunordered(s1[i], s2[i]) || s1[i] >= s2[i]);
+ CMP(_CMP_NLE_US, isunordered(s1[i], s2[i]) || s1[i] > s2[i]);
+ CMP(_CMP_ORD_Q, !isunordered(s1[i], s2[i]));
+
+ CMP(_CMP_EQ_UQ, isunordered(s1[i], s2[i]) || s1[i] == s2[i]);
+ CMP(_CMP_NGE_US, isunordered(s1[i], s2[i]) || s1[i] < s2[i]);
+ CMP(_CMP_NGT_US, isunordered(s1[i], s2[i]) || s1[i] <= s2[i]);
+
+ CMP(_CMP_FALSE_OQ, 0);
+ CMP(_CMP_NEQ_OQ, !isunordered(s1[i], s2[i]) && s1[i] != s2[i]);
+ CMP(_CMP_GE_OS, !isunordered(s1[i], s2[i]) && s1[i] >= s2[i]);
+ CMP(_CMP_GT_OS, !isunordered(s1[i], s2[i]) && s1[i] > s2[i]);
+ CMP(_CMP_TRUE_UQ, 1);
+
+ CMP(_CMP_EQ_OS, !isunordered(s1[i], s2[i]) && s1[i] == s2[i]);
+ CMP(_CMP_LT_OQ, !isunordered(s1[i], s2[i]) && s1[i] < s2[i]);
+ CMP(_CMP_LE_OQ, !isunordered(s1[i], s2[i]) && s1[i] <= s2[i]);
+ CMP(_CMP_UNORD_S, isunordered(s1[i], s2[i]));
+ CMP(_CMP_NEQ_US, isunordered(s1[i], s2[i]) || s1[i] != s2[i]);
+ CMP(_CMP_NLT_UQ, isunordered(s1[i], s2[i]) || s1[i] >= s2[i]);
+ CMP(_CMP_NLE_UQ, isunordered(s1[i], s2[i]) || s1[i] > s2[i]);
+ CMP(_CMP_ORD_S, !isunordered(s1[i], s2[i]));
+ CMP(_CMP_EQ_US, isunordered(s1[i], s2[i]) || s1[i] == s2[i]);
+ CMP(_CMP_NGE_UQ, isunordered(s1[i], s2[i]) || s1[i] < s2[i]);
+ CMP(_CMP_NGT_UQ, isunordered(s1[i], s2[i]) || s1[i] <= s2[i]);
+ CMP(_CMP_FALSE_OS, 0);
+ CMP(_CMP_NEQ_OS, !isunordered(s1[i], s2[i]) && s1[i] != s2[i]);
+ CMP(_CMP_GE_OQ, !isunordered(s1[i], s2[i]) && s1[i] >= s2[i]);
+ CMP(_CMP_GT_OQ, !isunordered(s1[i], s2[i]) && s1[i] > s2[i]);
+ CMP(_CMP_TRUE_US, 1)
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler "vcmpps\[ \\t\]+\[^\n\]*\[^\}\]%zmm\[0-9\]\[^\n^k\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vcmpps\[ \\t\]+\[^\n\]*\[^\}\]%zmm\[0-9\]\[^\n^k\]*%k\[1-7\]\{" } } */
+/* { dg-final { scan-assembler "vcmpps\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[^\n^k\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vcmpps\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[^\n^k\]*%k\[1-7\]\{" } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ m = _mm512_cmp_ps_mask (x, x, _CMP_FALSE_OQ);
+ m = _mm512_mask_cmp_ps_mask (m, x, x, _CMP_FALSE_OQ);
+ m = _mm512_cmp_round_ps_mask (x, x, _CMP_FALSE_OQ, _MM_FROUND_NO_EXC);
+ m = _mm512_mask_cmp_round_ps_mask (m, x, x, _CMP_FALSE_OQ, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#include <math.h>
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+#if AVX512F_LEN == 512
+#define CMP(imm, rel) \
+ dst_ref = 0; \
+ for (i = 0; i < 16; i++) \
+ { \
+ dst_ref = (((int) rel) << i) | dst_ref; \
+ } \
+ source1.x = _mm512_loadu_ps(s1); \
+ source2.x = _mm512_loadu_ps(s2); \
+ dst1 = _mm512_cmp_ps_mask(source1.x, source2.x, imm);\
+ dst2 = _mm512_mask_cmp_ps_mask(mask, source1.x, source2.x, imm);\
+ if (dst_ref != dst1) abort(); \
+ if ((dst_ref & mask) != dst2) abort();
+#endif
+
+static void
+TEST ()
+{
+ UNION_TYPE (AVX512F_LEN,) source1, source2;
+ MASK_TYPE dst1, dst2, dst_ref;
+ MASK_TYPE mask = MASK_VALUE;
+ int i;
+ float s1[16] = {2134.3343, 6678.346, 453.345635, 54646.464,
+ 231.23311, 5674.455, 111.111111, 23241.152,
+ 123.14811, 1245.124, 244.151353, 53454.141,
+ 926.16717, 3733.261, 643.161644, 23514.633};
+ float s2[16] = {41124.234, 6678.346, 8653.65635, 856.43576,
+ 231.23311, 4646.123, 111.111111, 124.12455,
+ 123.14811, 1245.124, 244.151353, 53454.141,
+ 2134.3343, 6678.346, 453.345635, 54646.464};
+
+ CMP(_CMP_EQ_OQ, !isunordered(s1[i], s2[i]) && s1[i] == s2[i]);
+ CMP(_CMP_LT_OS, !isunordered(s1[i], s2[i]) && s1[i] < s2[i]);
+ CMP(_CMP_LE_OS, !isunordered(s1[i], s2[i]) && s1[i] <= s2[i]);
+ CMP(_CMP_UNORD_Q, isunordered(s1[i], s2[i]));
+ CMP(_CMP_NEQ_UQ, isunordered(s1[i], s2[i]) || s1[i] != s2[i]);
+ CMP(_CMP_NLT_US, isunordered(s1[i], s2[i]) || s1[i] >= s2[i]);
+ CMP(_CMP_NLE_US, isunordered(s1[i], s2[i]) || s1[i] > s2[i]);
+ CMP(_CMP_ORD_Q, !isunordered(s1[i], s2[i]));
+
+ CMP(_CMP_EQ_UQ, isunordered(s1[i], s2[i]) || s1[i] == s2[i]);
+ CMP(_CMP_NGE_US, isunordered(s1[i], s2[i]) || s1[i] < s2[i]);
+ CMP(_CMP_NGT_US, isunordered(s1[i], s2[i]) || s1[i] <= s2[i]);
+
+ CMP(_CMP_FALSE_OQ, 0);
+ CMP(_CMP_NEQ_OQ, !isunordered(s1[i], s2[i]) && s1[i] != s2[i]);
+ CMP(_CMP_GE_OS, !isunordered(s1[i], s2[i]) && s1[i] >= s2[i]);
+ CMP(_CMP_GT_OS, !isunordered(s1[i], s2[i]) && s1[i] > s2[i]);
+ CMP(_CMP_TRUE_UQ, 1);
+
+ CMP(_CMP_EQ_OS, !isunordered(s1[i], s2[i]) && s1[i] == s2[i]);
+ CMP(_CMP_LT_OQ, !isunordered(s1[i], s2[i]) && s1[i] < s2[i]);
+ CMP(_CMP_LE_OQ, !isunordered(s1[i], s2[i]) && s1[i] <= s2[i]);
+ CMP(_CMP_UNORD_S, isunordered(s1[i], s2[i]));
+ CMP(_CMP_NEQ_US, isunordered(s1[i], s2[i]) || s1[i] != s2[i]);
+ CMP(_CMP_NLT_UQ, isunordered(s1[i], s2[i]) || s1[i] >= s2[i]);
+ CMP(_CMP_NLE_UQ, isunordered(s1[i], s2[i]) || s1[i] > s2[i]);
+ CMP(_CMP_ORD_S, !isunordered(s1[i], s2[i]));
+ CMP(_CMP_EQ_US, isunordered(s1[i], s2[i]) || s1[i] == s2[i]);
+ CMP(_CMP_NGE_UQ, isunordered(s1[i], s2[i]) || s1[i] < s2[i]);
+ CMP(_CMP_NGT_UQ, isunordered(s1[i], s2[i]) || s1[i] <= s2[i]);
+ CMP(_CMP_FALSE_OS, 0);
+ CMP(_CMP_NEQ_OS, !isunordered(s1[i], s2[i]) && s1[i] != s2[i]);
+ CMP(_CMP_GE_OQ, !isunordered(s1[i], s2[i]) && s1[i] >= s2[i]);
+ CMP(_CMP_GT_OQ, !isunordered(s1[i], s2[i]) && s1[i] > s2[i]);
+ CMP(_CMP_TRUE_US, 1)
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler "vcmpsd\[ \\t\]+\[^\n\]*\[^\}\]%xmm\[0-9\]\[^\n^k\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vcmpsd\[ \\t\]+\[^\n\]*\[^\}\]%xmm\[0-9\]\[^\n^k\]*%k\[1-7\]\{" } } */
+/* { dg-final { scan-assembler "vcmpsd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]\[^\n^k\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vcmpsd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]\[^\n^k\]*%k\[1-7\]\{" } } */
+
+#include <immintrin.h>
+
+volatile __m128d x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ m = _mm_cmp_sd_mask (x, x, _CMP_FALSE_OQ);
+ m = _mm_mask_cmp_sd_mask (m, x, x, _CMP_FALSE_OQ);
+ m = _mm_cmp_round_sd_mask (x, x, _CMP_FALSE_OQ, _MM_FROUND_NO_EXC);
+ m = _mm_mask_cmp_round_sd_mask (m, x, x, _CMP_FALSE_OQ, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-require-effective-target avx512f } */
+/* { dg-require-effective-target c99_runtime } */
+/* { dg-options "-O2 -mavx512f -std=c99" } */
+
+#include "avx512f-check.h"
+#include <math.h>
+
+double s1[2] = {2134.3343, 6678.346};
+double s2[2] = {1485.1288, 6678.346};
+
+__mmask8 dst_ref;
+
+#define CMP(imm, rel) \
+ dst_ref = 0; \
+ dst_ref = ((int) rel) | dst_ref; \
+ source1 = _mm_loadu_pd(s1); \
+ source2 = _mm_loadu_pd(s2); \
+ dst = _mm_cmp_sd_mask(source1, source2, imm); \
+ dst2 = _mm_mask_cmp_sd_mask(mask, source1, source2, imm);\
+ if (dst_ref != dst) abort(); \
+ if ((dst_ref & mask) != dst2) abort();
+
+static void
+avx512f_test ()
+{
+ __m128d source1, source2;
+ __mmask8 dst, dst2, mask;
+ mask = 1;
+ int i;
+
+ CMP(_CMP_EQ_OQ, !isunordered(s1[0], s2[0]) && s1[0] == s2[0]);
+ CMP(_CMP_LT_OS, !isunordered(s1[0], s2[0]) && s1[0] < s2[0]);
+ CMP(_CMP_LE_OS, !isunordered(s1[0], s2[0]) && s1[0] <= s2[0]);
+ CMP(_CMP_UNORD_Q, isunordered(s1[0], s2[0]));
+ CMP(_CMP_NEQ_UQ, isunordered(s1[0], s2[0]) || s1[0] != s2[0]);
+ CMP(_CMP_NLT_US, isunordered(s1[0], s2[0]) || s1[0] >= s2[0]);
+ CMP(_CMP_NLE_US, isunordered(s1[0], s2[0]) || s1[0] > s2[0]);
+ CMP(_CMP_ORD_Q, !isunordered(s1[0], s2[0]));
+
+ CMP(_CMP_EQ_UQ, isunordered(s1[0], s2[0]) || s1[0] == s2[0]);
+ CMP(_CMP_NGE_US, isunordered(s1[0], s2[0]) || s1[0] < s2[0]);
+ CMP(_CMP_NGT_US, isunordered(s1[0], s2[0]) || s1[0] <= s2[0]);
+
+ CMP(_CMP_FALSE_OQ, 0);
+ CMP(_CMP_NEQ_OQ, !isunordered(s1[0], s2[0]) && s1[0] != s2[0]);
+ CMP(_CMP_GE_OS, !isunordered(s1[0], s2[0]) && s1[0] >= s2[0]);
+ CMP(_CMP_GT_OS, !isunordered(s1[0], s2[0]) && s1[0] > s2[0]);
+ CMP(_CMP_TRUE_UQ, 1);
+
+ CMP(_CMP_EQ_OS, !isunordered(s1[0], s2[0]) && s1[0] == s2[0]);
+ CMP(_CMP_LT_OQ, !isunordered(s1[0], s2[0]) && s1[0] < s2[0]);
+ CMP(_CMP_LE_OQ, !isunordered(s1[0], s2[0]) && s1[0] <= s2[0]);
+ CMP(_CMP_UNORD_S, isunordered(s1[0], s2[0]));
+ CMP(_CMP_NEQ_US, isunordered(s1[0], s2[0]) || s1[0] != s2[0]);
+ CMP(_CMP_NLT_UQ, isunordered(s1[0], s2[0]) || s1[0] >= s2[0]);
+ CMP(_CMP_NLE_UQ, isunordered(s1[0], s2[0]) || s1[0] > s2[0]);
+ CMP(_CMP_ORD_S, !isunordered(s1[0], s2[0]));
+ CMP(_CMP_EQ_US, isunordered(s1[0], s2[0]) || s1[0] == s2[0]);
+ CMP(_CMP_NGE_UQ, isunordered(s1[0], s2[0]) || s1[0] < s2[0]);
+ CMP(_CMP_NGT_UQ, isunordered(s1[0], s2[0]) || s1[0] <= s2[0]);
+ CMP(_CMP_FALSE_OS, 0);
+ CMP(_CMP_NEQ_OS, !isunordered(s1[0], s2[0]) && s1[0] != s2[0]);
+ CMP(_CMP_GE_OQ, !isunordered(s1[0], s2[0]) && s1[0] >= s2[0]);
+ CMP(_CMP_GT_OQ, !isunordered(s1[0], s2[0]) && s1[0] > s2[0]);
+ CMP(_CMP_TRUE_US, 1)
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler "vcmpss\[ \\t\]+\[^\n\]*\[^\}\]%xmm\[0-9\]\[^\n^k\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vcmpss\[ \\t\]+\[^\n\]*\[^\}\]%xmm\[0-9\]\[^\n^k\]*%k\[1-7\]\{" } } */
+/* { dg-final { scan-assembler "vcmpss\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]\[^\n^k\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vcmpss\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]\[^\n^k\]*%k\[1-7\]\{" } } */
+
+#include <immintrin.h>
+
+volatile __m128 x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ m = _mm_cmp_ss_mask (x, x, _CMP_FALSE_OQ);
+ m = _mm_mask_cmp_ss_mask (m, x, x, _CMP_FALSE_OQ);
+ m = _mm_cmp_round_ss_mask (x, x, _CMP_FALSE_OQ, _MM_FROUND_NO_EXC);
+ m = _mm_mask_cmp_round_ss_mask (m, x, x, _CMP_FALSE_OQ, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-require-effective-target avx512f } */
+/* { dg-require-effective-target c99_runtime } */
+/* { dg-options "-O2 -mavx512f -std=c99" } */
+
+#include "avx512f-check.h"
+#include <math.h>
+
+float s1[4] = {2134.3343, 6678.346, 453.345635, 54646.464};
+float s2[4] = {1485.1288, 6678.346, 8653.65635, 856.43576};
+
+__mmask8 dst_ref;
+
+#define CMP(imm, rel) \
+ dst_ref = 0; \
+ dst_ref = ((int) rel) | dst_ref; \
+ source1 = _mm_loadu_ps(s1); \
+ source2 = _mm_loadu_ps(s2); \
+ dst = _mm_cmp_ss_mask(source1, source2, imm); \
+ dst2 = _mm_mask_cmp_ss_mask(mask, source1, source2, imm);\
+ if (dst_ref != dst) abort(); \
+ if ((dst_ref & mask)!= dst2) abort();
+
+static void
+avx512f_test ()
+{
+ __m128 source1, source2;
+ __mmask8 dst, dst2, mask;
+ int i;
+
+ mask = 1;
+
+ CMP(_CMP_EQ_OQ, !isunordered(s1[0], s2[0]) && s1[0] == s2[0]);
+ CMP(_CMP_LT_OS, !isunordered(s1[0], s2[0]) && s1[0] < s2[0]);
+ CMP(_CMP_LE_OS, !isunordered(s1[0], s2[0]) && s1[0] <= s2[0]);
+ CMP(_CMP_UNORD_Q, isunordered(s1[0], s2[0]));
+ CMP(_CMP_NEQ_UQ, isunordered(s1[0], s2[0]) || s1[0] != s2[0]);
+ CMP(_CMP_NLT_US, isunordered(s1[0], s2[0]) || s1[0] >= s2[0]);
+ CMP(_CMP_NLE_US, isunordered(s1[0], s2[0]) || s1[0] > s2[0]);
+ CMP(_CMP_ORD_Q, !isunordered(s1[0], s2[0]));
+
+ CMP(_CMP_EQ_UQ, isunordered(s1[0], s2[0]) || s1[0] == s2[0]);
+ CMP(_CMP_NGE_US, isunordered(s1[0], s2[0]) || s1[0] < s2[0]);
+ CMP(_CMP_NGT_US, isunordered(s1[0], s2[0]) || s1[0] <= s2[0]);
+
+ CMP(_CMP_FALSE_OQ, 0);
+ CMP(_CMP_NEQ_OQ, !isunordered(s1[0], s2[0]) && s1[0] != s2[0]);
+ CMP(_CMP_GE_OS, !isunordered(s1[0], s2[0]) && s1[0] >= s2[0]);
+ CMP(_CMP_GT_OS, !isunordered(s1[0], s2[0]) && s1[0] > s2[0]);
+ CMP(_CMP_TRUE_UQ, 1);
+
+ CMP(_CMP_EQ_OS, !isunordered(s1[0], s2[0]) && s1[0] == s2[0]);
+ CMP(_CMP_LT_OQ, !isunordered(s1[0], s2[0]) && s1[0] < s2[0]);
+ CMP(_CMP_LE_OQ, !isunordered(s1[0], s2[0]) && s1[0] <= s2[0]);
+ CMP(_CMP_UNORD_S, isunordered(s1[0], s2[0]));
+ CMP(_CMP_NEQ_US, isunordered(s1[0], s2[0]) || s1[0] != s2[0]);
+ CMP(_CMP_NLT_UQ, isunordered(s1[0], s2[0]) || s1[0] >= s2[0]);
+ CMP(_CMP_NLE_UQ, isunordered(s1[0], s2[0]) || s1[0] > s2[0]);
+ CMP(_CMP_ORD_S, !isunordered(s1[0], s2[0]));
+ CMP(_CMP_EQ_US, isunordered(s1[0], s2[0]) || s1[0] == s2[0]);
+ CMP(_CMP_NGE_UQ, isunordered(s1[0], s2[0]) || s1[0] < s2[0]);
+ CMP(_CMP_NGT_UQ, isunordered(s1[0], s2[0]) || s1[0] <= s2[0]);
+ CMP(_CMP_FALSE_OS, 0);
+ CMP(_CMP_NEQ_OS, !isunordered(s1[0], s2[0]) && s1[0] != s2[0]);
+ CMP(_CMP_GE_OQ, !isunordered(s1[0], s2[0]) && s1[0] >= s2[0]);
+ CMP(_CMP_GT_OQ, !isunordered(s1[0], s2[0]) && s1[0] > s2[0]);
+ CMP(_CMP_TRUE_US, 1)
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler "vcomisd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm" } } */
+
+#include <immintrin.h>
+
+volatile __m128d x;
+volatile int res;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm_comi_round_sd (x, x, _CMP_LT_OS, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcomiss\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm" 1 } } */
+/* { dg-final { scan-assembler-times "vcomiss\[ \\t\]+\[^{}\n\]*%xmm" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128 x;
+volatile int res;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm_comi_round_ss (x, x, _CMP_LT_OS, _MM_FROUND_NO_EXC);
+}
+
+void extern
+avx512f_test_2 (void)
+{
+ res = _mm_comi_round_ss (x, x, _CMP_LT_OS, _MM_FROUND_CUR_DIRECTION);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcompresspd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcompresspd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vcompresspd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*\\)\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+double *p;
+volatile __m512d x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mask_compress_pd (x, m, x);
+ x = _mm512_maskz_compress_pd (m, x);
+
+ _mm512_mask_compressstoreu_pd (p, m, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+#define MASK ((1 << SIZE) - 1)
+#include <x86intrin.h>
+
+static void
+CALC (double *s, double *r, MASK_TYPE mask)
+{
+ int i, k;
+
+ for (i = 0, k = 0; i < SIZE; i++)
+ {
+ if (mask & (1 << i))
+ r[k++] = s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s, res1, res2;
+ double res3[SIZE];
+ MASK_TYPE compressed_mask, mask = MASK_VALUE;
+ double res_ref[SIZE];
+ int i, mask_bit_count, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 12345 * (i + 200) * sign;
+ res1.a[i] = DEFAULT_VALUE;
+ res3[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+ res1.x = INTRINSIC (_mask_compress_pd) (res1.x, mask, s.x);
+ res2.x = INTRINSIC (_maskz_compress_pd) (mask, s.x);
+ INTRINSIC (_mask_compressstoreu_pd) (res3, mask, s.x);
+
+ mask_bit_count = __popcntd (mask & MASK);
+ compressed_mask = (1 << mask_bit_count) - 1;
+ CALC (s.a, res_ref, mask);
+
+ MASK_MERGE (d) (res_ref, compressed_mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, compressed_mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, compressed_mask, SIZE);
+ if (checkVd (res3, res_ref, SIZE))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcompressps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcompressps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vcompressps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*\\)\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+float *p;
+volatile __m512 x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mask_compress_ps (x, m, x);
+ x = _mm512_maskz_compress_ps (m, x);
+
+ _mm512_mask_compressstoreu_ps (p, m, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#define MASK ((1 << SIZE) - 1)
+#include <x86intrin.h>
+
+static void
+CALC (float *s, float *r, MASK_TYPE mask)
+{
+ int i, k;
+
+ for (i = 0, k = 0; i < SIZE; i++)
+ {
+ if (mask & (1 << i))
+ r[k++] = s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, ) s, res1, res2;
+ float res3[SIZE];
+ MASK_TYPE compressed_mask, mask = MASK_VALUE;
+ float res_ref[SIZE];
+ int i, mask_bit_count, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 12345 * (i + 200) * sign;
+ res1.a[i] = DEFAULT_VALUE;
+ res3[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+ res1.x = INTRINSIC (_mask_compress_ps) (res1.x, mask, s.x);
+ res2.x = INTRINSIC (_maskz_compress_ps) (mask, s.x);
+ INTRINSIC (_mask_compressstoreu_ps) (res3, mask, s.x);
+
+ mask_bit_count = __popcntd (mask & MASK);
+ compressed_mask = (1 << mask_bit_count) - 1;
+ CALC (s.a, res_ref, mask);
+
+ MASK_MERGE () (res_ref, compressed_mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res1, res_ref))
+ abort ();
+
+ MASK_ZERO () (res_ref, compressed_mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res2, res_ref))
+ abort ();
+
+ MASK_MERGE () (res_ref, compressed_mask, SIZE);
+ if (checkVf (res3, res_ref, SIZE))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcvtdq2pd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtdq2pd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtdq2pd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m256i s;
+volatile __m512d res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtepi32_pd (s);
+ res = _mm512_mask_cvtepi32_pd (res, m, s);
+ res = _mm512_maskz_cvtepi32_pd (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SRC_SIZE ((AVX512F_LEN_HALF) / 32)
+#include "avx512f-mask-type.h"
+#define DST_SIZE ((AVX512F_LEN) / 64)
+
+static void
+CALC (int *s, double *r)
+{
+ int i;
+
+ for (i = 0; i < DST_SIZE; i++)
+ {
+ r[i] = (double) s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN_HALF, i_d) s;
+ UNION_TYPE (AVX512F_LEN, d) res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[DST_SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SRC_SIZE; i++)
+ {
+ s.a[i] = 123456 * (i + 2000) * sign;
+ sign = -sign;
+ }
+
+ for (i = 0; i < DST_SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_cvtepi32_pd) (s.x);
+ res2.x = INTRINSIC (_mask_cvtepi32_pd) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_cvtepi32_pd) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, DST_SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, DST_SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcvtdq2ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vcvtdq2ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtdq2ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtdq2ps\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtdq2ps\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtdq2ps\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i s;
+volatile __m512 res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtepi32_ps (s);
+ res = _mm512_mask_cvtepi32_ps (res, m, s);
+ res = _mm512_maskz_cvtepi32_ps (m, s);
+ res = _mm512_cvt_roundepi32_ps (s, _MM_FROUND_TO_NEAREST_INT);
+ res = _mm512_mask_cvt_roundepi32_ps (res, m, s, _MM_FROUND_TO_POS_INF);
+ res = _mm512_maskz_cvt_roundepi32_ps (m, s, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (int *s, float *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = (float) s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) s;
+ UNION_TYPE (AVX512F_LEN, ) res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 123456 * (i + 2000) * sign;
+ res2.a[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+ res1.x = INTRINSIC (_cvtepi32_ps) (s.x);
+ res2.x = INTRINSIC (_mask_cvtepi32_ps) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_cvtepi32_ps) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, ) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcvtpd2dq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtpd2dq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtpd2dq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtpd2dq\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtpd2dq\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtpd2dq\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d s;
+volatile __m256i res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtpd_epi32 (s);
+ res = _mm512_mask_cvtpd_epi32 (res, m, s);
+ res = _mm512_maskz_cvtpd_epi32 (m, s);
+ res = _mm512_cvt_roundpd_epi32 (s, _MM_FROUND_TO_NEAREST_INT);
+ res = _mm512_mask_cvt_roundpd_epi32 (res, m, s, _MM_FROUND_TO_POS_INF);
+ res = _mm512_maskz_cvt_roundpd_epi32 (m, s, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SRC_SIZE ((AVX512F_LEN) / 64)
+#include "avx512f-mask-type.h"
+#define DST_SIZE ((AVX512F_LEN_HALF) / 32)
+
+static void
+CALC (double *s, unsigned *r)
+{
+ int i;
+
+ for (i = 0; i < SRC_SIZE; i++)
+ {
+ r[i] = (s[i] >= 0) ? (int) (s[i] + 0.5)
+ : (int) (s[i] - 0.5);
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s;
+ UNION_TYPE (AVX512F_LEN_HALF, i_d) res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[DST_SIZE] = { 0 };
+ int i, sign = 1;
+
+ for (i = 0; i < SRC_SIZE; i++)
+ {
+ s.a[i] = 123.456 * (i + 2000) * sign;
+ sign = -sign;
+ }
+
+ for (i = 0; i < DST_SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_cvtpd_epi32) (s.x);
+ res2.x = INTRINSIC (_mask_cvtpd_epi32) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_cvtpd_epi32) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SRC_SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SRC_SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vcvtpd2ps\[ \\t\]+\[^\n\]*%ymm\[0-9\]" 6 } } */
+/* { dg-final { scan-assembler-times "vcvtpd2ps\[ \\t\]+\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtpd2ps\[ \\t\]+\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtpd2ps\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%ymm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtpd2ps\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtpd2ps\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __m256 y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm512_cvtpd_ps (x);
+ y = _mm512_mask_cvtpd_ps (y, 4, x);
+ y = _mm512_maskz_cvtpd_ps (6, x);
+ y = _mm512_cvt_roundpd_ps (x, _MM_FROUND_TO_NEAREST_INT);
+ y = _mm512_mask_cvt_roundpd_ps (y, 4, x, _MM_FROUND_TO_NEG_INF);
+ y = _mm512_maskz_cvt_roundpd_ps (6, x, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+void static
+CALC (float *e, UNION_TYPE (AVX512F_LEN, d) s1)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ e[i] = (float) s1.a[i];
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s1;
+ UNION_TYPE (AVX512F_LEN_HALF,) u1, u2, u3;
+ MASK_TYPE mask = MASK_VALUE;
+ float e[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 0.12 * (i + 37.09);
+ u1.a[i] = DEFAULT_VALUE;
+ u2.a[i] = DEFAULT_VALUE;
+ u3.a[i] = DEFAULT_VALUE;
+ }
+
+ u1.x = INTRINSIC (_cvtpd_ps) (s1.x);
+ u2.x = INTRINSIC (_mask_cvtpd_ps) (u2.x, mask, s1.x);
+ u3.x = INTRINSIC (_maskz_cvtpd_ps) (mask, s1.x);
+
+ CALC (e, s1);
+
+ if (UNION_CHECK (AVX512F_LEN_HALF,) (u1, e))
+ abort ();
+
+ MASK_MERGE ()(e, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF,) (u2, e))
+ abort ();
+
+ MASK_ZERO ()(e, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF,) (u3, e))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcvtpd2udq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtpd2udq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtpd2udq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtpd2udq\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtpd2udq\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtpd2udq\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d s;
+volatile __m256i res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtpd_epu32 (s);
+ res = _mm512_mask_cvtpd_epu32 (res, m, s);
+ res = _mm512_maskz_cvtpd_epu32 (m, s);
+ res = _mm512_cvt_roundpd_epu32 (s, _MM_FROUND_TO_NEAREST_INT);
+ res = _mm512_mask_cvt_roundpd_epu32 (res, m, s, _MM_FROUND_TO_POS_INF);
+ res = _mm512_maskz_cvt_roundpd_epu32 (m, s, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SRC_SIZE ((AVX512F_LEN) / 64)
+#include "avx512f-mask-type.h"
+#define DST_SIZE ((AVX512F_LEN_HALF) / 32)
+
+static void
+CALC (double *s, unsigned *r)
+{
+ int i;
+
+ for (i = 0; i < DST_SIZE; i++)
+ {
+ r[i] = (unsigned) (s[i] + 0.5);
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s;
+ UNION_TYPE (AVX512F_LEN_HALF, i_d) res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned res_ref[DST_SIZE] = { 0 };
+ int i;
+
+ for (i = 0; i < SRC_SIZE; i++)
+ {
+ s.a[i] = 123.456 * (i + 2000);
+ }
+
+ for (i = 0; i < DST_SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_cvtpd_epu32) (s.x);
+ res2.x = INTRINSIC (_mask_cvtpd_epu32) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_cvtpd_epu32) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SRC_SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SRC_SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vcvtph2ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 6 } } */
+/* { dg-final { scan-assembler-times "vcvtph2ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtph2ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtph2ps\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vcvtph2ps\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtph2ps\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m256i x;
+volatile __m512 y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm512_cvtph_ps (x);
+ y = _mm512_mask_cvtph_ps (y, 4, x);
+ y = _mm512_maskz_cvtph_ps (6, x);
+ y = _mm512_cvt_roundph_ps (x, _MM_FROUND_NO_EXC);
+ y = _mm512_mask_cvt_roundph_ps (y, 4, x, _MM_FROUND_NO_EXC);
+ y = _mm512_maskz_cvt_roundph_ps (6, x, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN_HALF, i_w) val;
+ UNION_TYPE (AVX512F_LEN,) res1,res2,res3;
+ MASK_TYPE mask = MASK_VALUE;
+ float exp[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ res1.a[i] = DEFAULT_VALUE;
+ res2.a[i] = DEFAULT_VALUE;
+ res3.a[i] = DEFAULT_VALUE;
+ }
+
+ exp[0] = 1;
+ exp[1] = 2;
+ exp[2] = 4;
+ exp[3] = 8;
+#if AVX512F_LEN > 128
+ exp[4] = -1;
+ exp[5] = -2;
+ exp[6] = -4;
+ exp[7] = -8;
+#endif
+#if AVX512F_LEN > 256
+ exp[8] = 1;
+ exp[9] = 2;
+ exp[10] = 4;
+ exp[11] = 8;
+ exp[12] = -1;
+ exp[13] = -2;
+ exp[14] = -4;
+ exp[15] = -8;
+#endif
+
+ val.a[0] = 0x3c00;
+ val.a[1] = 0x4000;
+ val.a[2] = 0x4400;
+ val.a[3] = 0x4800;
+#if AVX512F_LEN > 128
+ val.a[4] = 0xbc00;
+ val.a[5] = 0xc000;
+ val.a[6] = 0xc400;
+ val.a[7] = 0xc800;
+#endif
+#if AVX512F_LEN > 256
+ val.a[8] = 0x3c00;
+ val.a[9] = 0x4000;
+ val.a[10] = 0x4400;
+ val.a[11] = 0x4800;
+ val.a[12] = 0xbc00;
+ val.a[13] = 0xc000;
+ val.a[14] = 0xc400;
+ val.a[15] = 0xc800;
+#endif
+
+ res1.x = _mm512_cvtph_ps (val.x);
+ res2.x = _mm512_mask_cvtph_ps (res2.x, mask, val.x);
+ res3.x = _mm512_maskz_cvtph_ps (mask, val.x);
+
+ if (UNION_CHECK (AVX512F_LEN,) (res1, exp))
+ abort ();
+
+ MASK_MERGE () (exp, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res2, exp))
+ abort ();
+
+ MASK_ZERO () (exp, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res3, exp))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcvtps2dq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[\\n\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtps2dq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtps2dq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtps2dq\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtps2dq\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtps2dq\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 s;
+volatile __m512i res;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtps_epi32 (s);
+ res = _mm512_mask_cvtps_epi32 (res, m, s);
+ res = _mm512_maskz_cvtps_epi32 (m, s);
+ res = _mm512_cvt_roundps_epi32 (s, _MM_FROUND_TO_NEAREST_INT);
+ res = _mm512_mask_cvt_roundps_epi32 (res, m, s, _MM_FROUND_TO_POS_INF);
+ res = _mm512_maskz_cvt_roundps_epi32 (m, s, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (int *r, float *s)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ r[i] = (s[i] >= 0) ? (int) (s[i] + 0.5) : (int) (s[i] - 0.5);
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN,) src;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ res2.a[i] = DEFAULT_VALUE;
+ src.a[i] = 1.5 + 34.67 * i * sign;
+ sign = sign * -1;
+ }
+
+ res1.x = INTRINSIC (_cvtps_epi32) (src.x);
+ res2.x = INTRINSIC (_mask_cvtps_epi32) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_cvtps_epi32) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcvtps2pd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtps2pd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtps2pd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtps2pd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtps2pd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtps2pd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m256 s;
+volatile __m512d res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtps_pd (s);
+ res = _mm512_mask_cvtps_pd (res, m, s);
+ res = _mm512_maskz_cvtps_pd (m, s);
+ res = _mm512_cvt_roundps_pd (s, _MM_FROUND_NO_EXC);
+ res = _mm512_mask_cvt_roundps_pd (res, m, s, _MM_FROUND_NO_EXC);
+ res = _mm512_maskz_cvt_roundps_pd (m, s, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SRC_SIZE ((AVX512F_LEN_HALF) / 32)
+#include "avx512f-mask-type.h"
+#define DST_SIZE ((AVX512F_LEN) / 64)
+
+static void
+CALC (float *s, double *r)
+{
+ int i;
+
+ for (i = 0; i < DST_SIZE; i++)
+ {
+ r[i] = (double) s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN_HALF, ) s;
+ UNION_TYPE (AVX512F_LEN, d) res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[DST_SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SRC_SIZE; i++)
+ {
+ s.a[i] = 123.456 * (i + 2000) * sign;
+ sign = -sign;
+ }
+
+ for (i = 0; i < DST_SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_cvtps_pd) (s.x);
+ res2.x = INTRINSIC (_mask_cvtps_pd) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_cvtps_pd) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, DST_SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, DST_SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vcvtps2ph\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vcvtps2ph\[ \\t\]+\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtps2ph\[ \\t\]+\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __m256i y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm512_cvtps_ph (x, 0);
+ y = _mm512_maskz_cvtps_ph (4, x, 0);
+ y = _mm512_mask_cvtps_ph (y, 2, x, 0);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN,) val;
+ UNION_TYPE (AVX512F_LEN_HALF, i_w) res1,res2,res3;
+ MASK_TYPE mask = MASK_VALUE;
+ short exp[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ res1.a[i] = DEFAULT_VALUE;
+ res2.a[i] = DEFAULT_VALUE;
+ res3.a[i] = DEFAULT_VALUE;
+ }
+
+ val.a[0] = 1;
+ val.a[1] = 2;
+ val.a[2] = 4;
+ val.a[3] = 8;
+#if AVX512F_LEN > 128
+ val.a[4] = -1;
+ val.a[5] = -2;
+ val.a[6] = -4;
+ val.a[7] = -8;
+#endif
+#if AVX512F_LEN > 256
+ val.a[8] = 1;
+ val.a[9] = 2;
+ val.a[10] = 4;
+ val.a[11] = 8;
+ val.a[12] = -1;
+ val.a[13] = -2;
+ val.a[14] = -4;
+ val.a[15] = -8;
+#endif
+
+ exp[0] = 0x3c00;
+ exp[1] = 0x4000;
+ exp[2] = 0x4400;
+ exp[3] = 0x4800;
+#if AVX512F_LEN > 128
+ exp[4] = 0xbc00;
+ exp[5] = 0xc000;
+ exp[6] = 0xc400;
+ exp[7] = 0xc800;
+#endif
+#if AVX512F_LEN > 256
+ exp[8] = 0x3c00;
+ exp[9] = 0x4000;
+ exp[10] = 0x4400;
+ exp[11] = 0x4800;
+ exp[12] = 0xbc00;
+ exp[13] = 0xc000;
+ exp[14] = 0xc400;
+ exp[15] = 0xc800;
+#endif
+
+ res1.x = _mm512_cvtps_ph (val.x, 0);
+ res2.x = _mm512_mask_cvtps_ph (res2.x, mask, val.x, 0);
+ res3.x = _mm512_maskz_cvtps_ph (mask, val.x, 0);
+
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_w) (res1, exp))
+ abort ();
+
+ MASK_MERGE (i_w) (exp, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_w) (res2, exp))
+ abort ();
+
+ MASK_ZERO (i_w) (exp, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_w) (res3, exp))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcvtps2udq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[\\n\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtps2udq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtps2udq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtps2udq\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtps2udq\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtps2udq\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 s;
+volatile __m512i res;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtps_epu32 (s);
+ res = _mm512_mask_cvtps_epu32 (res, m, s);
+ res = _mm512_maskz_cvtps_epu32 (m, s);
+ res = _mm512_cvt_roundps_epu32 (s, _MM_FROUND_TO_NEAREST_INT);
+ res = _mm512_mask_cvt_roundps_epu32 (res, m, s, _MM_FROUND_TO_NEG_INF);
+ res = _mm512_maskz_cvt_roundps_epu32 (m, s, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (unsigned *r, float *s)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ r[i] = (unsigned) (s[i] + 0.5);
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN,) src;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned res_ref[SIZE];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src.a[i] = 1.5 + 34.67 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtps_epu32) (src.x);
+ res2.x = INTRINSIC (_mask_cvtps_epu32) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_cvtps_epu32) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vcvtsd2si\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+#include <immintrin.h>
+
+volatile __m128d x;
+volatile unsigned y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm_cvt_roundsd_i32 (x, _MM_FROUND_TO_NEAREST_INT);
+}
--- /dev/null
+/* { dg-do compile { target { ! { ia32 } } } } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vcvtsd2siq\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128d x;
+volatile unsigned long long y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm_cvt_roundsd_i64 (x, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vcvtsd2usi\[ \\t\]+\[^\n\]*%xmm\[0-9\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtsd2usi\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+#include <immintrin.h>
+
+volatile __m128d x;
+volatile unsigned y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm_cvtsd_u32 (x);
+ y = _mm_cvt_roundsd_u32 (x, _MM_FROUND_TO_NEG_INF);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+void static
+avx512f_test (void)
+{
+ union128d s1;
+ unsigned int d;
+ unsigned int e;
+
+ s1.x = _mm_set_pd (24.43, 68.346);
+ d = _mm_cvtsd_u32 (s1.x);
+ e = (unsigned int)(s1.a[0] + 0.5);
+
+ if (e != d)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile { target { ! { ia32 } } } } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vcvtsd2usi\[ \\t\]+\[^\n\]*%xmm\[0-9\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtsd2usi\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128d x;
+volatile unsigned long long y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm_cvtsd_u64 (x);
+ y = _mm_cvt_roundsd_u64 (x, _MM_FROUND_TO_POS_INF);
+}
--- /dev/null
+/* { dg-do run { target { ! { ia32 } } } } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+void static
+avx512f_test (void)
+{
+ union128d s1;
+ unsigned long long d;
+ unsigned long long e;
+
+ s1.x = _mm_set_pd (24.43, 68.346);
+ d = _mm_cvtsd_u64 (s1.x);
+ e = (unsigned long long)(s1.a[0] + 0.5);
+
+ if (e != d)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile { target { ! { ia32 } } } } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcvtsi2sdq\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128d x;
+volatile long long n;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm_cvt_roundi64_sd (x, n, _MM_FROUND_TO_POS_INF);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcvtsi2ss\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128 x;
+volatile int n;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm_cvt_roundi32_ss (x, n, _MM_FROUND_TO_NEAREST_INT);
+}
--- /dev/null
+/* { dg-do compile { target { ! { ia32 } } } } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcvtsi2ssq\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128 x;
+volatile long long n;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm_cvt_roundi64_ss (x, n, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vcvtss2si\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+#include <immintrin.h>
+
+volatile __m128 x;
+volatile unsigned y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm_cvt_roundss_i32 (x, _MM_FROUND_TO_NEAREST_INT);
+}
--- /dev/null
+/* { dg-do compile { target { ! { ia32 } } } } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vcvtss2siq\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128 x;
+volatile unsigned long long y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm_cvt_roundss_i64 (x, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vcvtss2usi\[ \\t\]+\[^\n\]*%xmm\[0-9\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtss2usi\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+#include <immintrin.h>
+
+volatile __m128 x;
+volatile unsigned y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm_cvtss_u32 (x);
+ y = _mm_cvt_roundss_u32 (x, _MM_FROUND_TO_NEG_INF);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+void static
+avx512f_test (void)
+{
+ union128 s1;
+ unsigned int d;
+ unsigned int e;
+
+ s1.x = _mm_set_ps (24.43, 68.346, 35.7765, 34508.51);
+ d = _mm_cvtss_u32 (s1.x);
+ e = (unsigned int)(s1.a[0] + 0.5);
+
+ if (e != d)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile { target { ! { ia32 } } } } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vcvtss2usi\[ \\t\]+\[^\n\]*%xmm\[0-9\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtss2usi\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128 x;
+volatile unsigned long long y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm_cvtss_u64 (x);
+ y = _mm_cvt_roundss_u64 (x, _MM_FROUND_TO_POS_INF);
+}
--- /dev/null
+/* { dg-do run { target { ! { ia32 } } } } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+void static
+avx512f_test (void)
+{
+ union128 s1;
+ unsigned long long d;
+ unsigned long long e;
+
+ s1.x = _mm_set_ps (24.43, 68.346, 12.34, 80.67);
+ d = _mm_cvtss_u64 (s1.x);
+ e = (unsigned long long)(s1.a[0] + 0.5);
+
+ if (e != d)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcvttpd2dq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvttpd2dq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvttpd2dq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vcvttpd2dq\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvttpd2dq\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvttpd2dq\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d s;
+volatile __m256i res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvttpd_epi32 (s);
+ res = _mm512_mask_cvttpd_epi32 (res, m, s);
+ res = _mm512_maskz_cvttpd_epi32 (m, s);
+ res = _mm512_cvtt_roundpd_epi32 (s, _MM_FROUND_NO_EXC);
+ res = _mm512_mask_cvtt_roundpd_epi32 (res, m, s, _MM_FROUND_NO_EXC);
+ res = _mm512_maskz_cvtt_roundpd_epi32 (m, s, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SRC_SIZE ((AVX512F_LEN) / 64)
+#include "avx512f-mask-type.h"
+#define DST_SIZE ((AVX512F_LEN_HALF) / 32)
+
+static void
+CALC (double *s, int *r)
+{
+ int i;
+
+ for (i = 0; i < SRC_SIZE; i++)
+ {
+ r[i] = (int) s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s;
+ UNION_TYPE (AVX512F_LEN_HALF, i_d) res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[DST_SIZE] = { 0 };
+ int i, sign = 1;
+
+ for (i = 0; i < SRC_SIZE; i++)
+ {
+ s.a[i] = 123.456 * (i + 2000) * sign;
+ sign = -sign;
+ }
+
+ for (i = 0; i < DST_SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_cvttpd_epi32) (s.x);
+ res2.x = INTRINSIC (_mask_cvttpd_epi32) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_cvttpd_epi32) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SRC_SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SRC_SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcvttpd2udq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvttpd2udq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvttpd2udq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vcvttpd2udq\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvttpd2udq\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvttpd2udq\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d s;
+volatile __m256i res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvttpd_epu32 (s);
+ res = _mm512_mask_cvttpd_epu32 (res, m, s);
+ res = _mm512_maskz_cvttpd_epu32 (m, s);
+ res = _mm512_cvtt_roundpd_epu32 (s, _MM_FROUND_NO_EXC);
+ res = _mm512_mask_cvtt_roundpd_epu32 (res, m, s, _MM_FROUND_NO_EXC);
+ res = _mm512_maskz_cvtt_roundpd_epu32 (m, s, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SRC_SIZE ((AVX512F_LEN) / 64)
+#include "avx512f-mask-type.h"
+#define DST_SIZE ((AVX512F_LEN_HALF) / 32)
+
+static void
+CALC (double *s, unsigned *r)
+{
+ int i;
+
+ for (i = 0; i < DST_SIZE; i++)
+ {
+ r[i] = (unsigned) s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s;
+ UNION_TYPE (AVX512F_LEN_HALF, i_d) res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned res_ref[DST_SIZE] = { 0 };
+ int i;
+
+ for (i = 0; i < SRC_SIZE; i++)
+ {
+ s.a[i] = 123.456 * (i + 2000);
+ }
+
+ for (i = 0; i < DST_SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_cvttpd_epu32) (s.x);
+ res2.x = INTRINSIC (_mask_cvttpd_epu32) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_cvttpd_epu32) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SRC_SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SRC_SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcvttps2dq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[\\n\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvttps2dq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvttps2dq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vcvttps2dq\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[\\n\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvttps2dq\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvttps2dq\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 s;
+volatile __m512i res;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvttps_epi32 (s);
+ res = _mm512_mask_cvttps_epi32 (res, m, s);
+ res = _mm512_maskz_cvttps_epi32 (m, s);
+ res = _mm512_cvtt_roundps_epi32 (s, _MM_FROUND_NO_EXC);
+ res = _mm512_mask_cvtt_roundps_epi32 (res, m, s, _MM_FROUND_NO_EXC);
+ res = _mm512_maskz_cvtt_roundps_epi32 (m, s, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (int *r, float *s)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ r[i] = (int) s[i];
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN,) src;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ res2.a[i] = DEFAULT_VALUE;
+ src.a[i] = 1.5 + 34.67 * i * sign;
+ sign = sign * -1;
+ }
+
+ res1.x = INTRINSIC (_cvttps_epi32) (src.x);
+ res2.x = INTRINSIC (_mask_cvttps_epi32) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_cvttps_epi32) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcvttps2udq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[\\n\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvttps2udq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvttps2udq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vcvttps2udq\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[\\n\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvttps2udq\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvttps2udq\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 s;
+volatile __m512i res;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvttps_epu32 (s);
+ res = _mm512_mask_cvttps_epu32 (res, m, s);
+ res = _mm512_maskz_cvttps_epu32 (m, s);
+ res = _mm512_cvtt_roundps_epu32 (s, _MM_FROUND_NO_EXC);
+ res = _mm512_mask_cvtt_roundps_epu32 (res, m, s, _MM_FROUND_NO_EXC);
+ res = _mm512_maskz_cvtt_roundps_epu32 (m, s, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (unsigned *r, float *s)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ r[i] = (unsigned) s[i];
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN,) src;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned res_ref[SIZE];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src.a[i] = 1.5 + 34.67 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvttps_epu32) (src.x);
+ res2.x = INTRINSIC (_mask_cvttps_epu32) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_cvttps_epu32) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vcvttsd2si\[ \\t\]+\[^\n\]*%xmm\[0-9\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvttsd2si\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+#include <immintrin.h>
+
+volatile __m128d x;
+volatile y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm_cvttsd_i32 (x);
+ y = _mm_cvtt_roundsd_i32 (x, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+#include "avx512f-mask-type.h"
+#include "avx512f-helper.h"
+
+static int
+__attribute__ ((noinline, unused))
+test (__m128d x)
+{
+ return _mm_cvttsd_i32 (x);
+}
+
+static void
+avx512f_test (void)
+{
+ union128d s1;
+ int res, res_ref;
+
+ s1.x = _mm_set_pd (123.321, 456.987);
+ res = test (s1.x);
+ res_ref = (int) s1.a[0];
+
+ if (res != res_ref)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile { target { ! { ia32 } } } } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vcvttsd2siq\[ \\t\]+\[^\n\]*%xmm\[0-9\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvttsd2siq\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128d x;
+volatile long long y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm_cvttsd_i64 (x);
+ y = _mm_cvtt_roundsd_i64 (x, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run { target { ! { ia32 } } } } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+#include "avx512f-mask-type.h"
+#include "avx512f-helper.h"
+
+static int
+__attribute__ ((noinline, unused))
+test (__m128d x)
+{
+ return _mm_cvttsd_i64 (x);
+}
+
+static void
+avx512f_test (void)
+{
+ union128d s1;
+ long long res, res_ref;
+
+ s1.x = _mm_set_pd (123.321, 456.987);
+ res = test (s1.x);
+ res_ref = (long long) s1.a[0];
+
+ if (res != res_ref)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vcvttsd2usi\[ \\t\]+\[^\n\]*%xmm\[0-9\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvttsd2usi\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+#include <immintrin.h>
+
+volatile __m128d x;
+volatile unsigned y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm_cvttsd_u32 (x);
+ y = _mm_cvtt_roundsd_u32 (x, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static unsigned int
+__attribute__((noinline, unused))
+test (union128d s1)
+{
+ return _mm_cvttsd_u32 (s1.x);
+}
+
+void static
+avx512f_test (void)
+{
+ union128d s1;
+ unsigned int d;
+ unsigned int e;
+
+ s1.x = _mm_set_pd (24.43, 68.346);
+ d = test (s1);
+ e = (unsigned int)s1.a[0];
+
+ if (e != d)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile { target { ! { ia32 } } } } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vcvttsd2usi\[ \\t\]+\[^\n\]*%xmm\[0-9\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvttsd2usi\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128d x;
+volatile unsigned long long y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm_cvttsd_u64 (x);
+ y = _mm_cvtt_roundsd_u64 (x, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run { target { ! { ia32 } } } } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static unsigned long long
+__attribute__((noinline, unused))
+test (union128d s1)
+{
+ return _mm_cvttsd_u64 (s1.x);
+}
+
+void static
+avx512f_test (void)
+{
+ union128d s1;
+ unsigned long long d;
+ unsigned long long e;
+
+ s1.x = _mm_set_pd (24.43, 68.346);
+ d = test (s1);
+ e = (unsigned long long)s1.a[0];
+
+ if (e != d)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vcvttss2si\[ \\t\]+\[^\n\]*%xmm\[0-9\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvttss2si\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+#include <immintrin.h>
+
+volatile __m128 x;
+volatile y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm_cvttss_i32 (x);
+ y = _mm_cvtt_roundss_i32 (x, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+#include "avx512f-mask-type.h"
+#include "avx512f-helper.h"
+
+static int
+__attribute__ ((noinline, unused))
+test (__m128 x)
+{
+ return _mm_cvttss_i32 (x);
+}
+
+static void
+avx512f_test (void)
+{
+ union128 s1;
+ int res, res_ref;
+
+ s1.x = _mm_set_ps (24.43, 68.346, 43.35, 546.46);
+ res = test (s1.x);
+ res_ref = (int) s1.a[0];
+
+ if (res != res_ref)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile { target { ! { ia32 } } } } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vcvttss2siq\[ \\t\]+\[^\n\]*%xmm\[0-9\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvttss2siq\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128 x;
+volatile long long y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm_cvttss_i64 (x);
+ y = _mm_cvtt_roundss_i64 (x, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run { target { ! { ia32 } } } } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+#include "avx512f-mask-type.h"
+#include "avx512f-helper.h"
+
+static int
+__attribute__ ((noinline, unused))
+test (__m128 x)
+{
+ return _mm_cvttss_i64 (x);
+}
+
+static void
+avx512f_test (void)
+{
+ union128 s1;
+ long long res, res_ref;
+
+ s1.x = _mm_set_ps (24.43, 68.346, 43.35, 546.46);
+ res = test (s1.x);
+ res_ref = (long long) s1.a[0];
+
+ if (res != res_ref)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vcvttss2usi\[ \\t\]+\[^\n\]*%xmm\[0-9\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvttss2usi\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+#include <immintrin.h>
+
+volatile __m128 x;
+volatile unsigned y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm_cvttss_u32 (x);
+ y = _mm_cvtt_roundss_u32 (x, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static unsigned int
+__attribute__((noinline, unused))
+test (union128 s1)
+{
+ return _mm_cvttss_u32 (s1.x);
+}
+
+void static
+avx512f_test (void)
+{
+ union128 s1;
+ unsigned int d;
+ unsigned int e;
+
+ s1.x = _mm_set_ps (24.43, 68.346, 45.12, 90.97);
+ d = test (s1);
+ e = (unsigned int)s1.a[0];
+
+ if (e != d)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile { target { ! { ia32 } } } } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vcvttss2usi\[ \\t\]+\[^\n\]*%xmm\[0-9\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvttss2usi\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128 x;
+volatile unsigned long long y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm_cvttss_u64 (x);
+ y = _mm_cvtt_roundss_u64 (x, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run { target { ! { ia32 } } } } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static unsigned long long
+__attribute__((noinline, unused))
+test (union128 s1)
+{
+ return _mm_cvttss_u64 (s1.x);
+}
+
+void static
+avx512f_test (void)
+{
+ union128 s1;
+ unsigned long long d;
+ unsigned long long e;
+
+ s1.x = _mm_set_ps (24.43, 68.346, 10.756, 89.145);
+ d = test (s1);
+ e = (unsigned long long)s1.a[0];
+
+ if (e != d)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcvtudq2pd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtudq2pd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtudq2pd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m256i s;
+volatile __m512d res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtepu32_pd (s);
+ res = _mm512_mask_cvtepu32_pd (res, m, s);
+ res = _mm512_maskz_cvtepu32_pd (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SRC_SIZE ((AVX512F_LEN_HALF) / 32)
+#include "avx512f-mask-type.h"
+#define DST_SIZE ((AVX512F_LEN) / 64)
+
+static void
+CALC (unsigned *s, double *r)
+{
+ int i;
+
+ for (i = 0; i < DST_SIZE; i++)
+ {
+ r[i] = (double) s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN_HALF, i_d) s;
+ UNION_TYPE (AVX512F_LEN, d) res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[DST_SIZE];
+ int i;
+
+ for (i = 0; i < SRC_SIZE; i++)
+ {
+ s.a[i] = 123456 * (i + 2000);
+ }
+
+ for (i = 0; i < DST_SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_cvtepu32_pd) (s.x);
+ res2.x = INTRINSIC (_mask_cvtepu32_pd) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_cvtepu32_pd) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, DST_SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, DST_SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcvtudq2ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vcvtudq2ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtudq2ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtudq2ps\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtudq2ps\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtudq2ps\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i s;
+volatile __m512 res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtepu32_ps (s);
+ res = _mm512_mask_cvtepu32_ps (res, m, s);
+ res = _mm512_maskz_cvtepu32_ps (m, s);
+ res = _mm512_cvt_roundepu32_ps (s, _MM_FROUND_TO_NEAREST_INT);
+ res = _mm512_mask_cvt_roundepu32_ps (res, m, s, _MM_FROUND_TO_NEG_INF);
+ res = _mm512_maskz_cvt_roundepu32_ps (m, s, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (unsigned *s, float *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = (float) s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) s;
+ UNION_TYPE (AVX512F_LEN, ) res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 123456 * (i + 2000);
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtepu32_ps) (s.x);
+ res2.x = INTRINSIC (_mask_cvtepu32_ps) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_cvtepu32_ps) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, ) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler "vcvtusi2sd\[ \\t\]+\[^\n\]*%xmm\[0-9\]" } } */
+
+#include <immintrin.h>
+
+volatile __m128d x;
+volatile unsigned n;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm_cvtu32_sd (x, n);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static void
+ __attribute__ ((noinline, unused))
+compute_vcvtusi2sd (double *s1, unsigned s2, double *r)
+{
+ r[0] = (double) s2;
+ r[1] = s1[1];
+}
+
+static void
+avx512f_test (void)
+{
+ union128d s1, res;
+ unsigned s2;
+ double res_ref[2];
+
+ s1.x = _mm_set_pd (-24.43, -43.35);
+ s2 = 0xFEDCA987;
+
+ res.x = _mm_cvtu32_sd (s1.x, s2);
+
+ compute_vcvtusi2sd (s1.a, s2, res_ref);
+
+ if (check_union128d (res, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile { target { ! { ia32 } } } } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcvtusi2sd\[ \\t\]+\[^\n\]*%xmm\[0-9\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtusi2sd\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128d x;
+volatile unsigned long long n;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm_cvtu64_sd (x, n);
+ x = _mm_cvt_roundu64_sd (x, n, _MM_FROUND_TO_POS_INF);
+}
--- /dev/null
+/* { dg-do run { target { ! { ia32 } } } } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static void
+ __attribute__ ((noinline, unused))
+compute_vcvtusi2sd (double *s1, unsigned long long s2, double *r)
+{
+ r[0] = (double) s2;
+ r[1] = s1[1];
+}
+
+static void
+avx512f_test (void)
+{
+ union128d s1, res;
+ unsigned long long s2;
+ double res_ref[4];
+
+ s1.x = _mm_set_pd (-24.43, -43.35);
+ s2 = 0xFEDCBA9876543210;
+
+ res.x = _mm_cvtu64_sd (s1.x, s2);
+
+ compute_vcvtusi2sd (s1.a, s2, res_ref);
+
+ if (check_union128d (res, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcvtusi2ss\[ \\t\]+\[^\n\]*%xmm\[0-9\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtusi2ss\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128 x;
+volatile unsigned n;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm_cvtu32_ss (x, n);
+ x = _mm_cvt_roundu32_ss (x, n, _MM_FROUND_TO_NEAREST_INT);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static void
+ __attribute__ ((noinline, unused))
+compute_vcvtusi2ss (float *s1, unsigned s2, float *r)
+{
+ r[0] = (float) s2;
+ r[1] = s1[1];
+ r[2] = s1[2];
+ r[3] = s1[3];
+}
+
+static void
+avx512f_test (void)
+{
+ union128 s1, res;
+ unsigned s2;
+ float res_ref[4];
+
+ s1.x = _mm_set_ps (-24.43, 68.346, -43.35, 546.46);
+ s2 = 0xFEDCA987;
+
+ res.x = _mm_cvtu32_ss (s1.x, s2);
+
+ compute_vcvtusi2ss (s1.a, s2, res_ref);
+
+ if (check_union128 (res, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile { target { ! { ia32 } } } } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vcvtusi2ss\[ \\t\]+\[^\n\]*%xmm\[0-9\]" 2 } } */
+/* { dg-final { scan-assembler-times "vcvtusi2ss\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%xmm\[0-9\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128 x;
+volatile unsigned long long n;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm_cvtu64_ss (x, n);
+ x = _mm_cvt_roundu64_ss (x, n, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run { target { ! { ia32 } } } } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+static void
+ __attribute__ ((noinline, unused))
+compute_vcvtusi2ss (float *s1, unsigned long long s2, float *r)
+{
+ r[0] = (float) s2;
+ r[1] = s1[1];
+ r[2] = s1[2];
+ r[3] = s1[3];
+}
+
+static void
+avx512f_test (void)
+{
+ union128 s1, res;
+ unsigned long long s2;
+ float res_ref[4];
+
+ s1.x = _mm_set_ps (-24.43, 68.346, -43.35, 546.46);
+ s2 = 0xFEDCBA9876543210;
+
+ res.x = _mm_cvtu64_ss (s1.x, s2);
+
+ compute_vcvtusi2ss (s1.a, s2, res_ref);
+
+ if (check_union128 (res, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vdivpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vdivpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vdivpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vdivpd\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vdivpd\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vdivpd\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_div_pd (x, x);
+ x = _mm512_mask_div_pd (x, m, x, x);
+ x = _mm512_maskz_div_pd (m, x, x);
+ x = _mm512_div_round_pd (x, x, _MM_FROUND_TO_NEAREST_INT);
+ x = _mm512_mask_div_round_pd (x, m, x, x, _MM_FROUND_TO_NEG_INF);
+ x = _mm512_maskz_div_round_pd (m, x, x, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (double *r, double *s1, double *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[i] / s2[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, d) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 1.5 + 34.67 * i * sign;
+ src2.a[i] = -22.17 * i * sign + 1.0;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_div_pd) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_div_pd) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_div_pd) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vdivps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vdivps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vdivps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vdivps\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vdivps\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vdivps\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_div_ps (x, x);
+ x = _mm512_mask_div_ps (x, m, x, x);
+ x = _mm512_maskz_div_ps (m, x, x);
+ x = _mm512_div_round_ps (x, x, _MM_FROUND_TO_NEAREST_INT);
+ x = _mm512_mask_div_round_ps (x, m, x, x, _MM_FROUND_TO_POS_INF);
+ x = _mm512_maskz_div_round_ps (m, x, x, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (float *r, float *s1, float *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[i] / s2[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN,) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 1.5 + 34.67 * i * sign;
+ src2.a[i] = -22.17 * i * sign + 1.0;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_div_ps) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_div_ps) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_div_ps) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN,) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O3 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vmovdqa64\[ \\t\]+%zmm" 2 } } */
+/* { dg-final { scan-assembler-times "vpbroadcastd" 1 } } */
+/* { dg-final { scan-assembler-times "vpbroadcastq" 1 } } */
+/* { dg-final { scan-assembler-times "vpbroadcastb" 2 } } */
+/* { dg-final { scan-assembler-times "vpbroadcastw" 2 } } */
+/* { dg-final { scan-assembler-times "vbroadcastss" 1 } } */
+/* { dg-final { scan-assembler-times "vbroadcastsd" 1 } } */
+
+#include <x86intrin.h>
+
+typedef char __v64qi __attribute__ ((vector_size (64)));
+typedef short __v32hi __attribute__ ((vector_size (64)));
+
+__v64qi foo_1 (char c)
+{
+ __v64qi v1 = {
+ c, c, c, c, c, c, c, c,
+ c, c, c, c, c, c, c, c,
+ c, c, c, c, c, c, c, c,
+ c, c, c, c, c, c, c, c,
+ c, c, c, c, c, c, c, c,
+ c, c, c, c, c, c, c, c,
+ c, c, c, c, c, c, c, c,
+ c, c, c, c, c, c, c, c
+ };
+
+ return v1;
+}
+
+__v32hi foo_2 (short c)
+{
+ __v32hi v1 = {
+ c, c, c, c, c, c, c, c,
+ c, c, c, c, c, c, c, c,
+ c, c, c, c, c, c, c, c,
+ c, c, c, c, c, c, c, c
+ };
+
+ return v1;
+}
+
+__v16si foo_3 (int c)
+{
+ __v16si v1 = {
+ c, c, c, c, c, c, c, c,
+ c, c, c, c, c, c, c, c
+ };
+
+ return v1;
+}
+
+__v8di foo_4 (long long c)
+{
+ __v8di v1 = {
+ c, c, c, c, c, c, c, c
+ };
+
+ return v1;
+}
+
+__v32qi foo_5 (char c)
+{
+ __v32qi v1 = {
+ c, c, c, c, c, c, c, c,
+ c, c, c, c, c, c, c, c,
+ c, c, c, c, c, c, c, c,
+ c, c, c, c, c, c, c, c
+ };
+
+ return v1;
+}
+
+__v16hi foo_6 (short c)
+{
+ __v16hi v1 = {
+ c, c, c, c, c, c, c, c,
+ c, c, c, c, c, c, c, c
+ };
+
+ return v1;
+}
+
+__v8si foo_7 (int c)
+{
+ __v8si v1 = {
+ c, c, c, c, c, c, c, c
+ };
+
+ return v1;
+}
+
+__v4di foo_8 (long long c)
+{
+ __v4di v1 = {
+ c, c, c, c
+ };
+
+ return v1;
+}
+
+
+__v16qi foo_9 (char c)
+{
+ __v16qi v1 = {
+ c, c, c, c, c, c, c, c,
+ c, c, c, c, c, c, c, c
+ };
+
+ return v1;
+}
+
+__v8hi foo_10(short c)
+{
+ __v8hi v1 = {
+ c, c, c, c, c, c, c, c
+ };
+
+ return v1;
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O3 -mavx512f" } */
+
+long long *D;
+int *S;
+short *H;
+char *Q;
+
+long long foo_unpack_1 (int low, int high, int ind)
+{
+ int i;
+
+ for (i = low; i <= high; i++)
+ D[i] *= S[i];
+
+ return D[ind];
+}
+
+long long foo_unpack_2 (int low, int high, int ind)
+{
+ int i;
+
+ for (i = low; i <= high; i++)
+ D[i] *= H[i];
+
+ return D[ind];
+}
+
+long long foo_unpack_3 (int low, int high, int ind)
+{
+ int i;
+
+ for (i = low; i <= high; i++)
+ D[i] *= Q[i];
+
+ return D[ind];
+}
+
+int foo_unpack_4 (int low, int high, int ind)
+{
+ int i;
+
+ for (i = low; i <= high; i++)
+ S[i] *= H[i];
+
+ return S[ind];
+}
+
+int foo_unpack_5 (int low, int high, int ind)
+{
+ int i;
+
+ for (i = low; i <= high; i++)
+ S[i] *= Q[i];
+
+ return S[ind];
+}
+
+short foo_unpack_6 (int low, int high, int ind)
+{
+ int i;
+
+ for (i = low; i <= high; i++)
+ H[i] *= Q[i];
+
+ return H[ind];
+}
+
+int foo_expand_1 (int low, int high, int ind)
+{
+ int i;
+
+ for (i = low; i <= high; i++)
+ S[i] *= D[i];
+
+ return S[ind];
+}
+
+short foo_expand_2 (int low, int high, int ind)
+{
+ int i;
+
+ for (i = low; i <= high; i++)
+ H[i] *= D[i];
+
+ return H[ind];
+}
+
+char foo_expand_3 (int low, int high, int ind)
+{
+ int i;
+
+ for (i = low; i <= high; i++)
+ Q[i] *= D[i];
+
+ return Q[ind];
+}
+
+short foo_expand_4 (int low, int high, int ind)
+{
+ int i;
+
+ for (i = low; i <= high; i++)
+ H[i] *= S[i];
+
+ return H[ind];
+}
+
+char foo_expand_5 (int low, int high, int ind)
+{
+ int i;
+
+ for (i = low; i <= high; i++)
+ Q[i] *= S[i];
+
+ return Q[ind];
+}
+
+char foo_expand_6 (int low, int high, int ind)
+{
+ int i;
+
+ for (i = low; i <= high; i++)
+ Q[i] *= H[i];
+
+ return Q[ind];
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vexpandpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vexpandpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+
+#include <immintrin.h>
+
+double *p;
+volatile __m512d x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mask_expand_pd (x, m, x);
+ x = _mm512_maskz_expand_pd (m, x);
+
+ x = _mm512_mask_expandloadu_pd (x, m, p);
+ x = _mm512_maskz_expandloadu_pd (m, p);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (double *s, double *r, MASK_TYPE mask)
+{
+ int i, k;
+
+ for (i = 0, k = 0; i < SIZE; i++)
+ {
+ if (mask & (1 << i))
+ r[i] = s[k++];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s1, res1, res2, res3, res4;
+ MASK_TYPE mask = MASK_VALUE;
+ double s2[SIZE];
+ double res_ref1[SIZE];
+ double res_ref2[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 123.456 * (i + 200) * sign;
+ s2[i] = 789.012 * (i + 300) * sign;
+ res1.a[i] = DEFAULT_VALUE;
+ res3.a[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+ res1.x = INTRINSIC (_mask_expand_pd) (res1.x, mask, s1.x);
+ res2.x = INTRINSIC (_maskz_expand_pd) (mask, s1.x);
+ res3.x = INTRINSIC (_mask_expandloadu_pd) (res3.x, mask, s2);
+ res4.x = INTRINSIC (_maskz_expandloadu_pd) (mask, s2);
+
+ CALC (s1.a, res_ref1, mask);
+ CALC (s2, res_ref2, mask);
+
+ MASK_MERGE (d) (res_ref1, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref1))
+ abort ();
+
+ MASK_ZERO (d) (res_ref1, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref1))
+ abort ();
+
+ MASK_MERGE (d) (res_ref2, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref2))
+ abort ();
+
+ MASK_ZERO (d) (res_ref2, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res4, res_ref2))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vexpandps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vexpandps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+
+#include <immintrin.h>
+
+float *p;
+volatile __m512 x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mask_expand_ps (x, m, x);
+ x = _mm512_maskz_expand_ps (m, x);
+
+ x = _mm512_mask_expandloadu_ps (x, m, p);
+ x = _mm512_maskz_expandloadu_ps (m, p);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (float *s, float *r, MASK_TYPE mask)
+{
+ int i, k;
+
+ for (i = 0, k = 0; i < SIZE; i++)
+ {
+ if (mask & (1 << i))
+ r[i] = s[k++];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, ) s1, res1, res2, res3, res4;
+ MASK_TYPE mask = MASK_VALUE;
+ float s2[SIZE];
+ float res_ref1[SIZE];
+ float res_ref2[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 123.456 * (i + 200) * sign;
+ s2[i] = 789.012 * (i + 300) * sign;
+ res1.a[i] = DEFAULT_VALUE;
+ res3.a[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+ res1.x = INTRINSIC (_mask_expand_ps) (res1.x, mask, s1.x);
+ res2.x = INTRINSIC (_maskz_expand_ps) (mask, s1.x);
+ res3.x = INTRINSIC (_mask_expandloadu_ps) (res3.x, mask, s2);
+ res4.x = INTRINSIC (_maskz_expandloadu_ps) (mask, s2);
+
+ CALC (s1.a, res_ref1, mask);
+ CALC (s2, res_ref2, mask);
+
+ MASK_MERGE () (res_ref1, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res1, res_ref1))
+ abort ();
+
+ MASK_ZERO () (res_ref1, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res2, res_ref1))
+ abort ();
+
+ MASK_MERGE () (res_ref2, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res3, res_ref2))
+ abort ();
+
+ MASK_ZERO () (res_ref2, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res4, res_ref2))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vextractf32x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vextractf32x4\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vextractf32x4\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __m128 y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm512_extractf32x4_ps (x, 1);
+ y = _mm512_mask_extractf32x4_ps (y, 2, x, 1);
+ y = _mm512_maskz_extractf32x4_ps (2, x, 1);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vextractf64x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vextractf64x4\[ \\t\]+\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vextractf64x4\[ \\t\]+\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __m256d y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm512_extractf64x4_pd (x, 1);
+ y = _mm512_maskz_extractf64x4_pd (2, x, 1);
+ y = _mm512_mask_extractf64x4_pd (y, 2, x, 1);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-require-effective-target avx512f } */
+/* { dg-options "-O2 -mavx512f" } */
+
+#include <string.h>
+#include "avx512f-check.h"
+#include "avx512f-helper.h"
+
+void static
+avx512f_test (void)
+{
+ union512d s1;
+ union256d res1, res2, res3;
+ __mmask8 mask = 0xBA;
+ double res_ref[4];
+ int j;
+
+ for (j = 0; j < 8; j++)
+ {
+ s1.a[j] = j * j / 4.56;
+ }
+
+ for (j = 0; j < 4; j++)
+ {
+ res1.a[j] = DEFAULT_VALUE;
+ res2.a[j] = DEFAULT_VALUE;
+ res3.a[j] = DEFAULT_VALUE;
+ }
+
+ res1.x = _mm512_extractf64x4_pd (s1.x, 0);
+ res2.x = _mm512_mask_extractf64x4_pd (res2.x, mask, s1.x, 0);
+ res3.x = _mm512_maskz_extractf64x4_pd (mask, s1.x, 0);
+
+ memset (res_ref, 0, 32);
+ memcpy (res_ref, s1.a, 32);
+
+ if (check_union256d (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, 4);
+ if (check_union256d (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, 4);
+ if (check_union256d (res3, res_ref))
+ abort ();
+
+ res1.x = _mm512_extractf64x4_pd (s1.x, 1);
+ res2.x = _mm512_mask_extractf64x4_pd (res2.x, mask, s1.x, 1);
+ res3.x = _mm512_maskz_extractf64x4_pd (mask, s1.x, 1);
+
+ memset (res_ref, 0, 32);
+ memcpy (res_ref, s1.a + 4, 32);
+
+ if (check_union256d (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, 4);
+ if (check_union256d (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, 4);
+ if (check_union256d (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vextracti32x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vextracti32x4\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vextracti32x4\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __m128i y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm512_extracti32x4_epi32 (x, 1);
+ y = _mm512_mask_extracti32x4_epi32 (y, 2, x, 1);
+ y = _mm512_maskz_extracti32x4_epi32 (2, x, 1);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vextracti64x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vextracti64x4\[ \\t\]+\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vextracti64x4\[ \\t\]+\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __m256i y;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm512_extracti64x4_epi64 (x, 1);
+ y = _mm512_mask_extracti64x4_epi64 (y, 2, x, 1);
+ y = _mm512_maskz_extracti64x4_epi64 (2, x, 1);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-require-effective-target avx512f } */
+/* { dg-options "-O2 -mavx512f" } */
+
+#include <string.h>
+#include "avx512f-check.h"
+#include "avx512f-helper.h"
+
+void static
+avx512f_test (void)
+{
+ union512i_q s1;
+ union256i_q res1, res2, res3;
+ __mmask8 mask = 0xBA;
+ long long int res_ref[4];
+ int j;
+
+ for (j = 0; j < 8; j++)
+ s1.a[j] = j * j;
+
+ for (j = 0; j < 4; j++)
+ {
+ res1.a[j] = DEFAULT_VALUE;
+ res2.a[j] = DEFAULT_VALUE;
+ res3.a[j] = DEFAULT_VALUE;
+ }
+ res1.x = _mm512_extracti64x4_epi64 (s1.x, 0);
+ res2.x = _mm512_mask_extracti64x4_epi64 (res2.x, mask, s1.x, 0);
+ res3.x = _mm512_maskz_extracti64x4_epi64 (mask, s1.x, 0);
+
+ memset (res_ref, 0, 32);
+ memcpy (res_ref, s1.a, 32);
+
+ if (check_union256i_q (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, 4);
+ if (check_union256i_q (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, 4);
+ if (check_union256i_q (res3, res_ref))
+ abort ();
+
+ res1.x = _mm512_extracti64x4_epi64 (s1.x, 1);
+ res2.x = _mm512_mask_extracti64x4_epi64 (res2.x, mask, s1.x, 1);
+ res3.x = _mm512_maskz_extracti64x4_epi64 (mask, s1.x, 1);
+
+ memset (res_ref, 0, 32);
+ memcpy (res_ref, s1.a + 4, 32);
+
+ if (check_union256i_q (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, 4);
+ if (check_union256i_q (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, 4);
+ if (check_union256i_q (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vfixupimmpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vfixupimmpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vfixupimmpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vfixupimmpd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vfixupimmpd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfixupimmpd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x1, x2;
+volatile __m512i y;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x1 = _mm512_fixupimm_pd (x1, x2, y, 3);
+ x1 = _mm512_mask_fixupimm_pd (x1, m, x2, y, 3);
+ x1 = _mm512_maskz_fixupimm_pd (m, x1, x2, y, 3);
+ x1 = _mm512_fixupimm_round_pd (x1, x2, y, 3, _MM_FROUND_NO_EXC);
+ x1 = _mm512_mask_fixupimm_round_pd (x1, m, x2, y, 3, _MM_FROUND_NO_EXC);
+ x1 = _mm512_maskz_fixupimm_round_pd (m, x1, x2, y, 3, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+#include "math.h"
+#include "values.h"
+
+static void
+CALC (double *r, double src, long long tbl)
+{
+ switch (tbl & 0xf)
+ {
+ case 0:
+ *r = src;
+ break;
+ case 1:
+ *r = src;
+ break;
+ case 2:
+ *r = signbit (src) ? -NAN : NAN;
+ break;
+ case 3:
+ *r = -NAN;
+ break;
+ case 4:
+ *r = -INFINITY;
+ break;
+ case 5:
+ *r = INFINITY;
+ break;
+ case 6:
+ *r = signbit (src) ? -INFINITY : INFINITY;
+ break;
+ case 7:
+ *r = 1.0 / -INFINITY;
+ break;
+ case 8:
+ *r = 0.0;
+ break;
+ case 9:
+ *r = -1.0;
+ break;
+ case 10:
+ *r = 1.0;
+ break;
+ case 11:
+ *r = 1.0 / 2.0;
+ break;
+ case 12:
+ *r = 90.0;
+ break;
+ case 13:
+ *r = M_PI_2;
+ break;
+ case 14:
+ *r = MAXDOUBLE;
+ break;
+ case 15:
+ *r = -MAXDOUBLE;
+ break;
+ default:
+ abort ();
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, j, k;
+ UNION_TYPE (AVX512F_LEN, d) res1, res2, res3, s1;
+ UNION_TYPE (AVX512F_LEN, i_q) s2;
+ double res_ref[SIZE];
+
+
+ float vals[2] = { -10, 10 };
+ int controls[8] = {0x11111111, 0x77777777, 0x77777777, 0x88888888,
+ 0x99999999, 0xaaaaaaaa, 0xbbbbbbbb, 0xcccccccc};
+
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < 2; i++)
+ {
+ for (j = 0; j < SIZE; j++)
+ {
+ s1.a[j] = vals[i];
+ s2.a[j] = controls[j];
+ res1.a[j] = DEFAULT_VALUE;
+ res2.a[j] = DEFAULT_VALUE;
+ res3.a[j] = DEFAULT_VALUE;
+
+ CALC (&res_ref[j], s1.a[j], s2.a[j]);
+ }
+
+ res1.x = INTRINSIC (_fixupimm_pd) (res1.x, s1.x, s2.x, 0);
+ res2.x = INTRINSIC (_mask_fixupimm_pd) (res2.x, mask, s1.x, s2.x, 0);
+ res3.x = INTRINSIC (_maskz_fixupimm_pd) (mask, res3.x, s1.x, s2.x, 0);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE(d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+ MASK_ZERO(d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+ }
+}
+
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vfixupimmps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vfixupimmps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vfixupimmps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vfixupimmps\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vfixupimmps\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfixupimmps\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x1, x2;
+volatile __m512i y;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x1 = _mm512_fixupimm_ps (x1, x2, y, 3);
+ x1 = _mm512_mask_fixupimm_ps (x1, m, x2, y, 3);
+ x1 = _mm512_maskz_fixupimm_ps (m, x1, x2, y, 3);
+ x1 = _mm512_fixupimm_round_ps (x1, x2, y, 3, _MM_FROUND_NO_EXC);
+ x1 = _mm512_mask_fixupimm_round_ps (x1, m, x2, y, 3, _MM_FROUND_NO_EXC);
+ x1 = _mm512_maskz_fixupimm_round_ps (m, x1, x2, y, 3, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#include "math.h"
+#include "values.h"
+
+static void
+CALC (float *r, float src, int tbl)
+{
+ switch (tbl & 0xf)
+ {
+ case 0:
+ *r = src;
+ break;
+ case 1:
+ *r = src;
+ break;
+ case 2:
+ *r = signbit (src) ? -NAN : NAN;
+ break;
+ case 3:
+ *r = -NAN;
+ break;
+ case 4:
+ *r = -INFINITY;
+ break;
+ case 5:
+ *r = INFINITY;
+ break;
+ case 6:
+ *r = signbit (src) ? -INFINITY : INFINITY;
+ break;
+ case 7:
+ *r = 1.0 / -INFINITY;
+ break;
+ case 8:
+ *r = 0.0;
+ break;
+ case 9:
+ *r = -1.0;
+ break;
+ case 10:
+ *r = 1.0;
+ break;
+ case 11:
+ *r = 1.0 / 2.0;
+ break;
+ case 12:
+ *r = 90.0;
+ break;
+ case 13:
+ *r = M_PI_2;
+ break;
+ case 14:
+ *r = MAXFLOAT;
+ break;
+ case 15:
+ *r = -MAXFLOAT;
+ break;
+ default:
+ abort ();
+ }
+}
+
+
+void static
+TEST (void)
+{
+ int i, j, k;
+ UNION_TYPE (AVX512F_LEN,) res1, res2, res3, s1;
+ UNION_TYPE (AVX512F_LEN, i_d) s2;
+ float res_ref[SIZE];
+
+
+ float vals[2] = { -10, 10 };
+ int controls[16] = { 0x11111111,
+ 0x77777777, 0x88888888, 0x99999999,
+ 0xaaaaaaaa, 0xbbbbbbbb, 0xcccccccc,
+ 0x77777777, 0x88888888, 0x99999999,
+ 0xaaaaaaaa, 0xbbbbbbbb, 0xcccccccc,
+ 0xdddddddd, 0xeeeeeeee, 0xffffffff
+ };
+
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < 2; i++)
+ {
+ for (j = 0; j < SIZE; j++)
+ {
+ s1.a[j] = vals[i];
+ s2.a[j] = controls[j];
+ res1.a[j] = DEFAULT_VALUE;
+ res2.a[j] = DEFAULT_VALUE;
+ res3.a[j] = DEFAULT_VALUE;
+
+ CALC (&res_ref[j], s1.a[j], s2.a[j]);
+ }
+
+ res1.x = INTRINSIC (_fixupimm_ps) (res1.x, s1.x, s2.x, 0);
+ res2.x = INTRINSIC (_mask_fixupimm_ps) (res2.x, mask, s1.x, s2.x, 0);
+ res3.x = INTRINSIC (_maskz_fixupimm_ps) (mask, res3.x, s1.x, s2.x, 0);
+
+ if (UNION_CHECK (AVX512F_LEN,) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE() (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res2, res_ref))
+ abort ();
+ MASK_ZERO() (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res3, res_ref))
+ abort ();
+ }
+}
+
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vfixupimmsd\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vfixupimmsd\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vfixupimmsd\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vfixupimmsd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vfixupimmsd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfixupimmsd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128d x;
+volatile __m128i y;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm_fixupimm_sd (x, x, y, 3);
+ x = _mm_mask_fixupimm_sd (x, m, x, y, 3);
+ x = _mm_maskz_fixupimm_sd (m, x, x, y, 3);
+ x = _mm_fixupimm_round_sd (x, x, y, 3, _MM_FROUND_NO_EXC);
+ x = _mm_mask_fixupimm_round_sd (x, m, x, y, 3, _MM_FROUND_NO_EXC);
+ x = _mm_maskz_fixupimm_round_sd (m, x, x, y, 3, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+#include "avx512f-helper.h"
+#include <math.h>
+#include <values.h>
+#include "avx512f-mask-type.h"
+
+void
+compute_fixupimmpd (double *r, double src, long long tbl)
+{
+ switch (tbl & 0xf)
+ {
+ case 0:
+ *r = src;
+ break;
+ case 1:
+ *r = src;
+ break;
+ case 2:
+ *r = signbit (src) ? -NAN : NAN;
+ break;
+ case 3:
+ *r = -NAN;
+ break;
+ case 4:
+ *r = -INFINITY;
+ break;
+ case 5:
+ *r = INFINITY;
+ break;
+ case 6:
+ *r = signbit (src) ? -INFINITY : INFINITY;
+ break;
+ case 7:
+ *r = 1.0 / -INFINITY;
+ break;
+ case 8:
+ *r = 0.0;
+ break;
+ case 9:
+ *r = -1.0;
+ break;
+ case 10:
+ *r = 1.0;
+ break;
+ case 11:
+ *r = 1.0 / 2.0;
+ break;
+ case 12:
+ *r = 90.0;
+ break;
+ case 13:
+ *r = M_PI_2;
+ break;
+ case 14:
+ *r = MAXDOUBLE;
+ break;
+ case 15:
+ *r = -MAXDOUBLE;
+ break;
+ default:
+ abort ();
+ }
+}
+
+void static
+avx512f_test (void)
+{
+ union128d s1, res1, res2, res3;
+ union128i_q s2;
+ double res_ref[2];
+ int i, j, k;
+
+ float vals[2] = { -10, 10 };
+ int controls[10] = { 0x11111111,
+ 0x77777777, 0x88888888, 0x99999999,
+ 0xaaaaaaaa, 0xbbbbbbbb, 0xcccccccc,
+ 0xdddddddd, 0xeeeeeeee, 0xffffffff
+ };
+
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < 2; i++)
+ {
+ s1.a[0] = vals[i];
+ s1.a[1] = 1.0;
+ s2.a[1] = 1.0;
+
+ res_ref[0] = 1.0;
+ res_ref[1] = 1.0;
+ res1.a[0] = res2.a[0] = res3.a[0] = DEFAULT_VALUE;
+ res1.a[1] = res2.a[1] = res3.a[1] = DEFAULT_VALUE;
+
+ for (j = 0; j < 10; j++)
+ {
+ s2.a[0] = controls[j];
+ compute_fixupimmpd (&res_ref[0], s1.a[0], s2.a[0]);
+
+ res1.x = _mm_fixupimm_sd (res1.x, s1.x, s2.x, 0);
+ res2.x = _mm_mask_fixupimm_sd (res2.x, mask, s1.x, s2.x, 0);
+ res3.x = _mm_maskz_fixupimm_sd (mask, res3.x, s1.x, s2.x, 0);
+
+ if (check_union128d (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, 1);
+ if (check_union128d (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, 1);
+ if (check_union128d (res3, res_ref))
+ abort ();
+ }
+ }
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vfixupimmss\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vfixupimmss\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vfixupimmss\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vfixupimmss\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vfixupimmss\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfixupimmss\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128 x;
+volatile __m128i y;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm_fixupimm_ss (x, x, y, 3);
+ x = _mm_mask_fixupimm_ss (x, m, x, y, 3);
+ x = _mm_maskz_fixupimm_ss (m, x, x, y, 3);
+ x = _mm_fixupimm_round_ss (x, x, y, 3, _MM_FROUND_NO_EXC);
+ x = _mm_mask_fixupimm_round_ss (x, m, x, y, 3, _MM_FROUND_NO_EXC);
+ x = _mm_maskz_fixupimm_round_ss (m, x, x, y, 3, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+#include "avx512f-helper.h"
+#include <math.h>
+#include <values.h>
+#include "avx512f-mask-type.h"
+
+void
+compute_fixupimmps (float *r, float src, int tbl)
+{
+ switch (tbl & 0xf)
+ {
+ case 0:
+ *r = src;
+ break;
+ case 1:
+ *r = src;
+ break;
+ case 2:
+ *r = signbit (src) ? -NAN : NAN;
+ break;
+ case 3:
+ *r = -NAN;
+ break;
+ case 4:
+ *r = -INFINITY;
+ break;
+ case 5:
+ *r = INFINITY;
+ break;
+ case 6:
+ *r = signbit (src) ? -INFINITY : INFINITY;
+ break;
+ case 7:
+ *r = 1.0 / -INFINITY;
+ break;
+ case 8:
+ *r = 0.0;
+ break;
+ case 9:
+ *r = -1.0;
+ break;
+ case 10:
+ *r = 1.0;
+ break;
+ case 11:
+ *r = 1.0 / 2.0;
+ break;
+ case 12:
+ *r = 90.0;
+ break;
+ case 13:
+ *r = M_PI_2;
+ break;
+ case 14:
+ *r = MAXFLOAT;
+ break;
+ case 15:
+ *r = -MAXFLOAT;
+ break;
+ default:
+ abort ();
+ }
+}
+
+void static
+avx512f_test (void)
+{
+ union128 s1, res1, res2, res3;
+ union128i_d s2;
+ float res_ref[4];
+ int i, j, k;
+
+ float vals[2] = { -10, 10 };
+ int controls[10] = { 0x11111111,
+ 0x77777777, 0x88888888, 0x99999999,
+ 0xaaaaaaaa, 0xbbbbbbbb, 0xcccccccc,
+ 0xdddddddd, 0xeeeeeeee, 0xffffffff
+ };
+
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < 2; i++)
+ {
+ s1.a[0] = vals[i];
+ res1.a[0] = res2.a[0] = res3.a[0] = DEFAULT_VALUE;
+ for (k = 1; k < 4; k++)
+ {
+ s1.a[k] = k;
+ s2.a[k] = k;
+ res_ref[k] = k;
+ res1.a[k] = res2.a[k] = res3.a[k] = DEFAULT_VALUE;
+ }
+
+ for (j = 0; j < 10; j++)
+ {
+ s2.a[0] = controls[j];
+ compute_fixupimmps (&res_ref[0], s1.a[0], s2.a[0]);
+
+ res1.x = _mm_fixupimm_ss (res1.x, s1.x, s2.x, 0);
+ res2.x = _mm_mask_fixupimm_ss (res2.x, mask, s1.x, s2.x, 0);
+ res3.x = _mm_maskz_fixupimm_ss (mask, res3.x, s1.x, s2.x, 0);
+
+ if (check_union128 (res1, res_ref))
+ abort ();
+
+ MASK_MERGE () (res_ref, mask, 1);
+ if (check_union128 (res2, res_ref))
+ abort ();
+
+ MASK_ZERO () (res_ref, mask, 1);
+ if (check_union128 (res3, res_ref))
+ abort ();
+ }
+ }
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vfmadd...pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 8 } } */
+/* { dg-final { scan-assembler-times "vfmadd...pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 4 } } */
+/* { dg-final { scan-assembler-times "vfmadd231pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vfmadd...pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vfmadd...pd\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmadd...pd\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmadd231pd\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmadd...pd\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x1, x2, x3;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x1 = _mm512_fmadd_pd (x1, x2, x3);
+ x1 = _mm512_mask_fmadd_pd (x1, m, x2, x3);
+ x3 = _mm512_mask3_fmadd_pd (x1, x2, x3, m);
+ x1 = _mm512_maskz_fmadd_pd (m, x1, x2, x3);
+ x1 = _mm512_fmadd_round_pd (x1, x2, x3, _MM_FROUND_TO_NEAREST_INT);
+ x1 = _mm512_mask_fmadd_round_pd (x1, m, x2, x3, _MM_FROUND_TO_NEG_INF);
+ x3 = _mm512_mask3_fmadd_round_pd (x1, x2, x3, m, _MM_FROUND_TO_POS_INF);
+ x1 = _mm512_maskz_fmadd_round_pd (m, x1, x2, x3, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (double *s1, double *s2, double *s3, double *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[i] * s2[i] + s3[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s1, s2, s3, res1, res2, res3, res4;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref1[SIZE];
+ double res_ref2[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = DEFAULT_VALUE;
+ s2.a[i] = 56.78 * (i + 1) * sign;
+ s3.a[i] = 90.12 * (i + 2) * sign;
+ sign = -sign;
+ }
+
+#if AVX512F_LEN == 512
+ res1.x = INTRINSIC (_fmadd_pd) (s1.x, s2.x, s3.x);
+#endif
+ res2.x = INTRINSIC (_mask_fmadd_pd) (s1.x, mask, s2.x, s3.x);
+ res3.x = INTRINSIC (_mask3_fmadd_pd) (s2.x, s3.x, s1.x, mask);
+ res4.x = INTRINSIC (_maskz_fmadd_pd) (mask, s1.x, s2.x, s3.x);
+
+ CALC (s1.a, s2.a, s3.a, res_ref1);
+ CALC (s2.a, s3.a, s1.a, res_ref2);
+
+#if AVX512F_LEN == 512
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res1, res_ref1, 0.0001))
+ abort ();
+#endif
+
+ MASK_MERGE (d) (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res2, res_ref1, 0.0001))
+ abort ();
+
+ MASK_MERGE (d) (res_ref2, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res3, res_ref2, 0.0001))
+ abort ();
+
+ MASK_ZERO (d) (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res4, res_ref1, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vfmadd...ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 8 } } */
+/* { dg-final { scan-assembler-times "vfmadd...ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 4 } } */
+/* { dg-final { scan-assembler-times "vfmadd231ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vfmadd...ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vfmadd...ps\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmadd...ps\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmadd231ps\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmadd...ps\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x1, x2, x3;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x1 = _mm512_fmadd_ps (x1, x2, x3);
+ x1 = _mm512_mask_fmadd_ps (x1, m, x2, x3);
+ x3 = _mm512_mask3_fmadd_ps (x1, x2, x3, m);
+ x1 = _mm512_maskz_fmadd_ps (m, x1, x2, x3);
+ x1 = _mm512_fmadd_round_ps (x1, x2, x3, _MM_FROUND_TO_NEAREST_INT);
+ x1 = _mm512_mask_fmadd_round_ps (x1, m, x2, x3, _MM_FROUND_TO_NEG_INF);
+ x3 = _mm512_mask3_fmadd_round_ps (x1, x2, x3, m, _MM_FROUND_TO_POS_INF);
+ x1 = _mm512_maskz_fmadd_round_ps (m, x1, x2, x3, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (float *s1, float *s2, float *s3, float *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[i] * s2[i] + s3[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, ) s1, s2, s3, res1, res2, res3, res4;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref1[SIZE];
+ float res_ref2[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = DEFAULT_VALUE;
+ s2.a[i] = 56.78 * (i + 1) * sign;
+ s3.a[i] = 90.12 * (i + 2) * sign;
+ sign = -sign;
+ }
+
+#if AVX512F_LEN == 512
+ res1.x = INTRINSIC (_fmadd_ps) (s1.x, s2.x, s3.x);
+#endif
+ res2.x = INTRINSIC (_mask_fmadd_ps) (s1.x, mask, s2.x, s3.x);
+ res3.x = INTRINSIC (_mask3_fmadd_ps) (s2.x, s3.x, s1.x, mask);
+ res4.x = INTRINSIC (_maskz_fmadd_ps) (mask, s1.x, s2.x, s3.x);
+
+ CALC (s1.a, s2.a, s3.a, res_ref1);
+ CALC (s2.a, s3.a, s1.a, res_ref2);
+
+#if AVX512F_LEN == 512
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res1, res_ref1, 0.0001))
+ abort ();
+#endif
+
+ MASK_MERGE () (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res2, res_ref1, 0.0001))
+ abort ();
+
+ MASK_MERGE () (res_ref2, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res3, res_ref2, 0.0001))
+ abort ();
+
+ MASK_ZERO () (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res4, res_ref1, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vfmaddsub...pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 8 } } */
+/* { dg-final { scan-assembler-times "vfmaddsub...pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 4 } } */
+/* { dg-final { scan-assembler-times "vfmaddsub231pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vfmaddsub...pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vfmaddsub...pd\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmaddsub...pd\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmaddsub231pd\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmaddsub...pd\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x1, x2, x3;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x1 = _mm512_fmaddsub_pd (x1, x2, x3);
+ x1 = _mm512_mask_fmaddsub_pd (x1, m, x2, x3);
+ x3 = _mm512_mask3_fmaddsub_pd (x1, x2, x3, m);
+ x1 = _mm512_maskz_fmaddsub_pd (m, x1, x2, x3);
+ x1 = _mm512_fmaddsub_round_pd (x1, x2, x3, _MM_FROUND_TO_NEAREST_INT);
+ x1 = _mm512_mask_fmaddsub_round_pd (x1, m, x2, x3, _MM_FROUND_TO_NEG_INF);
+ x3 = _mm512_mask3_fmaddsub_round_pd (x1, x2, x3, m, _MM_FROUND_TO_POS_INF);
+ x1 = _mm512_maskz_fmaddsub_round_pd (m, x1, x2, x3, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (double *s1, double *s2, double *s3, double *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ if (i % 2)
+ r[i] = s1[i] * s2[i] + s3[i];
+ else
+ r[i] = s1[i] * s2[i] - s3[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s1, s2, s3, res1, res2, res3, res4;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref1[SIZE];
+ double res_ref2[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = DEFAULT_VALUE;
+ s2.a[i] = 56.78 * (i + 1) * sign;
+ s3.a[i] = 90.12 * (i + 2) * sign;
+ sign = -sign;
+ }
+
+#if AVX512F_LEN == 512
+ res1.x = INTRINSIC (_fmaddsub_pd) (s1.x, s2.x, s3.x);
+#endif
+ res2.x = INTRINSIC (_mask_fmaddsub_pd) (s1.x, mask, s2.x, s3.x);
+ res3.x = INTRINSIC (_mask3_fmaddsub_pd) (s2.x, s3.x, s1.x, mask);
+ res4.x = INTRINSIC (_maskz_fmaddsub_pd) (mask, s1.x, s2.x, s3.x);
+
+ CALC (s1.a, s2.a, s3.a, res_ref1);
+ CALC (s2.a, s3.a, s1.a, res_ref2);
+
+#if AVX512F_LEN == 512
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res1, res_ref1, 0.0001))
+ abort ();
+#endif
+
+ MASK_MERGE (d) (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res2, res_ref1, 0.0001))
+ abort ();
+
+ MASK_MERGE (d) (res_ref2, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res3, res_ref2, 0.0001))
+ abort ();
+
+ MASK_ZERO (d) (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res4, res_ref1, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vfmaddsub...ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 8 } } */
+/* { dg-final { scan-assembler-times "vfmaddsub...ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 4 } } */
+/* { dg-final { scan-assembler-times "vfmaddsub231ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vfmaddsub...ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vfmaddsub...ps\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmaddsub...ps\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmaddsub231ps\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmaddsub...ps\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x1, x2, x3;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x1 = _mm512_fmaddsub_ps (x1, x2, x3);
+ x1 = _mm512_mask_fmaddsub_ps (x1, m, x2, x3);
+ x3 = _mm512_mask3_fmaddsub_ps (x1, x2, x3, m);
+ x1 = _mm512_maskz_fmaddsub_ps (m, x1, x2, x3);
+ x1 = _mm512_fmaddsub_round_ps (x1, x2, x3, _MM_FROUND_TO_NEAREST_INT);
+ x1 = _mm512_mask_fmaddsub_round_ps (x1, m, x2, x3, _MM_FROUND_TO_NEG_INF);
+ x3 = _mm512_mask3_fmaddsub_round_ps (x1, x2, x3, m, _MM_FROUND_TO_POS_INF);
+ x1 = _mm512_maskz_fmaddsub_round_ps (m, x1, x2, x3, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (float *s1, float *s2, float *s3, float *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ if (i % 2)
+ r[i] = s1[i] * s2[i] + s3[i];
+ else
+ r[i] = s1[i] * s2[i] - s3[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, ) s1, s2, s3, res1, res2, res3, res4;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref1[SIZE];
+ float res_ref2[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = DEFAULT_VALUE;
+ s2.a[i] = 56.78 * (i + 1) * sign;
+ s3.a[i] = 90.12 * (i + 2) * sign;
+ sign = -sign;
+ }
+
+#if AVX512F_LEN == 512
+ res1.x = INTRINSIC (_fmaddsub_ps) (s1.x, s2.x, s3.x);
+#endif
+ res2.x = INTRINSIC (_mask_fmaddsub_ps) (s1.x, mask, s2.x, s3.x);
+ res3.x = INTRINSIC (_mask3_fmaddsub_ps) (s2.x, s3.x, s1.x, mask);
+ res4.x = INTRINSIC (_maskz_fmaddsub_ps) (mask, s1.x, s2.x, s3.x);
+
+ CALC (s1.a, s2.a, s3.a, res_ref1);
+ CALC (s2.a, s3.a, s1.a, res_ref2);
+
+#if AVX512F_LEN == 512
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res1, res_ref1, 0.0001))
+ abort ();
+#endif
+
+ MASK_MERGE () (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res2, res_ref1, 0.0001))
+ abort ();
+
+ MASK_MERGE () (res_ref2, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res3, res_ref2, 0.0001))
+ abort ();
+
+ MASK_ZERO () (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res4, res_ref1, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vfmsub...pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 8 } } */
+/* { dg-final { scan-assembler-times "vfmsub...pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 4 } } */
+/* { dg-final { scan-assembler-times "vfmsub231pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vfmsub...pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vfmsub...pd\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmsub...pd\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmsub231pd\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmsub...pd\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x1, x2, x3;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x1 = _mm512_fmsub_pd (x1, x2, x3);
+ x1 = _mm512_mask_fmsub_pd (x1, m, x2, x3);
+ x3 = _mm512_mask3_fmsub_pd (x1, x2, x3, m);
+ x1 = _mm512_maskz_fmsub_pd (m, x1, x2, x3);
+ x1 = _mm512_fmsub_round_pd (x1, x2, x3, _MM_FROUND_TO_NEAREST_INT);
+ x1 = _mm512_mask_fmsub_round_pd (x1, m, x2, x3, _MM_FROUND_TO_NEG_INF);
+ x3 = _mm512_mask3_fmsub_round_pd (x1, x2, x3, m, _MM_FROUND_TO_POS_INF);
+ x1 = _mm512_maskz_fmsub_round_pd (m, x1, x2, x3, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (double *s1, double *s2, double *s3, double *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[i] * s2[i] - s3[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s1, s2, s3, res1, res2, res3, res4;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref1[SIZE];
+ double res_ref2[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = DEFAULT_VALUE;
+ s2.a[i] = 56.78 * (i + 1) * sign;
+ s3.a[i] = 90.12 * (i + 2) * sign;
+ sign = -sign;
+ }
+
+#if AVX512F_LEN == 512
+ res1.x = INTRINSIC (_fmsub_pd) (s1.x, s2.x, s3.x);
+#endif
+ res2.x = INTRINSIC (_mask_fmsub_pd) (s1.x, mask, s2.x, s3.x);
+ res3.x = INTRINSIC (_mask3_fmsub_pd) (s2.x, s3.x, s1.x, mask);
+ res4.x = INTRINSIC (_maskz_fmsub_pd) (mask, s1.x, s2.x, s3.x);
+
+ CALC (s1.a, s2.a, s3.a, res_ref1);
+ CALC (s2.a, s3.a, s1.a, res_ref2);
+
+#if AVX512F_LEN == 512
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res1, res_ref1, 0.0001))
+ abort ();
+#endif
+
+ MASK_MERGE (d) (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res2, res_ref1, 0.0001))
+ abort ();
+
+ MASK_MERGE (d) (res_ref2, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res3, res_ref2, 0.0001))
+ abort ();
+
+ MASK_ZERO (d) (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res4, res_ref1, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vfmsub...ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 8 } } */
+/* { dg-final { scan-assembler-times "vfmsub...ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 4 } } */
+/* { dg-final { scan-assembler-times "vfmsub231ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vfmsub...ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vfmsub...ps\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmsub...ps\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmsub231ps\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmsub...ps\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x1, x2, x3;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x1 = _mm512_fmsub_ps (x1, x2, x3);
+ x1 = _mm512_mask_fmsub_ps (x1, m, x2, x3);
+ x3 = _mm512_mask3_fmsub_ps (x1, x2, x3, m);
+ x1 = _mm512_maskz_fmsub_ps (m, x1, x2, x3);
+ x1 = _mm512_fmsub_round_ps (x1, x2, x3, _MM_FROUND_TO_NEAREST_INT);
+ x1 = _mm512_mask_fmsub_round_ps (x1, m, x2, x3, _MM_FROUND_TO_NEG_INF);
+ x3 = _mm512_mask3_fmsub_round_ps (x1, x2, x3, m, _MM_FROUND_TO_POS_INF);
+ x1 = _mm512_maskz_fmsub_round_ps (m, x1, x2, x3, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (float *s1, float *s2, float *s3, float *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[i] * s2[i] - s3[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, ) s1, s2, s3, res1, res2, res3, res4;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref1[SIZE];
+ float res_ref2[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = DEFAULT_VALUE;
+ s2.a[i] = 56.78 * (i + 1) * sign;
+ s3.a[i] = 90.12 * (i + 2) * sign;
+ sign = -sign;
+ }
+
+#if AVX512F_LEN == 512
+ res1.x = INTRINSIC (_fmsub_ps) (s1.x, s2.x, s3.x);
+#endif
+ res2.x = INTRINSIC (_mask_fmsub_ps) (s1.x, mask, s2.x, s3.x);
+ res3.x = INTRINSIC (_mask3_fmsub_ps) (s2.x, s3.x, s1.x, mask);
+ res4.x = INTRINSIC (_maskz_fmsub_ps) (mask, s1.x, s2.x, s3.x);
+
+ CALC (s1.a, s2.a, s3.a, res_ref1);
+ CALC (s2.a, s3.a, s1.a, res_ref2);
+
+#if AVX512F_LEN == 512
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res1, res_ref1, 0.0001))
+ abort ();
+#endif
+
+ MASK_MERGE () (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res2, res_ref1, 0.0001))
+ abort ();
+
+ MASK_MERGE () (res_ref2, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res3, res_ref2, 0.0001))
+ abort ();
+
+ MASK_ZERO () (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res4, res_ref1, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vfmsubadd...pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 8 } } */
+/* { dg-final { scan-assembler-times "vfmsubadd...pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 4 } } */
+/* { dg-final { scan-assembler-times "vfmsubadd231pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vfmsubadd...pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vfmsubadd...pd\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmsubadd...pd\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmsubadd231pd\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmsubadd...pd\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x1, x2, x3;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x1 = _mm512_fmsubadd_pd (x1, x2, x3);
+ x1 = _mm512_mask_fmsubadd_pd (x1, m, x2, x3);
+ x3 = _mm512_mask3_fmsubadd_pd (x1, x2, x3, m);
+ x1 = _mm512_maskz_fmsubadd_pd (m, x1, x2, x3);
+ x1 = _mm512_fmsubadd_round_pd (x1, x2, x3, _MM_FROUND_TO_NEAREST_INT);
+ x1 = _mm512_mask_fmsubadd_round_pd (x1, m, x2, x3, _MM_FROUND_TO_NEG_INF);
+ x3 = _mm512_mask3_fmsubadd_round_pd (x1, x2, x3, m, _MM_FROUND_TO_POS_INF);
+ x1 = _mm512_maskz_fmsubadd_round_pd (m, x1, x2, x3, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (double *s1, double *s2, double *s3, double *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ if (i % 2)
+ r[i] = s1[i] * s2[i] - s3[i];
+ else
+ r[i] = s1[i] * s2[i] + s3[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s1, s2, s3, res1, res2, res3, res4;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref1[SIZE];
+ double res_ref2[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = DEFAULT_VALUE;
+ s2.a[i] = 56.78 * (i + 1) * sign;
+ s3.a[i] = 90.12 * (i + 2) * sign;
+ sign = -sign;
+ }
+
+#if AVX512F_LEN == 512
+ res1.x = INTRINSIC (_fmsubadd_pd) (s1.x, s2.x, s3.x);
+#endif
+ res2.x = INTRINSIC (_mask_fmsubadd_pd) (s1.x, mask, s2.x, s3.x);
+ res3.x = INTRINSIC (_mask3_fmsubadd_pd) (s2.x, s3.x, s1.x, mask);
+ res4.x = INTRINSIC (_maskz_fmsubadd_pd) (mask, s1.x, s2.x, s3.x);
+
+ CALC (s1.a, s2.a, s3.a, res_ref1);
+ CALC (s2.a, s3.a, s1.a, res_ref2);
+
+#if AVX512F_LEN == 512
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res1, res_ref1, 0.0001))
+ abort ();
+#endif
+
+ MASK_MERGE (d) (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res2, res_ref1, 0.0001))
+ abort ();
+
+ MASK_MERGE (d) (res_ref2, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res3, res_ref2, 0.0001))
+ abort ();
+
+ MASK_ZERO (d) (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res4, res_ref1, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vfmsubadd...ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 8 } } */
+/* { dg-final { scan-assembler-times "vfmsubadd...ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 4 } } */
+/* { dg-final { scan-assembler-times "vfmsubadd231ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vfmsubadd...ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vfmsubadd...ps\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmsubadd...ps\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmsubadd231ps\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfmsubadd...ps\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x1, x2, x3;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x1 = _mm512_fmsubadd_ps (x1, x2, x3);
+ x1 = _mm512_mask_fmsubadd_ps (x1, m, x2, x3);
+ x3 = _mm512_mask3_fmsubadd_ps (x1, x2, x3, m);
+ x1 = _mm512_maskz_fmsubadd_ps (m, x1, x2, x3);
+ x1 = _mm512_fmsubadd_round_ps (x1, x2, x3, _MM_FROUND_TO_NEAREST_INT);
+ x1 = _mm512_mask_fmsubadd_round_ps (x1, m, x2, x3, _MM_FROUND_TO_NEG_INF);
+ x3 = _mm512_mask3_fmsubadd_round_ps (x1, x2, x3, m, _MM_FROUND_TO_POS_INF);
+ x1 = _mm512_maskz_fmsubadd_round_ps (m, x1, x2, x3, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (float *s1, float *s2, float *s3, float *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ if (i % 2)
+ r[i] = s1[i] * s2[i] - s3[i];
+ else
+ r[i] = s1[i] * s2[i] + s3[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, ) s1, s2, s3, res1, res2, res3, res4;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref1[SIZE];
+ float res_ref2[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = DEFAULT_VALUE;
+ s2.a[i] = 56.78 * (i + 1) * sign;
+ s3.a[i] = 90.12 * (i + 2) * sign;
+ sign = -sign;
+ }
+
+#if AVX512F_LEN == 512
+ res1.x = INTRINSIC (_fmsubadd_ps) (s1.x, s2.x, s3.x);
+#endif
+ res2.x = INTRINSIC (_mask_fmsubadd_ps) (s1.x, mask, s2.x, s3.x);
+ res3.x = INTRINSIC (_mask3_fmsubadd_ps) (s2.x, s3.x, s1.x, mask);
+ res4.x = INTRINSIC (_maskz_fmsubadd_ps) (mask, s1.x, s2.x, s3.x);
+
+ CALC (s1.a, s2.a, s3.a, res_ref1);
+ CALC (s2.a, s3.a, s1.a, res_ref2);
+
+#if AVX512F_LEN == 512
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res1, res_ref1, 0.0001))
+ abort ();
+#endif
+
+ MASK_MERGE () (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res2, res_ref1, 0.0001))
+ abort ();
+
+ MASK_MERGE () (res_ref2, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res3, res_ref2, 0.0001))
+ abort ();
+
+ MASK_ZERO () (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res4, res_ref1, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vfnmadd...pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 8 } } */
+/* { dg-final { scan-assembler-times "vfnmadd...pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 4 } } */
+/* { dg-final { scan-assembler-times "vfnmadd231pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vfnmadd...pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vfnmadd...pd\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfnmadd...pd\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfnmadd231pd\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfnmadd...pd\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x1, x2, x3;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x1 = _mm512_fnmadd_pd (x1, x2, x3);
+ x1 = _mm512_mask_fnmadd_pd (x1, m, x2, x3);
+ x3 = _mm512_mask3_fnmadd_pd (x1, x2, x3, m);
+ x1 = _mm512_maskz_fnmadd_pd (m, x1, x2, x3);
+ x1 = _mm512_fnmadd_round_pd (x1, x2, x3, _MM_FROUND_TO_NEAREST_INT);
+ x1 = _mm512_mask_fnmadd_round_pd (x1, m, x2, x3, _MM_FROUND_TO_NEG_INF);
+ x3 = _mm512_mask3_fnmadd_round_pd (x1, x2, x3, m, _MM_FROUND_TO_POS_INF);
+ x1 = _mm512_maskz_fnmadd_round_pd (m, x1, x2, x3, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (double *s1, double *s2, double *s3, double *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = -s1[i] * s2[i] + s3[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s1, s2, s3, res1, res2, res3, res4;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref1[SIZE];
+ double res_ref2[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = DEFAULT_VALUE;
+ s2.a[i] = 56.78 * (i + 1) * sign;
+ s3.a[i] = 90.12 * (i + 2) * sign;
+ sign = -sign;
+ }
+
+#if AVX512F_LEN == 512
+ res1.x = INTRINSIC (_fnmadd_pd) (s1.x, s2.x, s3.x);
+#endif
+ res2.x = INTRINSIC (_mask_fnmadd_pd) (s1.x, mask, s2.x, s3.x);
+ res3.x = INTRINSIC (_mask3_fnmadd_pd) (s2.x, s3.x, s1.x, mask);
+ res4.x = INTRINSIC (_maskz_fnmadd_pd) (mask, s1.x, s2.x, s3.x);
+
+ CALC (s1.a, s2.a, s3.a, res_ref1);
+ CALC (s2.a, s3.a, s1.a, res_ref2);
+
+#if AVX512F_LEN == 512
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res1, res_ref1, 0.0001))
+ abort ();
+#endif
+
+ MASK_MERGE (d) (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res2, res_ref1, 0.0001))
+ abort ();
+
+ MASK_MERGE (d) (res_ref2, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res3, res_ref2, 0.0001))
+ abort ();
+
+ MASK_ZERO (d) (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res4, res_ref1, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vfnmadd...ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 8 } } */
+/* { dg-final { scan-assembler-times "vfnmadd...ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 4 } } */
+/* { dg-final { scan-assembler-times "vfnmadd231ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vfnmadd...ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vfnmadd...ps\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfnmadd...ps\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfnmadd231ps\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfnmadd...ps\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x1, x2, x3;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x1 = _mm512_fnmadd_ps (x1, x2, x3);
+ x1 = _mm512_mask_fnmadd_ps (x1, m, x2, x3);
+ x3 = _mm512_mask3_fnmadd_ps (x1, x2, x3, m);
+ x1 = _mm512_maskz_fnmadd_ps (m, x1, x2, x3);
+ x1 = _mm512_fnmadd_round_ps (x1, x2, x3, _MM_FROUND_TO_NEAREST_INT);
+ x1 = _mm512_mask_fnmadd_round_ps (x1, m, x2, x3, _MM_FROUND_TO_NEG_INF);
+ x3 = _mm512_mask3_fnmadd_round_ps (x1, x2, x3, m, _MM_FROUND_TO_POS_INF);
+ x1 = _mm512_maskz_fnmadd_round_ps (m, x1, x2, x3, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (float *s1, float *s2, float *s3, float *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = -s1[i] * s2[i] + s3[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, ) s1, s2, s3, res1, res2, res3, res4;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref1[SIZE];
+ float res_ref2[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = DEFAULT_VALUE;
+ s2.a[i] = 56.78 * (i + 1) * sign;
+ s3.a[i] = 90.12 * (i + 2) * sign;
+ sign = -sign;
+ }
+
+#if AVX512F_LEN == 512
+ res1.x = INTRINSIC (_fnmadd_ps) (s1.x, s2.x, s3.x);
+#endif
+ res2.x = INTRINSIC (_mask_fnmadd_ps) (s1.x, mask, s2.x, s3.x);
+ res3.x = INTRINSIC (_mask3_fnmadd_ps) (s2.x, s3.x, s1.x, mask);
+ res4.x = INTRINSIC (_maskz_fnmadd_ps) (mask, s1.x, s2.x, s3.x);
+
+ CALC (s1.a, s2.a, s3.a, res_ref1);
+ CALC (s2.a, s3.a, s1.a, res_ref2);
+
+#if AVX512F_LEN == 512
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res1, res_ref1, 0.0001))
+ abort ();
+#endif
+
+ MASK_MERGE () (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res2, res_ref1, 0.0001))
+ abort ();
+
+ MASK_MERGE () (res_ref2, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res3, res_ref2, 0.0001))
+ abort ();
+
+ MASK_ZERO () (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res4, res_ref1, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vfnmsub...pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 8 } } */
+/* { dg-final { scan-assembler-times "vfnmsub...pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 4 } } */
+/* { dg-final { scan-assembler-times "vfnmsub231pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vfnmsub...pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vfnmsub...pd\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfnmsub...pd\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfnmsub231pd\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfnmsub...pd\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x1, x2, x3;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x1 = _mm512_fnmsub_pd (x1, x2, x3);
+ x1 = _mm512_mask_fnmsub_pd (x1, m, x2, x3);
+ x3 = _mm512_mask3_fnmsub_pd (x1, x2, x3, m);
+ x1 = _mm512_maskz_fnmsub_pd (m, x1, x2, x3);
+ x1 = _mm512_fnmsub_round_pd (x1, x2, x3, _MM_FROUND_TO_NEAREST_INT);
+ x1 = _mm512_mask_fnmsub_round_pd (x1, m, x2, x3, _MM_FROUND_TO_NEG_INF);
+ x3 = _mm512_mask3_fnmsub_round_pd (x1, x2, x3, m, _MM_FROUND_TO_POS_INF);
+ x1 = _mm512_maskz_fnmsub_round_pd (m, x1, x2, x3, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (double *s1, double *s2, double *s3, double *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = -s1[i] * s2[i] - s3[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s1, s2, s3, res1, res2, res3, res4;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref1[SIZE];
+ double res_ref2[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = DEFAULT_VALUE;
+ s2.a[i] = 56.78 * (i + 1) * sign;
+ s3.a[i] = 90.12 * (i + 2) * sign;
+ sign = -sign;
+ }
+
+#if AVX512F_LEN == 512
+ res1.x = INTRINSIC (_fnmsub_pd) (s1.x, s2.x, s3.x);
+#endif
+ res2.x = INTRINSIC (_mask_fnmsub_pd) (s1.x, mask, s2.x, s3.x);
+ res3.x = INTRINSIC (_mask3_fnmsub_pd) (s2.x, s3.x, s1.x, mask);
+ res4.x = INTRINSIC (_maskz_fnmsub_pd) (mask, s1.x, s2.x, s3.x);
+
+ CALC (s1.a, s2.a, s3.a, res_ref1);
+ CALC (s2.a, s3.a, s1.a, res_ref2);
+
+#if AVX512F_LEN == 512
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res1, res_ref1, 0.0001))
+ abort ();
+#endif
+
+ MASK_MERGE (d) (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res2, res_ref1, 0.0001))
+ abort ();
+
+ MASK_MERGE (d) (res_ref2, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res3, res_ref2, 0.0001))
+ abort ();
+
+ MASK_ZERO (d) (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res4, res_ref1, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vfnmsub...ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 8 } } */
+/* { dg-final { scan-assembler-times "vfnmsub...ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 4 } } */
+/* { dg-final { scan-assembler-times "vfnmsub231ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vfnmsub...ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vfnmsub...ps\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfnmsub...ps\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfnmsub231ps\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vfnmsub...ps\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x1, x2, x3;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x1 = _mm512_fnmsub_ps (x1, x2, x3);
+ x1 = _mm512_mask_fnmsub_ps (x1, m, x2, x3);
+ x3 = _mm512_mask3_fnmsub_ps (x1, x2, x3, m);
+ x1 = _mm512_maskz_fnmsub_ps (m, x1, x2, x3);
+ x1 = _mm512_fnmsub_round_ps (x1, x2, x3, _MM_FROUND_TO_NEAREST_INT);
+ x1 = _mm512_mask_fnmsub_round_ps (x1, m, x2, x3, _MM_FROUND_TO_NEG_INF);
+ x3 = _mm512_mask3_fnmsub_round_ps (x1, x2, x3, m, _MM_FROUND_TO_POS_INF);
+ x1 = _mm512_maskz_fnmsub_round_ps (m, x1, x2, x3, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (float *s1, float *s2, float *s3, float *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = -s1[i] * s2[i] - s3[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, ) s1, s2, s3, res1, res2, res3, res4;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref1[SIZE];
+ float res_ref2[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = DEFAULT_VALUE;
+ s2.a[i] = 56.78 * (i + 1) * sign;
+ s3.a[i] = 90.12 * (i + 2) * sign;
+ sign = -sign;
+ }
+
+#if AVX512F_LEN == 512
+ res1.x = INTRINSIC (_fnmsub_ps) (s1.x, s2.x, s3.x);
+#endif
+ res2.x = INTRINSIC (_mask_fnmsub_ps) (s1.x, mask, s2.x, s3.x);
+ res3.x = INTRINSIC (_mask3_fnmsub_ps) (s2.x, s3.x, s1.x, mask);
+ res4.x = INTRINSIC (_maskz_fnmsub_ps) (mask, s1.x, s2.x, s3.x);
+
+ CALC (s1.a, s2.a, s3.a, res_ref1);
+ CALC (s2.a, s3.a, s1.a, res_ref2);
+
+#if AVX512F_LEN == 512
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res1, res_ref1, 0.0001))
+ abort ();
+#endif
+
+ MASK_MERGE () (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res2, res_ref1, 0.0001))
+ abort ();
+
+ MASK_MERGE () (res_ref2, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res3, res_ref2, 0.0001))
+ abort ();
+
+ MASK_ZERO () (res_ref1, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, ) (res4, res_ref1, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vgetexppd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6} } */
+/* { dg-final { scan-assembler-times "vgetexppd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2} } */
+/* { dg-final { scan-assembler-times "vgetexppd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2} } */
+/* { dg-final { scan-assembler-times "vgetexppd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[^\{\]" 3} } */
+/* { dg-final { scan-assembler-times "vgetexppd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1} } */
+/* { dg-final { scan-assembler-times "vgetexppd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1} } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_getexp_pd (x);
+ x = _mm512_mask_getexp_pd (x, m, x);
+ x = _mm512_maskz_getexp_pd (m, x);
+ x = _mm512_getexp_round_pd (x, _MM_FROUND_NO_EXC);
+ x = _mm512_mask_getexp_round_pd (x, m, x, _MM_FROUND_NO_EXC);
+ x = _mm512_maskz_getexp_round_pd (m, x, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+#include "math.h"
+
+static void
+CALC (double *s, double *r)
+{
+ int i = 0;
+ for (i = 0; i < SIZE; i++)
+ r[i] = floor (log (s[i]) / log (2));
+}
+
+void static
+TEST (void)
+{
+ int j;
+ UNION_TYPE (AVX512F_LEN, d) res1, res2, res3, s;
+ double res_ref[SIZE];
+ double res_ref_mask[SIZE];
+
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (j = 0; j < SIZE; j++)
+ {
+ s.a[j] = j * (j + 12.0231);
+ res1.a[j] = DEFAULT_VALUE;
+ res2.a[j] = DEFAULT_VALUE;
+ res3.a[j] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_getexp_pd) (s.x);
+ res2.x = INTRINSIC (_mask_getexp_pd) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_getexp_pd) (mask, s.x);
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE(d) (res_ref, mask, SIZE);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO(d) (res_ref, mask, SIZE);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+
+}
+
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vgetexpps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6} } */
+/* { dg-final { scan-assembler-times "vgetexpps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2} } */
+/* { dg-final { scan-assembler-times "vgetexpps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2} } */
+/* { dg-final { scan-assembler-times "vgetexpps\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[^\{\]" 3} } */
+/* { dg-final { scan-assembler-times "vgetexpps\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1} } */
+/* { dg-final { scan-assembler-times "vgetexpps\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1} } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_getexp_ps (x);
+ x = _mm512_mask_getexp_ps (x, m, x);
+ x = _mm512_maskz_getexp_ps (m, x);
+ x = _mm512_getexp_round_ps (x, _MM_FROUND_NO_EXC);
+ x = _mm512_mask_getexp_round_ps (x, m, x, _MM_FROUND_NO_EXC);
+ x = _mm512_maskz_getexp_round_ps (m, x, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#include "math.h"
+
+static void
+CALC (float *s, float *r)
+{
+ int i = 0;
+ for (i = 0; i < SIZE; i++)
+ r[i] = floor (log (s[i]) / log (2));
+}
+
+void static
+TEST (void)
+{
+ int j;
+ UNION_TYPE (AVX512F_LEN, ) res1,res2,res3,s;
+ float res_ref[SIZE];
+ float res_ref_mask[SIZE];
+
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (j = 0; j < SIZE; j++)
+ {
+ s.a[j] = j * (j + 12.0231);
+ res1.a[j] = DEFAULT_VALUE;
+ res2.a[j] = DEFAULT_VALUE;
+ res3.a[j] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_getexp_ps) (s.x);
+ res2.x = INTRINSIC (_mask_getexp_ps) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_getexp_ps) (mask, s.x);
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, ) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE() (res_ref,mask,SIZE );
+
+ if (UNION_CHECK (AVX512F_LEN, ) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO() (res_ref,mask,SIZE );
+
+ if (UNION_CHECK (AVX512F_LEN, ) (res3, res_ref))
+ abort ();
+
+}
+
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vgetmantpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[\\n\]" 2 } } */
+/* { dg-final { scan-assembler-times "vgetmantpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vgetmantpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vgetmantpd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[\\n\]" 1 } } */
+/* { dg-final { scan-assembler-times "vgetmantpd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vgetmantpd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x, y;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_getmant_pd (y, _MM_MANT_NORM_p75_1p5, _MM_MANT_SIGN_src);
+ x =
+ _mm512_mask_getmant_pd (x, m, y, _MM_MANT_NORM_p75_1p5,
+ _MM_MANT_SIGN_src);
+ x =
+ _mm512_maskz_getmant_pd (m, y, _MM_MANT_NORM_p75_1p5,
+ _MM_MANT_SIGN_src);
+ x = _mm512_getmant_round_pd (y, _MM_MANT_NORM_p75_1p5, _MM_MANT_SIGN_src,
+ _MM_FROUND_NO_EXC);
+ x =
+ _mm512_mask_getmant_round_pd (x, m, y, _MM_MANT_NORM_p75_1p5,
+ _MM_MANT_SIGN_src, _MM_FROUND_NO_EXC);
+ x =
+ _mm512_maskz_getmant_round_pd (m, y, _MM_MANT_NORM_p75_1p5,
+ _MM_MANT_SIGN_src, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+#include <math.h>
+
+#ifndef GET_NORM_MANT
+#define GET_NORM_MANT
+
+union fp_int_t
+{
+ long long int int_val;
+ double fp_val;
+};
+
+double
+get_norm_mant (double source, int signctrl, int interv)
+{
+ long long src, sign, exp, fraction;
+ union fp_int_t bin_conv;
+
+ bin_conv.fp_val = source;
+ src = bin_conv.int_val;
+ sign = (signctrl & 0x1) ? 0 : (src >> 63);
+ exp = (src & 0x7ff0000000000000) >> 52;
+ fraction = (src & 0xfffffffffffff);
+
+ if (isnan (source))
+ return signbit (source) ? -NAN : NAN;
+ if (source == 0.0 || source == -0.0 || isinf (source))
+ return sign ? -1.0 : 1.0;
+ if (signbit (source) && (signctrl & 0x2))
+ return -NAN;
+ if (!isnormal (source))
+ {
+ src = (src & 0xfff7ffffffffffff);
+ exp = 0x3ff;
+ while (!(src & 0x8000000000000))
+ {
+ src += fraction & 0x8000000000000;
+ fraction = fraction << 1;
+ exp--;
+ }
+ }
+
+ switch (interv)
+ {
+ case 0:
+ exp = 0x3ff;
+ break;
+ case 1:
+ exp = ((exp - 0x3ff) & 0x1) ? 0x3fe : 0x3ff;
+ break;
+ case 2:
+ exp = 0x3fe;
+ break;
+ case 3:
+ exp = (fraction & 0x8000000000000) ? 0x3fe : 0x3ff;
+ break;
+ default:
+ abort ();
+ }
+
+ bin_conv.int_val = (sign << 63) | (exp << 52) | fraction;
+ return bin_conv.fp_val;
+}
+#endif
+
+CALC (double *r, double *s, int interv, int signctrl)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = get_norm_mant (s[i], signctrl, interv);
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, d) res1, res2, res3, src;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[SIZE];
+ int interv = _MM_MANT_NORM_p5_1;
+ int signctrl = _MM_MANT_SIGN_src;
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src.a[i] = 34.67 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_getmant_pd) (src.x, interv, signctrl);
+ res2.x =
+ INTRINSIC (_mask_getmant_pd) (res2.x, mask, src.x, interv,
+ signctrl);
+ res3.x =
+ INTRINSIC (_maskz_getmant_pd) (mask, src.x, interv, signctrl);
+
+ CALC (res_ref, src.a, interv, signctrl);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vgetmantps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[\\n\]" 2 } } */
+/* { dg-final { scan-assembler-times "vgetmantps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vgetmantps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vgetmantps\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[\\n\]" 1 } } */
+/* { dg-final { scan-assembler-times "vgetmantps\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vgetmantps\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x, y;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_getmant_ps (y, _MM_MANT_NORM_p75_1p5, _MM_MANT_SIGN_src);
+ x =
+ _mm512_mask_getmant_ps (x, m, y, _MM_MANT_NORM_p75_1p5,
+ _MM_MANT_SIGN_src);
+ x =
+ _mm512_maskz_getmant_ps (m, y, _MM_MANT_NORM_p75_1p5,
+ _MM_MANT_SIGN_src);
+ x = _mm512_getmant_round_ps (y, _MM_MANT_NORM_p75_1p5, _MM_MANT_SIGN_src,
+ _MM_FROUND_NO_EXC);
+ x =
+ _mm512_mask_getmant_round_ps (x, m, y, _MM_MANT_NORM_p75_1p5,
+ _MM_MANT_SIGN_src, _MM_FROUND_NO_EXC);
+ x =
+ _mm512_maskz_getmant_round_ps (m, y, _MM_MANT_NORM_p75_1p5,
+ _MM_MANT_SIGN_src, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#include <math.h>
+
+#ifndef GET_NORM_MANT
+#define GET_NORM_MANT
+
+union fp_int_t
+{
+ int int_val;
+ float fp_val;
+};
+
+float
+get_norm_mant (float source, int signctrl, int interv)
+{
+ int src, sign, exp, fraction;
+ union fp_int_t bin_conv;
+
+ bin_conv.fp_val = source;
+ src = bin_conv.int_val;
+ sign = (signctrl & 0x1) ? 0 : (src >> 31);
+ exp = (src & 0x7f800000) >> 23;
+ fraction = (src & 0x7fffff);
+
+ if (isnan (source))
+ return signbit (source) ? -NAN : NAN;
+ if (source == 0.0 || source == -0.0 || isinf (source))
+ return sign ? -1.0 : 1.0;
+ if (signbit (source) && (signctrl & 0x2))
+ return -NAN;
+ if (!isnormal (source))
+ {
+ src = (src & 0xffbfffff);
+ exp = 0x7f;
+ while (!(src & 0x400000))
+ {
+ src += fraction & 0x400000;
+ fraction = fraction << 1;
+ exp--;
+ }
+ }
+
+ switch (interv)
+ {
+ case 0:
+ exp = 0x7f;
+ break;
+ case 1:
+ exp = ((exp - 0x7f) & 0x1) ? 0x7e : 0x7f;
+ break;
+ case 2:
+ exp = 0x7e;
+ break;
+ case 3:
+ exp = (fraction & 0x400000) ? 0x7e : 0x7f;
+ break;
+ default:
+ abort ();
+ }
+
+ bin_conv.int_val = (sign << 31) | (exp << 23) | fraction;
+
+ return bin_conv.fp_val;
+}
+#endif
+
+CALC (float *r, float *s, int interv, int signctrl)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = get_norm_mant (s[i], signctrl, interv);
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN,) res1, res2, res3, src;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+ int interv = _MM_MANT_NORM_p5_1;
+ int signctrl = _MM_MANT_SIGN_src;
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src.a[i] = 34.67 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_getmant_ps) (src.x, interv, signctrl);
+ res2.x =
+ INTRINSIC (_mask_getmant_ps) (res2.x, mask, src.x, interv,
+ signctrl);
+ res3.x =
+ INTRINSIC (_maskz_getmant_ps) (mask, src.x, interv, signctrl);
+
+ CALC (res_ref, src.a, interv, signctrl);
+
+ if (UNION_CHECK (AVX512F_LEN,) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE ()(res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO ()(res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vinsertf32x4\[^\n\]*zmm" 3 } } */
+/* { dg-final { scan-assembler-times "vinsertf32x4\[^\n\]*\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vinsertf32x4\[^\n\]*\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+__m128 y;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_insertf32x4 (x, y, 1);
+ x = _mm512_maskz_insertf32x4 (6, x, y, 1);
+ x = _mm512_mask_insertf32x4 (x, 2, x, y, 1);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#include "string.h"
+
+void static
+CALC (UNION_TYPE (AVX512F_LEN,) s1, union128 s2, float *res_ref, int imm)
+{
+ memcpy (res_ref, s1.a, SIZE * sizeof (float));
+ memcpy (res_ref + imm * 4, s2.a, 16);
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN,) s1, res1, res2, res3;
+ union128 s2;
+ float res_ref[SIZE];
+ int j;
+
+ MASK_TYPE mask = 6 ^ (0xffd >> SIZE);
+
+ for (j = 0; j < SIZE; j++)
+ {
+ s1.a[j] = j * j / 10.2;
+ res1.a[j] = DEFAULT_VALUE;
+ res2.a[j] = DEFAULT_VALUE;
+ res3.a[j] = DEFAULT_VALUE;
+ }
+
+ for (j = 0; j < 4; j++)
+ s2.a[j] = j * j * j / 2.03;
+
+ res1.x = INTRINSIC (_insertf32x4) (s1.x, s2.x, 1);
+ res2.x = INTRINSIC (_mask_insertf32x4) (res2.x, mask, s1.x, s2.x, 1);
+ res3.x = INTRINSIC (_maskz_insertf32x4) (mask, s1.x, s2.x, 1);
+
+ CALC (s1, s2, res_ref, 1);
+
+ if (UNION_CHECK (AVX512F_LEN,) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE () (res_ref, mask, SIZE);
+
+ if (UNION_CHECK (AVX512F_LEN,) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO () (res_ref, mask, SIZE);
+
+ if (UNION_CHECK (AVX512F_LEN,) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vinsertf64x4\[ \\t\]+\[^\n\]+" 3 } } */
+/* { dg-final { scan-assembler-times "vinsertf64x4\[ \\t\]+\[^\n\]+\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vinsertf64x4\[ \\t\]+\[^\n\]+\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __m256d y;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_insertf64x4 (x, y, 1);
+ x = _mm512_mask_insertf64x4 (x, 2, x, y, 1);
+ x = _mm512_maskz_insertf64x4 (2, x, y, 1);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-require-effective-target avx512f } */
+/* { dg-options "-O2 -mavx512f" } */
+
+#define SIZE (512 / 64)
+#include "avx512f-mask-type.h"
+#include <string.h>
+#include "avx512f-check.h"
+#include "avx512f-mask-type.h"
+#include "avx512f-helper.h"
+
+void static
+avx512f_test (void)
+{
+ union512d s1, res, res2, res3;
+ union256d s2;
+ double res_ref[8];
+ MASK_TYPE mask = MASK_VALUE;
+ int j;
+
+ for (j = 0; j < 8; j++)
+ {
+ s1.a[j] = j * j + 1.6;
+ res2.a[j] = DEFAULT_VALUE;
+ }
+
+ for (j = 0; j < 4; j++)
+ s2.a[j] = j * j * j / 2.7;
+
+ res.x = _mm512_insertf64x4 (s1.x, s2.x, 0);
+ res2.x = _mm512_mask_insertf64x4 (res2.x, mask, s1.x, s2.x, 0);
+ res3.x = _mm512_maskz_insertf64x4 (mask, s1.x, s2.x, 0);
+
+ memcpy (res_ref, s1.a, 64);
+ memcpy (res_ref, s2.a, 32);
+
+ if (check_union512d (res, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (check_union512d (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (check_union512d (res3, res_ref))
+ abort ();
+
+ res.x = _mm512_insertf64x4 (s1.x, s2.x, 1);
+ res2.x = _mm512_mask_insertf64x4 (res2.x, mask, s1.x, s2.x, 1);
+ res3.x = _mm512_maskz_insertf64x4 (mask, s1.x, s2.x, 1);
+
+ memcpy (res_ref, s1.a, 64);
+ memcpy (res_ref + 4, s2.a, 32);
+
+ if (check_union512d (res, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (check_union512d (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (check_union512d (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vinserti32x4\[^\n\]*xmm\[^\n\]*zmm\[^\n\]*zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vinserti32x4\[^\n\]*\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vinserti32x4\[^\n\]*\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x,a;
+volatile __m128i y;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_maskz_inserti32x4 (6, x, y, 1);
+ x = _mm512_mask_inserti32x4 (a, 6, x, y, 1);
+ x = _mm512_inserti32x4 (x, y, 1);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#include "string.h"
+
+void static
+CALC (UNION_TYPE (AVX512F_LEN, i_d) s1, union128i_d s2, int *res_ref, int imm)
+{
+ memcpy (res_ref, s1.a, SIZE * sizeof (int));
+ memcpy (res_ref + imm * 4, s2.a, 16);
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) s1, res1, res2, res3;
+ union128i_d s2;
+ int res_ref[SIZE];
+ int j;
+
+ MASK_TYPE mask = 6 ^ (0xffd >> SIZE);
+
+ for (j = 0; j < SIZE; j++)
+ {
+ s1.a[j] = j * j;
+ res1.a[j] = DEFAULT_VALUE;
+ res2.a[j] = DEFAULT_VALUE;
+ res3.a[j] = DEFAULT_VALUE;
+ }
+
+ for (j = 0; j < 4; j++)
+ s2.a[j] = j * j * j;
+
+ res1.x = INTRINSIC (_inserti32x4) (s1.x, s2.x, 1);
+ res2.x = INTRINSIC (_mask_inserti32x4) (res2.x, mask, s1.x, s2.x, 1);
+ res3.x = INTRINSIC (_maskz_inserti32x4) (mask, s1.x, s2.x, 1);
+
+ CALC (s1, s2, res_ref, 1);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vinserti64x4\[ \\t\]+\[^\n\]+\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vinserti64x4\[ \\t\]+\[^\n\]+\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vinserti64x4\[ \\t\]+\[^\n\]+\[^\n\]" 3 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __m256i y;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_inserti64x4 (x, y, 1);
+ x = _mm512_mask_inserti64x4 (x, 2, x, y, 1);
+ x = _mm512_maskz_inserti64x4 (2, x, y, 1);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-require-effective-target avx512f } */
+/* { dg-options "-O2 -mavx512f" } */
+
+#define SIZE (512 / 64)
+#include "avx512f-mask-type.h"
+#include <string.h>
+#include "avx512f-check.h"
+#include "avx512f-mask-type.h"
+#include "avx512f-helper.h"
+
+void static
+avx512f_test (void)
+{
+ union512i_q s1, res, res2, res3;
+ union256i_q s2;
+ long long int res_ref[8];
+ MASK_TYPE mask = MASK_VALUE;
+ int j;
+
+ for (j = 0; j < 8; j++)
+ {
+ s1.a[j] = j * j;
+ res2.a[j] = DEFAULT_VALUE;
+ }
+
+ for (j = 0; j < 4; j++)
+ s2.a[j] = j * j * j;
+
+ res.x = _mm512_inserti64x4 (s1.x, s2.x, 0);
+ res2.x = _mm512_mask_inserti64x4 (res2.x, mask, s1.x, s2.x, 0);
+ res3.x = _mm512_maskz_inserti64x4 (mask, s1.x, s2.x, 0);
+
+ memcpy (res_ref, s1.a, 64);
+ memcpy (res_ref, s2.a, 32);
+
+ if (check_union512i_q (res, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (check_union512i_q (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (check_union512i_q (res3, res_ref))
+ abort ();
+
+ res.x = _mm512_inserti64x4 (s1.x, s2.x, 1);
+ res2.x = _mm512_mask_inserti64x4 (res2.x, mask, s1.x, s2.x, 1);
+ res3.x = _mm512_maskz_inserti64x4 (mask, s1.x, s2.x, 1);
+
+ memcpy (res_ref, s1.a, 64);
+ memcpy (res_ref + 4, s2.a, 32);
+
+ if (check_union512i_q (res, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (check_union512i_q (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (check_union512i_q (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vmaxpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vmaxpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vmaxpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vmaxpd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vmaxpd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmaxpd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_max_pd (x, x);
+ x = _mm512_mask_max_pd (x, m, x, x);
+ x = _mm512_maskz_max_pd (m, x, x);
+ x = _mm512_max_round_pd (x, x, _MM_FROUND_NO_EXC);
+ x = _mm512_mask_max_round_pd (x, m, x, x, _MM_FROUND_NO_EXC);
+ x = _mm512_maskz_max_round_pd (m, x, x, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (double *r, double *s1, double *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[i] > s2[i] ? s1[i] : s2[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, d) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 1.5 + 34.67 * i * sign;
+ src2.a[i] = -22.17 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_max_pd) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_max_pd) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_max_pd) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vmaxps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vmaxps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vmaxps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vmaxps\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vmaxps\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmaxps\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_max_ps (x, x);
+ x = _mm512_mask_max_ps (x, m, x, x);
+ x = _mm512_maskz_max_ps (m, x, x);
+ x = _mm512_max_round_ps (x, x, _MM_FROUND_NO_EXC);
+ x = _mm512_mask_max_round_ps (x, m, x, x, _MM_FROUND_NO_EXC);
+ x = _mm512_maskz_max_round_ps (m, x, x, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (float *r, float *s1, float *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[i] > s2[i] ? s1[i] : s2[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, ) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 1.5 + 34.67 * i * sign;
+ src2.a[i] = -22.17 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_max_ps) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_max_ps) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_max_ps) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, ) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vminpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vminpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vminpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vminpd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vminpd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vminpd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_min_pd (x, x);
+ x = _mm512_mask_min_pd (x, m, x, x);
+ x = _mm512_maskz_min_pd (m, x, x);
+ x = _mm512_min_round_pd (x, x, _MM_FROUND_NO_EXC);
+ x = _mm512_mask_min_round_pd (x, m, x, x, _MM_FROUND_NO_EXC);
+ x = _mm512_maskz_min_round_pd (m, x, x, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (double *r, double *s1, double *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[i] < s2[i] ? s1[i] : s2[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, d) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 1.5 + 34.67 * i * sign;
+ src2.a[i] = -22.17 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_min_pd) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_min_pd) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_min_pd) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vminps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vminps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vminps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vminps\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vminps\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vminps\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_min_ps (x, x);
+ x = _mm512_mask_min_ps (x, m, x, x);
+ x = _mm512_maskz_min_ps (m, x, x);
+ x = _mm512_min_round_ps (x, x, _MM_FROUND_NO_EXC);
+ x = _mm512_mask_min_round_ps (x, m, x, x, _MM_FROUND_NO_EXC);
+ x = _mm512_maskz_min_round_ps (m, x, x, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (float *r, float *s1, float *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[i] < s2[i] ? s1[i] : s2[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, ) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 1.5 + 34.67 * i * sign;
+ src2.a[i] = -22.17 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_min_ps) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_min_ps) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_min_ps) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, ) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vmovapd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmovapd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vmovapd\[ \\t\]+\[^\n\]*\\)\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmovapd\[ \\t\]+\[^\n\]*\\)\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vmovapd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*\\)\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+double *p;
+volatile __m512d x1, x2;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x1 = _mm512_mask_mov_pd (x1, m, x2);
+ x1 = _mm512_maskz_mov_pd (m, x2);
+
+ x1 = _mm512_load_pd (p);
+ x1 = _mm512_mask_load_pd (x1, m, p);
+ x1 = _mm512_maskz_load_pd (m, p);
+
+ _mm512_store_pd (p, x1);
+ _mm512_mask_store_pd (p, m, x1);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE ((AVX512F_LEN) / 64)
+#include "avx512f-mask-type.h"
+#define ALIGN ((AVX512F_LEN) / 8)
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s2, s3, res1, res3, res4, res5, res6;
+ MASK_TYPE mask = MASK_VALUE;
+ double s1[SIZE] __attribute__ ((aligned (ALIGN)));
+ double res2[SIZE] __attribute__ ((aligned (ALIGN)));
+ double res7[SIZE] __attribute__ ((aligned (ALIGN)));
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1[i] = 12.34 * (i + 2000) * sign;
+ s2.a[i] = 56.78 * (i - 30) * sign;
+ s3.a[i] = 90.12 * (i + 40) * sign;
+ res3.a[i] = DEFAULT_VALUE;
+ res5.a[i] = DEFAULT_VALUE;
+ res7[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+#if AVX512F_LEN == 512
+ res1.x = INTRINSIC (_load_pd) (s1);
+ INTRINSIC (_store_pd) (res2, s2.x);
+#endif
+ res3.x = INTRINSIC (_mask_mov_pd) (res3.x, mask, s3.x);
+ res4.x = INTRINSIC (_maskz_mov_pd) (mask, s3.x);
+ res5.x = INTRINSIC (_mask_load_pd) (res5.x, mask, s1);
+ res6.x = INTRINSIC (_maskz_load_pd) (mask, s1);
+ INTRINSIC (_mask_store_pd) (res7, mask, s2.x);
+
+#if AVX512F_LEN == 512
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, s1))
+ abort ();
+
+ if (UNION_CHECK (AVX512F_LEN, d) (s2, res2))
+ abort ();
+#endif
+
+ MASK_MERGE (d) (s3.a, mask, SIZE);
+ if (checkVd (res3.a, s3.a, SIZE))
+ abort ();
+
+ MASK_ZERO (d) (s3.a, mask, SIZE);
+ if (checkVd (res4.a, s3.a, SIZE))
+ abort ();
+
+ MASK_MERGE (d) (s1, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res5, s1))
+ abort ();
+
+ MASK_ZERO (d) (s1, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res6, s1))
+ abort ();
+
+ MASK_MERGE (d) (s2.a, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (s2, res7))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vmovaps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmovaps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vmovaps\[ \\t\]+\[^\n\]*\\)\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmovaps\[ \\t\]+\[^\n\]*\\)\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vmovaps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*\\)\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+float *p;
+volatile __m512 x1, x2;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x1 = _mm512_mask_mov_ps (x1, m, x2);
+ x1 = _mm512_maskz_mov_ps (m, x2);
+
+ x1 = _mm512_load_ps (p);
+ x1 = _mm512_mask_load_ps (x1, m, p);
+ x1 = _mm512_maskz_load_ps (m, p);
+
+ _mm512_store_ps (p, x1);
+ _mm512_mask_store_ps (p, m, x1);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE ((AVX512F_LEN) / 32)
+#include "avx512f-mask-type.h"
+#define ALIGN ((AVX512F_LEN) / 8)
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, ) s2, s3, res1, res3, res4, res5, res6;
+ MASK_TYPE mask = MASK_VALUE;
+ float s1[SIZE] __attribute__ ((aligned (ALIGN)));
+ float res2[SIZE] __attribute__ ((aligned (ALIGN)));
+ float res7[SIZE] __attribute__ ((aligned (ALIGN)));
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1[i] = 12.34 * (i + 2000) * sign;
+ s2.a[i] = 56.78 * (i - 30) * sign;
+ s3.a[i] = 90.12 * (i + 40) * sign;
+ res3.a[i] = DEFAULT_VALUE;
+ res5.a[i] = DEFAULT_VALUE;
+ res7[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+#if AVX512F_LEN == 512
+ res1.x = INTRINSIC (_load_ps) (s1);
+ INTRINSIC (_store_ps) (res2, s2.x);
+#endif
+ res3.x = INTRINSIC (_mask_mov_ps) (res3.x, mask, s3.x);
+ res4.x = INTRINSIC (_maskz_mov_ps) (mask, s3.x);
+ res5.x = INTRINSIC (_mask_load_ps) (res5.x, mask, s1);
+ res6.x = INTRINSIC (_maskz_load_ps) (mask, s1);
+ INTRINSIC (_mask_store_ps) (res7, mask, s2.x);
+
+#if AVX512F_LEN == 512
+ if (UNION_CHECK (AVX512F_LEN, ) (res1, s1))
+ abort ();
+
+ if (UNION_CHECK (AVX512F_LEN, ) (s2, res2))
+ abort ();
+#endif
+
+ MASK_MERGE () (s3.a, mask, SIZE);
+ if (checkVf (res3.a, s3.a, SIZE))
+ abort ();
+
+ MASK_ZERO () (s3.a, mask, SIZE);
+ if (checkVf (res4.a, s3.a, SIZE))
+ abort ();
+
+ MASK_MERGE () (s1, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res5, s1))
+ abort ();
+
+ MASK_ZERO () (s1, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res6, s1))
+ abort ();
+
+ MASK_MERGE () (s2.a, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (s2, res7))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vmovddup\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]|vunpcklpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vmovddup\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]|vunpcklpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmovddup\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}|vunpcklpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x1, x2;
+volatile __mmask8 m8;
+
+void extern
+avx512f_test (void)
+{
+ x1 = _mm512_movedup_pd (x2);
+ x1 = _mm512_mask_movedup_pd (x1, m8, x2);
+ x1 = _mm512_maskz_movedup_pd (m8, x2);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+void static
+CALC (double *s, double *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE/2; i++)
+ {
+ r[2 * i] = s[2 * i];
+ r[2 * i + 1] = s[2 * i];
+ }
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = i * 123.2 + 32.6;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_movedup_pd) (s.x);
+ res2.x = INTRINSIC (_mask_movedup_pd) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_movedup_pd) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vmovdqa32\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmovdqa32\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vmovdqa32\[ \\t\]+\[^\n\]*\\)\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmovdqa32\[ \\t\]+\[^\n\]*\\)\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vmovdqa32\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*\\)\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+int *p;
+volatile __m512i x1, x2;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x1 = _mm512_mask_mov_epi32 (x1, m, x2);
+ x1 = _mm512_maskz_mov_epi32 (m, x2);
+
+ x1 = _mm512_load_si512 (p);
+ x1 = _mm512_load_epi32 (p);
+ x1 = _mm512_mask_load_epi32 (x1, m, p);
+ x1 = _mm512_maskz_load_epi32 (m, p);
+
+ _mm512_store_si512 (p, x1);
+ _mm512_store_epi32 (p, x1);
+ _mm512_mask_store_epi32 (p, m, x1);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE ((AVX512F_LEN) / 32)
+#include "avx512f-mask-type.h"
+#define ALIGN ((AVX512F_LEN) / 8)
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) s2, s3, res1, res2, res5, res6, res7, res8;
+ MASK_TYPE mask = MASK_VALUE;
+ int s1[SIZE] __attribute__ ((aligned (ALIGN)));
+ int res3[SIZE] __attribute__ ((aligned (ALIGN)));
+ int res4[SIZE] __attribute__ ((aligned (ALIGN)));
+ int res9[SIZE] __attribute__ ((aligned (ALIGN)));
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1[i] = 1234 * (i + 2000) * sign;
+ s2.a[i] = 5678 * (i - 30) * sign;
+ s3.a[i] = 9012 * (i + 40) * sign;
+ res5.a[i] = DEFAULT_VALUE;
+ res7.a[i] = DEFAULT_VALUE;
+ res9[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+#if AVX512F_LEN == 512
+ res1.x = INTRINSIC (_load_si512) (s1);
+ res2.x = INTRINSIC (_load_epi32) (s1);
+ INTRINSIC (_store_si512) (res3, s2.x);
+ INTRINSIC (_store_epi32) (res4, s2.x);
+#endif
+ res5.x = INTRINSIC (_mask_mov_epi32) (res5.x, mask, s3.x);
+ res6.x = INTRINSIC (_maskz_mov_epi32) (mask, s3.x);
+ res7.x = INTRINSIC (_mask_load_epi32) (res7.x, mask, s1);
+ res8.x = INTRINSIC (_maskz_load_epi32) (mask, s1);
+ INTRINSIC (_mask_store_epi32) (res9, mask, s2.x);
+
+#if AVX512F_LEN == 512
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, s1))
+ abort ();
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, s1))
+ abort ();
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (s2, res3))
+ abort ();
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (s2, res4))
+ abort ();
+#endif
+
+ MASK_MERGE (i_d) (s3.a, mask, SIZE);
+ if (checkVi (res5.a, s3.a, SIZE))
+ abort ();
+
+ MASK_ZERO (i_d) (s3.a, mask, SIZE);
+ if (checkVi (res6.a, s3.a, SIZE))
+ abort ();
+
+ MASK_MERGE (i_d) (s1, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res7, s1))
+ abort ();
+
+ MASK_ZERO (i_d) (s1, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res8, s1))
+ abort ();
+
+ MASK_MERGE (i_d) (s2.a, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (s2, res9))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vmovdqa64\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmovdqa64\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vmovdqa64\[ \\t\]+\[^\n\]*\\)\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmovdqa64\[ \\t\]+\[^\n\]*\\)\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vmovdqa64\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*\\)\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+long long *p;
+volatile __m512i x1, x2;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x1 = _mm512_mask_mov_epi64 (x1, m, x2);
+ x1 = _mm512_maskz_mov_epi64 (m, x2);
+
+ x1 = _mm512_load_epi64 (p);
+ x1 = _mm512_mask_load_epi64 (x1, m, p);
+ x1 = _mm512_maskz_load_epi64 (m, p);
+
+ _mm512_store_epi64 (p, x1);
+ _mm512_mask_store_epi64 (p, m, x1);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE ((AVX512F_LEN) / 64)
+#include "avx512f-mask-type.h"
+#define ALIGN ((AVX512F_LEN) / 8)
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_q) s2, s3, res1, res3, res4, res5, res6;
+ MASK_TYPE mask = MASK_VALUE;
+ long long s1[SIZE] __attribute__ ((aligned (ALIGN)));
+ long long res2[SIZE] __attribute__ ((aligned (ALIGN)));
+ long long res7[SIZE] __attribute__ ((aligned (ALIGN)));
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1[i] = 1234 * (i + 2000) * sign;
+ s2.a[i] = 5678 * (i - 30) * sign;
+ s3.a[i] = 9012 * (i + 40) * sign;
+ res3.a[i] = DEFAULT_VALUE;
+ res5.a[i] = DEFAULT_VALUE;
+ res7[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+#if AVX512F_LEN == 512
+ res1.x = INTRINSIC (_load_epi64) (s1);
+ INTRINSIC (_store_epi64) (res2, s2.x);
+#endif
+ res3.x = INTRINSIC (_mask_mov_epi64) (res3.x, mask, s3.x);
+ res4.x = INTRINSIC (_maskz_mov_epi64) (mask, s3.x);
+ res5.x = INTRINSIC (_mask_load_epi64) (res5.x, mask, s1);
+ res6.x = INTRINSIC (_maskz_load_epi64) (mask, s1);
+ INTRINSIC (_mask_store_epi64) (res7, mask, s2.x);
+
+#if AVX512F_LEN == 512
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, s1))
+ abort ();
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (s2, res2))
+ abort ();
+#endif
+
+ MASK_MERGE (i_q) (s3.a, mask, SIZE);
+ if (checkVl (res3.a, s3.a, SIZE))
+ abort ();
+
+ MASK_ZERO (i_q) (s3.a, mask, SIZE);
+ if (checkVl (res4.a, s3.a, SIZE))
+ abort ();
+
+ MASK_MERGE (i_q) (s1, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res5, s1))
+ abort ();
+
+ MASK_ZERO (i_q) (s1, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res6, s1))
+ abort ();
+
+ MASK_MERGE (i_q) (s2.a, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (s2, res7))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vmovdqu32\[ \\t\]+\[^\n\]*\\)\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmovdqu32\[ \\t\]+\[^\n\]*\\)\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmovdqu32\[ \\t\]+\[^\n\]*\\)\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vmovdqu32\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*\\)\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmovdqu32\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*\\)\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+int *p;
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_loadu_si512 (p);
+ x = _mm512_mask_loadu_epi32 (x, m, p);
+ x = _mm512_maskz_loadu_epi32 (m, p);
+
+ _mm512_storeu_si512 (p, x);
+ _mm512_mask_storeu_epi32 (p, m, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE ((AVX512F_LEN) / 32)
+#include "avx512f-mask-type.h"
+
+typedef struct
+{
+ char c;
+ int a[SIZE];
+} __attribute__ ((packed)) EVAL(unaligned_array, AVX512F_LEN,);
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) s2, res1, res3, res4;
+ EVAL(unaligned_array, AVX512F_LEN,) s1, res2, res5;
+ MASK_TYPE mask = MASK_VALUE;
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 12345 * (i + 2000) * sign;
+ s2.a[i] = 67890 * (i + 2000) * sign;
+ res3.a[i] = DEFAULT_VALUE;
+ res5.a[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+#if AVX512F_LEN == 512
+ res1.x = _mm512_loadu_si512 (s1.a);
+ _mm512_storeu_si512 (res2.a, s2.x);
+#endif
+ res3.x = INTRINSIC (_mask_loadu_epi32) (res3.x, mask, s1.a);
+ res4.x = INTRINSIC (_maskz_loadu_epi32) (mask, s1.a);
+ INTRINSIC (_mask_storeu_epi32) (res5.a, mask, s2.x);
+
+#if AVX512F_LEN == 512
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, s1.a))
+ abort ();
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (s2, res2.a))
+ abort ();
+#endif
+
+ MASK_MERGE (i_d) (s1.a, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, s1.a))
+ abort ();
+
+ MASK_ZERO (i_d) (s1.a, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res4, s1.a))
+ abort ();
+
+ MASK_MERGE (i_d) (s2.a, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (s2, res5.a))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vmovdqu64\[ \\t\]+\[^\n\]*\\)\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmovdqu64\[ \\t\]+\[^\n\]*\\)\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vmovdqu64\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*\\)\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+long long *p;
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mask_loadu_epi64 (x, m, p);
+ x = _mm512_maskz_loadu_epi64 (m, p);
+
+ _mm512_mask_storeu_epi64 (p, m, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE ((AVX512F_LEN) / 64)
+#include "avx512f-mask-type.h"
+
+typedef struct
+{
+ char c;
+ long long a[SIZE];
+} __attribute__ ((packed)) EVAL(unaligned_array, AVX512F_LEN,);
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_q) s2, res1, res2;
+ EVAL(unaligned_array, AVX512F_LEN,) s1, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 12345 * (i + 2000) * sign;
+ s2.a[i] = 67890 * (i + 2000) * sign;
+ res1.a[i] = DEFAULT_VALUE;
+ res3.a[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+ res1.x = INTRINSIC (_mask_loadu_epi64) (res1.x, mask, s1.a);
+ res2.x = INTRINSIC (_maskz_loadu_epi64) (mask, s1.a);
+ INTRINSIC (_mask_storeu_epi64) (res3.a, mask, s2.x);
+
+ MASK_MERGE (i_q) (s1.a, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, s1.a))
+ abort ();
+
+ MASK_ZERO (i_q) (s1.a, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, s1.a))
+ abort ();
+
+ MASK_MERGE (i_q) (s2.a, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (s2, res3.a))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler "vmovntdq\[ \\t\]+\[^\n\]*%zmm\[0-9\]" } } */
+
+#include <immintrin.h>
+
+__m512i *x;
+volatile __m512i y;
+
+void extern
+avx512f_test (void)
+{
+ _mm512_stream_si512 (x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+void static
+avx512f_test (void)
+{
+ union512i_q s, res;
+
+ s.x = _mm512_set_epi64 (39578, -429496, 7856, 0, 85632, -1234, 47563, -1);
+ _mm512_stream_si512 (&res.x, s.x);
+
+ if (check_union512i_q (s, res.a))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler "vmovntpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]" } } */
+
+#include <immintrin.h>
+
+double *x;
+volatile __m512d y;
+
+void extern
+avx512f_test (void)
+{
+ _mm512_stream_pd (x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+void static
+avx512f_test (void)
+{
+ union512d s;
+ double res[8];
+
+ s.x = _mm512_set_pd (-39578.467285, 4294967295.1, -7856.342941, 0,
+ 85632.783567, 1234.9999, 47563.234215, -1.07);
+ _mm512_stream_pd (res, s.x);
+
+ if (check_union512d (s, res))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler "vmovntps\[ \\t\]+\[^\n\]*%zmm\[0-9\]" } } */
+
+#include <immintrin.h>
+
+float *x;
+volatile __m512 y;
+
+void extern
+avx512f_test (void)
+{
+ _mm512_stream_ps (x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-require-effective-target avx512f } */
+
+#include "avx512f-check.h"
+
+void static
+avx512f_test (void)
+{
+ union512 s;
+ float res[16];
+
+ s.x = _mm512_set_ps (-39578.467285, 4294967295.1, -7856.342941, 0,
+ 85632.783567, 1234.9999, 47563.234215, -1.07,
+ 3453.65743, -1234.9999, 67.234, -1,
+ 0.336624, 34534543, 4345.234234, -1.07234234);
+
+ _mm512_stream_ps (res, s.x);
+
+ if (check_union512 (s, res))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vmovshdup\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vmovshdup\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmovshdup\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_movehdup_ps (x);
+ x = _mm512_mask_movehdup_ps (x, m, x);
+ x = _mm512_maskz_movehdup_ps (m, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+void static
+CALC (float *s, float *r)
+{
+ int i;
+
+ for (i = 1; i < SIZE; i += 2)
+ {
+ r[i - 1] = r[i] = s[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, ) s, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = i * 123.2 + 32.6;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_movehdup_ps) (s.x);
+ res2.x = INTRINSIC (_mask_movehdup_ps) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_movehdup_ps) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, ) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vmovsldup\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vmovsldup\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmovsldup\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_moveldup_ps (x);
+ x = _mm512_mask_moveldup_ps (x, m, x);
+ x = _mm512_maskz_moveldup_ps (m, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+void static
+CALC (float *s, float *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i += 2)
+ {
+ r[i] = r[i + 1] = s[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, ) s, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = i * 123.2 + 32.6;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_moveldup_ps) (s.x);
+ res2.x = INTRINSIC (_mask_moveldup_ps) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_moveldup_ps) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, ) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vmovupd\[ \\t\]+\[^\n\]*\\)\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmovupd\[ \\t\]+\[^\n\]*\\)\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmovupd\[ \\t\]+\[^\n\]*\\)\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vmovupd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*\\)\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmovupd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*\\)\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+double *p;
+volatile __m512d x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_loadu_pd (p);
+ x = _mm512_mask_loadu_pd (x, m, p);
+ x = _mm512_maskz_loadu_pd (m, p);
+
+ _mm512_storeu_pd (p, x);
+ _mm512_mask_storeu_pd (p, m, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) res1, res2, res3, s2;
+ MASK_TYPE mask = MASK_VALUE;
+ double s1[SIZE];
+ double res4[SIZE];
+ double res5[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1[i] = 123.456 * (i + 2000) * sign;
+ s2.a[i] = 789.012 * (i + 3000) * sign;
+ res2.a[i] = DEFAULT_VALUE;
+ res5[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+ res1.x = INTRINSIC (_loadu_pd) (s1);
+ res2.x = INTRINSIC (_mask_loadu_pd) (res2.x, mask, s1);
+ res3.x = INTRINSIC (_maskz_loadu_pd) (mask, s1);
+ INTRINSIC (_storeu_pd) (res4, s2.x);
+ INTRINSIC (_mask_storeu_pd) (res5, mask, s2.x);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, s1))
+ abort ();
+
+ MASK_MERGE (d) (s1, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, s1))
+ abort ();
+
+ MASK_ZERO (d) (s1, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, s1))
+ abort ();
+
+ if (UNION_CHECK (AVX512F_LEN, d) (s2, res4))
+ abort ();
+
+ MASK_MERGE (d) (s2.a, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (s2, res5))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vmovups\[ \\t\]+\[^\n\]*\\)\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmovups\[ \\t\]+\[^\n\]*\\)\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmovups\[ \\t\]+\[^\n\]*\\)\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vmovups\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*\\)\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmovups\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*\\)\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+float *p;
+volatile __m512 x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_loadu_ps (p);
+ x = _mm512_mask_loadu_ps (x, m, p);
+ x = _mm512_maskz_loadu_ps (m, p);
+
+ _mm512_storeu_ps (p, x);
+ _mm512_mask_storeu_ps (p, m, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, ) res1, res2, res3, s2;
+ MASK_TYPE mask = MASK_VALUE;
+ float s1[SIZE];
+ float res4[SIZE];
+ float res5[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1[i] = 123.456 * (i + 2000) * sign;
+ s2.a[i] = 789.012 * (i + 3000) * sign;
+ res2.a[i] = DEFAULT_VALUE;
+ res5[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+ res1.x = INTRINSIC (_loadu_ps) (s1);
+ res2.x = INTRINSIC (_mask_loadu_ps) (res2.x, mask, s1);
+ res3.x = INTRINSIC (_maskz_loadu_ps) (mask, s1);
+ INTRINSIC (_storeu_ps) (res4, s2.x);
+ INTRINSIC (_mask_storeu_ps) (res5, mask, s2.x);
+
+ if (UNION_CHECK (AVX512F_LEN, ) (res1, s1))
+ abort ();
+
+ MASK_MERGE () (s1, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res2, s1))
+ abort ();
+
+ MASK_ZERO () (s1, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res3, s1))
+ abort ();
+
+ if (UNION_CHECK (AVX512F_LEN, ) (s2, res4))
+ abort ();
+
+ MASK_MERGE () (s2.a, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (s2, res5))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vmulpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vmulpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vmulpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vmulpd\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmulpd\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmulpd\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mul_pd (x, x);
+ x = _mm512_mask_mul_pd (x, m, x, x);
+ x = _mm512_maskz_mul_pd (m, x, x);
+ x = _mm512_mul_round_pd (x, x, _MM_FROUND_TO_NEAREST_INT);
+ x = _mm512_mask_mul_round_pd (x, m, x, x, _MM_FROUND_TO_NEG_INF);
+ x = _mm512_maskz_mul_round_pd (m, x, x, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (double *r, double *s1, double *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[i] * s2[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, d) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 1.5 + 34.67 * i * sign;
+ src2.a[i] = -22.17 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_mul_pd) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_mul_pd) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_mul_pd) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vmulps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vmulps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vmulps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vmulps\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmulps\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vmulps\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mul_ps (x, x);
+ x = _mm512_mask_mul_ps (x, m, x, x);
+ x = _mm512_maskz_mul_ps (m, x, x);
+ x = _mm512_mul_round_ps (x, x, _MM_FROUND_TO_NEAREST_INT);
+ x = _mm512_mask_mul_round_ps (x, m, x, x, _MM_FROUND_TO_POS_INF);
+ x = _mm512_maskz_mul_round_ps (m, x, x, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (float *r, float *s1, float *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[i] * s2[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN,) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 1.5 + 34.67 * i * sign;
+ src2.a[i] = -22.17 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_mul_ps) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_mul_ps) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_mul_ps) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN,) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (int *i1, int *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ if (i1[i] < 0)
+ r[i] = -i1[i];
+ else
+ r[i] = i1[i];
+}
+
+static void
+TEST (void)
+{
+ int ck[SIZE];
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_d) s, d, dm, dz;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = i * 7 + (i << 15) + 356;
+ d.a[i] = DEFAULT_VALUE;
+ dm.a[i] = DEFAULT_VALUE;
+ dz.a[i] = DEFAULT_VALUE;
+ }
+
+ CALC (s.a, ck);
+
+ d.x = INTRINSIC (_abs_epi32) (s.x);
+ dz.x = INTRINSIC (_maskz_abs_epi32) (mask, s.x);
+ dm.x = INTRINSIC (_mask_abs_epi32) (dm.x, mask, s.x);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (d, ck))
+ abort ();
+
+ MASK_MERGE (i_d) (ck, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (dm, ck))
+ abort ();
+
+ MASK_ZERO (i_d) (ck, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (dz, ck))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpabsd\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpabsd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpabsd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_abs_epi32 (x);
+ x = _mm512_maskz_abs_epi32 (7, x);
+ x = _mm512_mask_abs_epi32 (x, 6, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (long long *i1, long long *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ if (i1[i] < 0)
+ r[i] = -i1[i];
+ else
+ r[i] = i1[i];
+}
+
+static void
+TEST (void)
+{
+ long long ck[SIZE];
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_q) s, d, dm, dz;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = i * 7 + (i << 15) + 356;
+ d.a[i] = DEFAULT_VALUE;
+ dm.a[i] = DEFAULT_VALUE;
+ dz.a[i] = DEFAULT_VALUE;
+ }
+
+ CALC (s.a, ck);
+
+ d.x = INTRINSIC (_abs_epi64) (s.x);
+ dz.x = INTRINSIC (_maskz_abs_epi64) (mask, s.x);
+ dm.x = INTRINSIC (_mask_abs_epi64) (dm.x, mask, s.x);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (d, ck))
+ abort ();
+
+ MASK_MERGE (i_q) (ck, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (dm, ck))
+ abort ();
+
+ MASK_ZERO (i_q) (ck, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (dz, ck))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpabsq\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpabsq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpabsq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_abs_epi64 (x);
+ x = _mm512_maskz_abs_epi64 (2, x);
+ x = _mm512_mask_abs_epi64 (x, 3, x);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpaddd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpaddd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpaddd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_add_epi32 (x, x);
+ x = _mm512_mask_add_epi32 (x, m, x, x);
+ x = _mm512_maskz_add_epi32 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (int *r, int *s1, int *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[i] + s2[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 1.5 + 34.67 * i * sign;
+ src2.a[i] = -22.17 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_add_epi32) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_add_epi32) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_add_epi32) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpaddq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpaddq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpaddq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_add_epi64 (x, x);
+ x = _mm512_mask_add_epi64 (x, m, x, x);
+ x = _mm512_maskz_add_epi64 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (long long *r, long long *s1, long long *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[i] + s2[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 1.5 + 34.67 * i * sign;
+ src2.a[i] = -22.17 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_add_epi64) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_add_epi64) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_add_epi64) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpandd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 4 } } */
+/* { dg-final { scan-assembler-times "vpandd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpandd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_and_si512 (x, x);
+ x = _mm512_and_epi32 (x, x);
+ x = _mm512_mask_and_epi32 (x, m, x, x);
+ x = _mm512_maskz_and_epi32 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (int *s1, int *s2, int *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ r[i] = s1[i] & s2[i];
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) s1, s2, res1, res2, res3, res4;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = i * sign;
+ s2.a[i] = (i + 20) * sign;
+ sign = -sign;
+ res3.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_and_si512) (s1.x, s2.x);
+ res2.x = INTRINSIC (_and_epi32) (s1.x, s2.x);
+ res3.x = INTRINSIC (_mask_and_epi32) (res3.x, mask, s1.x, s2.x);
+ res4.x = INTRINSIC (_maskz_and_epi32) (mask, s1.x, s2.x);
+
+ CALC (s1.a, s2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res4, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpandnd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 4 } } */
+/* { dg-final { scan-assembler-times "vpandnd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpandnd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_andnot_si512 (x, x);
+ x = _mm512_andnot_epi32 (x, x);
+ x = _mm512_mask_andnot_epi32 (x, m, x, x);
+ x = _mm512_maskz_andnot_epi32 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (int *s1, int *s2, int *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ r[i] = (~s1[i]) & s2[i];
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) s1, s2, res1, res2, res3, res4;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = i * sign;
+ s2.a[i] = (i + 20) * sign;
+ sign = -sign;
+ res3.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_andnot_si512) (s1.x, s2.x);
+ res2.x = INTRINSIC (_andnot_epi32) (s1.x, s2.x);
+ res3.x = INTRINSIC (_mask_andnot_epi32) (res3.x, mask, s1.x, s2.x);
+ res4.x = INTRINSIC (_maskz_andnot_epi32) (mask, s1.x, s2.x);
+
+ CALC (s1.a, s2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res4, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpandnq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpandnq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpandnq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_andnot_epi64 (x, x);
+ x = _mm512_mask_andnot_epi64 (x,m, x, x);
+ x = _mm512_maskz_andnot_epi64 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (long long *s1, long long *s2, long long *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ r[i] = (~s1[i]) & s2[i];
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_q) s1, s2, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = i * sign;
+ s2.a[i] = (i + 20) * sign;
+ sign = -sign;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_andnot_epi64) (s1.x, s2.x);
+ res2.x = INTRINSIC (_mask_andnot_epi64) (res2.x, mask, s1.x, s2.x);
+ res3.x = INTRINSIC (_maskz_andnot_epi64) (mask, s1.x, s2.x);
+
+ CALC (s1.a, s2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpandq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpandq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpandq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_and_epi64 (x, x);
+ x = _mm512_mask_and_epi64 (x,m, x, x);
+ x = _mm512_maskz_and_epi64 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (long long *s1, long long *s2, long long *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ r[i] = s1[i] & s2[i];
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_q) s1, s2, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = i * sign;
+ s2.a[i] = (i + 20) * sign;
+ sign = -sign;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_and_epi64) (s1.x, s2.x);
+ res2.x = INTRINSIC (_mask_and_epi64) (res2.x, mask, s1.x, s2.x);
+ res3.x = INTRINSIC (_maskz_and_epi64) (mask, s1.x, s2.x);
+
+ CALC (s1.a, s2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler "(vpblendmd|vmovdqa32)\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mask_blend_epi32 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (int *r, int *s1, int *s2, MASK_TYPE mask)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = (mask & (1LL << i)) ? s2[i] : s1[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 15 + 3467 * i * sign;
+ src2.a[i] = -2217 * i * sign;
+ sign = sign * -1;
+ }
+
+ res1.x = INTRINSIC (_mask_blend_epi32) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a, mask);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler "(vpblendmq|vmovdqa64)\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mask_blend_epi64 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (long long *r, long long *s1, long long *s2, MASK_TYPE mask)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = (mask & (1LL << i)) ? s2[i] : s1[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 15 + 3467 * i * sign;
+ src2.a[i] = -2217 * i * sign;
+ sign = sign * -1;
+ }
+
+ res1.x = INTRINSIC (_mask_blend_epi64) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a, mask);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpbroadcastd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vpbroadcastd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vpbroadcastd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vpbroadcastd\[ \\t\]+%e\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpbroadcastd\[ \\t\]+%e\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpbroadcastd\[ \\t\]+%e\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __m128i y;
+volatile int z;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_broadcastd_epi32 (y);
+ x = _mm512_mask_broadcastd_epi32 (x, m, y);
+ x = _mm512_maskz_broadcastd_epi32 (m, y);
+
+ x = _mm512_set1_epi32 (z);
+ x = _mm512_mask_set1_epi32 (x, m, z);
+ x = _mm512_maskz_set1_epi32 (m, z);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (int *r, int *s)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s[0];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3;
+ UNION_TYPE (128, i_d) src;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < 4; i++)
+ {
+ src.a[i] = 1 + 34 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_broadcastd_epi32) (src.x);
+ res2.x = INTRINSIC (_mask_broadcastd_epi32) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_broadcastd_epi32) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+
+ res1.x = INTRINSIC (_set1_epi32) (src.a[0]);
+ res2.x = INTRINSIC (_mask_set1_epi32) (res2.x, mask, src.a[0]);
+ res3.x = INTRINSIC (_maskz_set1_epi32) (mask, src.a[0]);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpbroadcastq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vpbroadcastq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vpbroadcastq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vpbroadcastq\[ \\t\]+%r\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 { target { ! { ia32 } } } } } */
+/* { dg-final { scan-assembler-times "vpbroadcastq\[ \\t\]+%r\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 { target { ! { ia32 } } } } } */
+/* { dg-final { scan-assembler-times "vpbroadcastq\[ \\t\]+%r\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 { target { ! { ia32 } } } } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __m128i y;
+volatile long long z;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_broadcastq_epi64 (y);
+ x = _mm512_mask_broadcastq_epi64 (x, m, y);
+ x = _mm512_maskz_broadcastq_epi64 (m, y);
+
+ x = _mm512_set1_epi64 (z);
+ x = _mm512_mask_set1_epi64 (x, m, z);
+ x = _mm512_maskz_set1_epi64 (m, z);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (long long *r, long long *s)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s[0];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3;
+ UNION_TYPE (128, i_q) src;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < 2; i++)
+ {
+ src.a[i] = 1 + 34 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_broadcastq_epi64) (src.x);
+ res2.x = INTRINSIC (_mask_broadcastq_epi64) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_broadcastq_epi64) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+
+ res1.x = INTRINSIC (_set1_epi64) (src.a[0]);
+ res2.x = INTRINSIC (_mask_set1_epi64) (res2.x, mask, src.a[0]);
+ res3.x = INTRINSIC (_maskz_set1_epi64) (mask, src.a[0]);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler "vpcmpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n^k\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n^k\]*%k\[1-7\]\{" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ m = _mm512_cmp_epi32_mask (x, x, _MM_CMPINT_GE);
+ m = _mm512_mask_cmp_epi32_mask (m, x, x, _MM_CMPINT_NLE);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#include <math.h>
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+#if AVX512F_LEN == 512
+#define CMP(imm, rel) \
+ dst_ref = 0; \
+ for (i = 0; i < 16; i++) \
+ { \
+ dst_ref = ((rel) << i) | dst_ref; \
+ } \
+ source1.x = _mm512_loadu_si512 (s1); \
+ source2.x = _mm512_loadu_si512 (s2); \
+ dst1 = _mm512_cmp_epi32_mask (source1.x, source2.x, imm);\
+ dst2 = _mm512_mask_cmp_epi32_mask (mask, source1.x, source2.x, imm);\
+ if (dst_ref != dst1) abort(); \
+ if ((mask & dst_ref) != dst2) abort();
+#endif
+
+static void
+TEST ()
+{
+ UNION_TYPE (AVX512F_LEN, i_d) source1, source2;
+ MASK_TYPE dst1, dst2, dst_ref;
+ MASK_TYPE mask = MASK_VALUE;
+ int i;
+ int s1[16] = {2134, 6678, 453, 54646,
+ 231, 5674, 111, 23241,
+ 12314, 145, 671, 77575,
+ 23455, 166, 5321, 5673};
+ int s2[16] = {41124, 6678, 8653, 856,
+ 231, 4646, 111, 124,
+ 2745, 4567, 3676, 123,
+ 714, 3589, 5683, 5673};
+
+ CMP(0x00, s1[i] == s2[i]);
+ CMP(0x01, s1[i] < s2[i]);
+ CMP(0x02, s1[i] <= s2[i]);
+ CMP(0x03, 0);
+ CMP(0x04, s1[i] != s2[i]);
+ CMP(0x05, s1[i] >= s2[i]);
+ CMP(0x06, s1[i] > s2[i]);
+ CMP(0x07, 1);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpcmpeqd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n^k\]*%k\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpcmpeqd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n^k\]*%k\[0-9\]\{" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ m = _mm512_cmpeq_epi32_mask (x, x);
+ m = _mm512_mask_cmpeq_epi32_mask (3, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (MASK_TYPE *r, int *s1, int *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] == s2[i])
+ *r = *r | (one << i);
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_d) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+ res1 = 0;
+ res2 = 0;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmpeq_epi32_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmpeq_epi32_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpcmpeqq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n^k\]*%k\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpcmpeqq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n^k\]*%k\[0-9\]\{" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ m = _mm512_cmpeq_epi64_mask (x, x);
+ m = _mm512_mask_cmpeq_epi64_mask (3, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (MASK_TYPE *r, long long *s1, long long *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] == s2[i])
+ *r = *r | (one << i);
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_q) src1, src2;
+ MASK_TYPE res1, res2, res_ref;
+ MASK_TYPE mask = MASK_VALUE;
+ res1 = 0;
+ res2 = 0;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmpeq_epi64_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmpeq_epi64_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res1 != res_ref)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res2 != res_ref)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpcmpgtd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n^k\]*%k\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpcmpgtd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n^k\]*%k\[0-9\]\{" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ m = _mm512_cmpgt_epi32_mask (x, x);
+ m = _mm512_mask_cmpgt_epi32_mask (3, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (MASK_TYPE *r, int *s1, int *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] > s2[i])
+ *r = *r | (one << i);
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_d) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+ res1 = 0;
+ res2 = 0;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmpgt_epi32_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmpgt_epi32_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpcmpgtq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n^k\]*%k\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpcmpgtq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n^k\]*%k\[0-9\]\{" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ m = _mm512_cmpgt_epi64_mask (x, x);
+ m = _mm512_mask_cmpgt_epi64_mask (3, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (MASK_TYPE *r, long long *s1, long long *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] > s2[i])
+ *r = *r | (one << i);
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_q) src1, src2;
+ MASK_TYPE res1, res2, res_ref;
+ MASK_TYPE mask = MASK_VALUE;
+ res1 = 0;
+ res2 = 0;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmpgt_epi64_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmpgt_epi64_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res1 != res_ref)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res2 != res_ref)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler "vpcmpq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n^k\]*%k\[1-7\]\{" } } */
+/* { dg-final { scan-assembler "vpcmpq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n^k\]*%k\[1-7\]\[^\{\]" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ m = _mm512_cmp_epi64_mask (x, x, _MM_CMPINT_NE);
+ m = _mm512_mask_cmp_epi64_mask (m, x, x, _MM_CMPINT_NLT);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#include <math.h>
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+__mmask8 dst_ref;
+
+#define CMP(imm, rel) \
+ dst_ref = 0; \
+ for (i = 0; i < 8; i++) \
+ { \
+ dst_ref = ((rel) << i) | dst_ref; \
+ } \
+ source1.x = _mm512_loadu_si512 (s1); \
+ source2.x = _mm512_loadu_si512 (s2); \
+ dst1 = _mm512_cmp_epi64_mask (source1.x, source2.x, imm);\
+ dst2 = _mm512_mask_cmp_epi64_mask (mask, source1.x, source2.x, imm);\
+ if (dst_ref != dst1) abort(); \
+ if ((mask & dst_ref) != dst2) abort();
+
+static void
+TEST ()
+{
+ UNION_TYPE (AVX512F_LEN, i_d) source1, source2;
+ MASK_TYPE dst1, dst2, dst_ref;
+ MASK_TYPE mask = MASK_VALUE;
+ long long s1[8] = {2134, 6678, 453, 54646,
+ 231, 5674, 111, 23241};
+ long long s2[8] = {41124, 6678, 8653, 856,
+ 231, 4646, 111, 124};
+ int i;
+
+ CMP(0x00, s1[i] == s2[i]);
+ CMP(0x01, s1[i] < s2[i]);
+ CMP(0x02, s1[i] <= s2[i]);
+ CMP(0x03, 0);
+ CMP(0x04, s1[i] != s2[i]);
+ CMP(0x05, s1[i] >= s2[i]);
+ CMP(0x06, s1[i] > s2[i]);
+ CMP(0x07, 1);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n^k\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n^k\]*%k\[1-7\]\{" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ m = _mm512_cmp_epu32_mask (x, x, _MM_CMPINT_EQ);
+ m = _mm512_mask_cmp_epu32_mask (m, x, x, _MM_CMPINT_LT);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#include <math.h>
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+#if AVX512F_LEN == 512
+#define CMP(imm, rel) \
+ dst_ref = 0; \
+ for (i = 0; i < 16; i++) \
+ { \
+ dst_ref = ((rel) << i) | dst_ref; \
+ } \
+ source1.x = _mm512_loadu_si512 (s1); \
+ source2.x = _mm512_loadu_si512 (s2); \
+ dst1 = _mm512_cmp_epu32_mask (source1.x, source2.x, imm);\
+ dst2 = _mm512_mask_cmp_epu32_mask (mask, source1.x, source2.x, imm);\
+ if (dst_ref != dst1) abort(); \
+ if ((mask & dst_ref) != dst2) abort();
+#endif
+
+static void
+TEST ()
+{
+ unsigned int s1[16] = {2134, 6678, 453, 54646,
+ 231, 5674, 111, 23241,
+ 12314, 145, 671, 77575,
+ 23455, 166, 5321, 5673};
+ unsigned int s2[16] = {41124, 6678, 8653, 856,
+ 231, 4646, 111, 124,
+ 2745, 4567, 3676, 123,
+ 714, 3589, 5683, 5673};
+ UNION_TYPE (AVX512F_LEN, i_d) source1, source2;
+ MASK_TYPE dst1, dst2, dst_ref;
+ MASK_TYPE mask = MASK_VALUE;
+ int i;
+
+ CMP(0x00, s1[i] == s2[i]);
+ CMP(0x01, s1[i] < s2[i]);
+ CMP(0x02, s1[i] <= s2[i]);
+ CMP(0x03, 0);
+ CMP(0x04, s1[i] != s2[i]);
+ CMP(0x05, s1[i] >= s2[i]);
+ CMP(0x06, s1[i] > s2[i]);
+ CMP(0x07, 1);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n^k\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n^k\]*%k\[1-7\]\{" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ m = _mm512_cmp_epu64_mask (x, x, _MM_CMPINT_LE);
+ m = _mm512_mask_cmp_epu64_mask (m, x, x, _MM_CMPINT_UNUSED);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#include <math.h>
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+#if AVX512F_LEN == 512
+#define CMP(imm, rel) \
+ dst_ref = 0; \
+ for (i = 0; i < 8; i++) \
+ { \
+ dst_ref = ((rel) << i) | dst_ref; \
+ } \
+ source1.x = _mm512_loadu_si512 (s1); \
+ source2.x = _mm512_loadu_si512 (s2); \
+ dst1 = _mm512_cmp_epu64_mask (source1.x, source2.x, imm);\
+ dst2 = _mm512_mask_cmp_epu64_mask (mask, source1.x, source2.x, imm);\
+ if (dst_ref != dst1) abort(); \
+ if ((mask & dst_ref) != dst2) abort();
+#endif
+
+static void
+TEST ()
+{
+ UNION_TYPE (AVX512F_LEN, i_q) source1, source2;
+ MASK_TYPE dst1, dst2, dst_ref;
+ MASK_TYPE mask = MASK_VALUE;
+ int i;
+ unsigned long long s1[8] = {2134, 6678, 453, 54646,
+ 231, 5674, 111, 23241};
+ unsigned long long s2[8] = {41124, 6678, 8653, 856,
+ 231, 4646, 111, 124};
+
+ CMP(0x00, s1[i] == s2[i]);
+ CMP(0x01, s1[i] < s2[i]);
+ CMP(0x02, s1[i] <= s2[i]);
+ CMP(0x03, 0);
+ CMP(0x04, s1[i] != s2[i]);
+ CMP(0x05, s1[i] >= s2[i]);
+ CMP(0x06, s1[i] > s2[i]);
+ CMP(0x07, 1);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpcompressd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpcompressd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vpcompressd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*\\)\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+int *p;
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mask_compress_epi32 (x, m, x);
+ x = _mm512_maskz_compress_epi32 (m, x);
+
+ _mm512_mask_compressstoreu_epi32 (p, m, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#define MASK ((1 << SIZE) - 1)
+#include <x86intrin.h>
+
+static void
+CALC (int *s, int *r, MASK_TYPE mask)
+{
+ int i, k;
+
+ for (i = 0, k = 0; i < SIZE; i++)
+ {
+ if (mask & (1 << i))
+ r[k++] = s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) s, res1, res2;
+ int res3[SIZE];
+ MASK_TYPE compressed_mask, mask = MASK_VALUE;
+ int res_ref[SIZE];
+ int i, mask_bit_count, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 12345 * (i + 200) * sign;
+ res1.a[i] = DEFAULT_VALUE;
+ res3[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+ res1.x = INTRINSIC (_mask_compress_epi32) (res1.x, mask, s.x);
+ res2.x = INTRINSIC (_maskz_compress_epi32) (mask, s.x);
+ INTRINSIC (_mask_compressstoreu_epi32) (res3, mask, s.x);
+
+ mask_bit_count = __popcntd (mask & MASK);
+ compressed_mask = (1 << mask_bit_count) - 1;
+ CALC (s.a, res_ref, mask);
+
+ MASK_MERGE (i_d) (res_ref, compressed_mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, compressed_mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, compressed_mask, SIZE);
+ if (checkVi (res3, res_ref, SIZE))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpcompressq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpcompressq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vpcompressq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*\\)\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+long long *p;
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mask_compress_epi64 (x, m, x);
+ x = _mm512_maskz_compress_epi64 (m, x);
+
+ _mm512_mask_compressstoreu_epi64 (p, m, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+#define MASK ((1 << SIZE) - 1)
+#include <x86intrin.h>
+
+static void
+CALC (long long *s, long long *r, MASK_TYPE mask)
+{
+ int i, k;
+
+ for (i = 0, k = 0; i < SIZE; i++)
+ {
+ if (mask & (1 << i))
+ r[k++] = s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_q) s, res1, res2;
+ long long res3[SIZE];
+ MASK_TYPE compressed_mask, mask = MASK_VALUE;
+ long long res_ref[SIZE];
+ int i, mask_bit_count, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 12345 * (i + 200) * sign;
+ res1.a[i] = DEFAULT_VALUE;
+ res3[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+ res1.x = INTRINSIC (_mask_compress_epi64) (res1.x, mask, s.x);
+ res2.x = INTRINSIC (_maskz_compress_epi64) (mask, s.x);
+ INTRINSIC (_mask_compressstoreu_epi64) (res3, mask, s.x);
+
+ mask_bit_count = __popcntd (mask & MASK);
+ compressed_mask = (1 << mask_bit_count) - 1;
+ CALC (s.a, res_ref, mask);
+
+ MASK_MERGE (i_q) (res_ref, compressed_mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, compressed_mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, compressed_mask, SIZE);
+ if (checkVl (res3, res_ref, SIZE))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpermd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpermd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vpermd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_permutexvar_epi32 (x, x);
+ x = _mm512_maskz_permutexvar_epi32 (m, x, x);
+ x = _mm512_mask_permutexvar_epi32 (x, m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (int *src1, int *mask, int *dst)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ dst[i] = src1[mask[i] & 15];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = (i + 10) * (i + 10) * sign;
+ src2.a[i] = (i + 30);
+ sign = -sign;
+ res3.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_permutexvar_epi32) (src1.x, src2.x);
+ res2.x = INTRINSIC (_maskz_permutexvar_epi32) (mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_mask_permutexvar_epi32) (res3.x, mask, src1.x, src2.x);
+
+ CALC (src1.a, src2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpermi2d\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mask2_permutex2var_epi32 (x, x, m, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#include "math.h"
+#include "values.h"
+
+static void
+CALC (int *dst, int *src1, int *ind, int *src2)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ unsigned long long offset = ind[i] & (SIZE - 1);
+ unsigned long long cond = ind[i] & SIZE;
+
+ dst[i] = cond ? src2[offset] : src1[offset];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_d) s1, s2, res, ind;
+ int res_ref[SIZE];
+
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ ind.a[i] = DEFAULT_VALUE;
+ s1.a[i] = 34 * i + 1;
+ s2.a[i] = 34 * i;
+
+ res.a[i] = DEFAULT_VALUE;
+ }
+
+ CALC (res_ref, s1.a, ind.a, s2.a);
+
+ res.x =
+ INTRINSIC (_mask2_permutex2var_epi32) (s1.x, ind.x, mask, s2.x);
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpermi2pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __m512i y;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mask2_permutex2var_pd (x, y, m, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+#include "math.h"
+#include "values.h"
+
+static void
+CALC (double *dst, double *src1, long long *ind, double *src2)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ unsigned long long offset = ind[i] & (SIZE - 1);
+ unsigned long long cond = ind[i] & SIZE;
+
+ dst[i] = cond ? src2[offset] : src1[offset];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, k;
+ UNION_TYPE (AVX512F_LEN, d) s1, s2, res;
+ UNION_TYPE (AVX512F_LEN, i_q) ind;
+ double res_ref[SIZE];
+
+ union
+ {
+ double f;
+ long long i;
+ } ind_copy[SIZE];
+
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ /* Some of the integer indexes may be interpreted as floating point
+ values in mask-merge mode, that's why we use IND_COPY. */
+ ind.a[i] = ind_copy[i].i = 17 * (i << 1);
+ s1.a[i] = 42.5 * i + 1;
+ s2.a[i] = 22.5 * i;
+
+ res.a[i] = DEFAULT_VALUE;
+ }
+
+ CALC (res_ref, s1.a, ind.a, s2.a);
+
+ res.x = INTRINSIC (_mask2_permutex2var_pd) (s1.x, ind.x, mask, s2.x);
+
+ /* Standard MASK_MERGE cannot be used since VPERMI2PD in mask-merge mode
+ merges vectors of two different types (_m512d and __m512i). */
+ for (k = 0; k < SIZE; k++)
+ res_ref[k] = (mask & (1LL << k)) ? res_ref[k] : ind_copy[k].f;
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpermi2ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __m512i y;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mask2_permutex2var_ps (x, y, m, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#include "math.h"
+#include "values.h"
+
+static void
+CALC (float *dst, float *src1, int *ind, float *src2)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ unsigned long long offset = ind[i] & (SIZE - 1);
+ unsigned long long cond = ind[i] & SIZE;
+
+ dst[i] = cond ? src2[offset] : src1[offset];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, k;
+ UNION_TYPE (AVX512F_LEN,) s1, s2, res;
+ UNION_TYPE (AVX512F_LEN, i_d) ind;
+ float res_ref[SIZE];
+
+ union
+ {
+ float f;
+ int i;
+ } ind_copy[SIZE];
+
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ /* Some of the integer indexes may be interpreted as floating point
+ values in mask-merge mode, that's why we use IND_COPY. */
+ ind.a[i] = ind_copy[i].i = 17 * (i << 1);
+ s1.a[i] = 42.5 * i + 1;
+ s2.a[i] = 22.5 * i;
+
+ res.a[i] = DEFAULT_VALUE;
+ }
+
+ CALC (res_ref, s1.a, ind.a, s2.a);
+
+ res.x = INTRINSIC (_mask2_permutex2var_ps) (s1.x, ind.x, mask, s2.x);
+
+ /* Standard MASK_MERGE cannot be used since VPERMI2PS in mask-merge mode
+ merges vectors of two different types (_m512 and __m512i). */
+ for (k = 0; k < SIZE; k++)
+ res_ref[k] = (mask & (1LL << k)) ? res_ref[k] : ind_copy[k].f;
+
+ if (UNION_CHECK (AVX512F_LEN,) (res, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpermi2q\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mask2_permutex2var_epi64 (x, x, m, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+#include "math.h"
+#include "values.h"
+
+static void
+CALC (long long *dst, long long *src1, long long *ind, long long *src2)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ unsigned long long offset = ind[i] & (SIZE - 1);
+ unsigned long long cond = ind[i] & SIZE;
+
+ dst[i] = cond ? src2[offset] : src1[offset];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_q) s1, s2, res, ind;
+ long long res_ref[SIZE];
+
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ ind.a[i] = DEFAULT_VALUE;
+ s1.a[i] = 34 * i + 1;
+ s2.a[i] = 34 * i;
+
+ res.a[i] = DEFAULT_VALUE;
+ }
+
+ CALC (res_ref, s1.a, ind.a, s2.a);
+
+ res.x =
+ INTRINSIC (_mask2_permutex2var_epi64) (s1.x, ind.x, mask, s2.x);
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpermilpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpermilpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpermilpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __m512i c;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_permutevar_pd (x, c);
+ x = _mm512_mask_permutevar_pd (x, m, x, c);
+ x = _mm512_maskz_permutevar_pd (m, x, c);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+#ifndef CTRL
+#define CTRL 6
+#endif
+
+#undef mask_v
+#define mask_v(pos) (((CTRL & (1ULL << (pos))) >> (pos)) << 1)
+
+static void
+CALC (double *s1, long long *s2, double *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ r[i] = s1[(2 * (i / 2)) + ((s2[i] & 0x02) >> 1)];
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s1, res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN, i_q) s2;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = i + 10.;
+ s2.a[i] = mask_v (i);
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_permutevar_pd) (s1.x, s2.x);
+ res2.x = INTRINSIC (_mask_permutevar_pd) (res2.x, mask, s1.x, s2.x);
+ res3.x = INTRINSIC (_maskz_permutevar_pd) (mask, s1.x, s2.x);
+
+ CALC (s1.a, s2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpermilpd\[ \\t\]+\[^\n\]*13\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpermilpd\[ \\t\]+\[^\n\]*13\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpermilpd\[ \\t\]+\[^\n\]*13\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_permute_pd (x, 13);
+ x = _mm512_mask_permute_pd (x, m, x, 13);
+ x = _mm512_maskz_permute_pd (m, x, 13);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+#ifndef CTRL
+#define CTRL 129
+#endif
+
+static void
+CALC (double *s1, int s2, double *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = (s2 & (1 << i)) ? s1[1 + 2 * (i / 2)] : s1[2 * (i / 2)];
+ }
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s1, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = i + 10.;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_permute_pd) (s1.x, CTRL);
+ res2.x = INTRINSIC (_mask_permute_pd) (res2.x, mask, s1.x, CTRL);
+ res3.x = INTRINSIC (_maskz_permute_pd) (mask, s1.x, CTRL);
+
+ CALC (s1.a, CTRL, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpermilps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpermilps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpermilps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __m512i c;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_permutevar_ps (x, c);
+ x = _mm512_mask_permutevar_ps (x, m, x, c);
+ x = _mm512_maskz_permutevar_ps (m, x, c);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+#ifndef CTRL
+#define CTRL 233
+#endif
+
+#undef mask_v
+#define mask_v(pos) ((CTRL & (0x3 << (pos))) >> (pos))
+
+static void
+CALC (float *s1, int *s2, float *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ r[i] = s1[(4 * (i / 4)) + (s2[i] & 0x03)];
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN,) s1, res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN, i_d) s2;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = i + 10.;
+ s2.a[i] = mask_v (i);
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_permutevar_ps) (s1.x, s2.x);
+ res2.x = INTRINSIC (_mask_permutevar_ps) (res2.x, mask, s1.x, s2.x);
+ res3.x = INTRINSIC (_maskz_permutevar_ps) (mask, s1.x, s2.x);
+
+ CALC (s1.a, s2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN,) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE ()(res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO ()(res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpermilps\[ \\t\]+\[^\n\]*13\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpermilps\[ \\t\]+\[^\n\]*13\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpermilps\[ \\t\]+\[^\n\]*13\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_permute_ps (x, 13);
+ x = _mm512_mask_permute_ps (x, m, x, 13);
+ x = _mm512_maskz_permute_ps (m, x, 13);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+#ifndef CTRL
+#define CTRL 129
+#endif
+
+#ifndef SELECT4_DEFINED
+#define SELECT4_DEFINED
+static int
+select4 (int i, unsigned ctrl)
+{
+ int res;
+ switch (i % 4)
+ {
+ case 0:
+ res = (CTRL & 0x03);
+ break;
+ case 1:
+ res = ((CTRL & 0x0c) >> 2);
+ break;
+ case 2:
+ res = ((CTRL & 0x30) >> 4);
+ break;
+ case 3:
+ res = ((CTRL & 0xc0) >> 6);
+ break;
+ }
+ return res;
+}
+#endif
+
+static void
+CALC (float *s, float *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s[(4 * (i / 4)) + select4 (i, CTRL)];
+ }
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN,) s1, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = i + 10.;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_permute_ps) (s1.x, CTRL);
+ res2.x = INTRINSIC (_mask_permute_ps) (res2.x, mask, s1.x, CTRL);
+ res3.x = INTRINSIC (_maskz_permute_ps) (mask, s1.x, CTRL);
+
+ CALC (s1.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN,) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE ()(res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO ()(res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpermpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpermpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpermpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __m512d y;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm512_permutexvar_pd (x, y);
+ y = _mm512_mask_permutexvar_pd (y, m, x, y);
+ y = _mm512_maskz_permutexvar_pd (m, x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (double *s1, long long *mask, double *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[mask[i] & 7 % SIZE];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) src1, res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN, i_q) src2;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = i * sign;
+ src2.a[i] = i + 20;
+ sign = -sign;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_permutexvar_pd) (src2.x, src1.x);
+ res2.x = INTRINSIC (_mask_permutexvar_pd) (res2.x, mask, src2.x, src1.x);
+ res3.x = INTRINSIC (_maskz_permutexvar_pd) (mask, src2.x, src1.x);
+
+ CALC (src1.a, src2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpermpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpermpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpermpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_permutex_pd (x, 13);
+ x = _mm512_mask_permutex_pd (x, m, x, 13);
+ x = _mm512_maskz_permutex_pd (m, x, 13);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+#define N 0x7c
+
+static void
+CALC (double *s1, double *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ int index = (N >> ((i % 4) * 2)) & 3;
+ int base = i / 4;
+ r[i] = s1[4 * base + index];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) src1, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = i * i * sign;
+ res2.a[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+ res1.x = INTRINSIC (_permutex_pd) (src1.x, N);
+ res2.x = INTRINSIC (_mask_permutex_pd) (res2.x, mask, src1.x, N);
+ res3.x = INTRINSIC (_maskz_permutex_pd) (mask, src1.x, N);
+
+ CALC (src1.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpermps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpermps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpermps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __m512 y;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ y = _mm512_permutexvar_ps (x, y);
+ y = _mm512_mask_permutexvar_ps (y, m, x, y);
+ y = _mm512_maskz_permutexvar_ps (m, x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (float *s1, int *mask, float *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[mask[i] & 15 % SIZE];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, ) src1, res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN, i_d) src2;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = i * sign;
+ src2.a[i] = i + 20;
+ sign = -sign;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_permutexvar_ps) (src2.x, src1.x);
+ res2.x = INTRINSIC (_mask_permutexvar_ps) (res2.x, mask, src2.x, src1.x);
+ res3.x = INTRINSIC (_maskz_permutexvar_ps) (mask, src2.x, src1.x);
+
+ CALC (src1.a, src2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, ) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpermq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpermq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpermq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_permutex_epi64 (x, 13);
+ x = _mm512_mask_permutex_epi64 (x, m, x, 13);
+ x = _mm512_maskz_permutex_epi64 (m, x, 13);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+#define IMM_MASK 0x7c
+
+static void
+CALC (long long *src1, int mask, long long *dst)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ int index = ((mask >> (2 * (i % 4))) & 3);
+ int base = i / 4;
+ dst[i] = src1[4 * base + index];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3, src1;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = (i + 10) * (i + 10) * sign;
+ sign = -sign;
+ res3.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_permutex_epi64) (src1.x, IMM_MASK);
+ res2.x = INTRINSIC (_maskz_permutex_epi64) (mask, src1.x, IMM_MASK);
+ res3.x = INTRINSIC (_mask_permutex_epi64) (res3.x, mask, src1.x, IMM_MASK);
+
+ CALC (src1.a, IMM_MASK, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpermq\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpermq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vpermq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_permutexvar_epi64 (x, x);
+ x = _mm512_maskz_permutexvar_epi64 (m, x, x);
+ x = _mm512_mask_permutexvar_epi64 (x, m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (long long *src1, long long *mask, long long *dst)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ dst[i] = src1[mask[i] & 7];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = (i + 10) * (i + 10) * sign;
+ src2.a[i] = 2 * i + 10;
+ sign = -sign;
+ res3.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_permutexvar_epi64) (src1.x, src2.x);
+ res2.x = INTRINSIC (_maskz_permutexvar_epi64) (mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_mask_permutexvar_epi64) (res3.x, mask, src1.x, src2.x);
+
+ CALC (src1.a, src2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpermt2d\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpermt2d\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpermt2d\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_permutex2var_epi32 (x, x, x);
+ x = _mm512_mask_permutex2var_epi32 (x, m, x, x);
+ x = _mm512_maskz_permutex2var_epi32 (m, x, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#include "math.h"
+#include "values.h"
+
+static void
+CALC (int *dst, int *src1, int *ind, int *src2)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ unsigned long long offset = ind[i] & (SIZE - 1);
+ unsigned long long cond = ind[i] & SIZE;
+
+ dst[i] = cond ? src2[offset] : src1[offset];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_d) s1, s2, res1, res2, res3, ind;
+ int res_ref[SIZE];
+
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ ind.a[i] = 17 * (i << 1);
+ s1.a[i] = DEFAULT_VALUE;
+ s2.a[i] = 34 * i;
+
+ res1.a[i] = DEFAULT_VALUE;
+ res2.a[i] = DEFAULT_VALUE;
+ res3.a[i] = DEFAULT_VALUE;
+ }
+
+ CALC (res_ref, s1.a, ind.a, s2.a);
+
+ res1.x = INTRINSIC (_permutex2var_epi32) (s1.x, ind.x, s2.x);
+ res2.x =
+ INTRINSIC (_mask_permutex2var_epi32) (s1.x, mask, ind.x, s2.x);
+ res3.x =
+ INTRINSIC (_maskz_permutex2var_epi32) (mask, s1.x, ind.x, s2.x);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpermt2pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpermt2pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpermt2pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __m512i y;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_permutex2var_pd (x, y, x);
+ x = _mm512_mask_permutex2var_pd (x, m, y, x);
+ x = _mm512_maskz_permutex2var_pd (m, x, y, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+#include "math.h"
+#include "values.h"
+
+static void
+CALC (double *dst, double *src1, long long *ind, double *src2)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ unsigned long long offset = ind[i] & (SIZE - 1);
+ unsigned long long cond = ind[i] & SIZE;
+
+ dst[i] = cond ? src2[offset] : src1[offset];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, d) s1, s2, res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN, i_q) ind;
+ double res_ref[SIZE];
+
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ ind.a[i] = 17 * (i << 1);
+ s1.a[i] = DEFAULT_VALUE;
+ s2.a[i] = 22.5 * i;
+
+ res1.a[i] = DEFAULT_VALUE;
+ res2.a[i] = DEFAULT_VALUE;
+ res3.a[i] = DEFAULT_VALUE;
+ }
+
+ CALC (res_ref, s1.a, ind.a, s2.a);
+
+ res1.x = INTRINSIC (_permutex2var_pd) (s1.x, ind.x, s2.x);
+ res2.x = INTRINSIC (_mask_permutex2var_pd) (s1.x, mask, ind.x, s2.x);
+ res3.x =
+ INTRINSIC (_maskz_permutex2var_pd) (mask, s1.x, ind.x, s2.x);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpermt2ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpermt2ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpermt2ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __m512i y;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_permutex2var_ps (x, y, x);
+ x = _mm512_mask_permutex2var_ps (x, m, y, x);
+ x = _mm512_maskz_permutex2var_ps (m, x, y, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#include "math.h"
+#include "values.h"
+
+static void
+CALC (float *dst, float *src1, int *ind, float *src2)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ unsigned long long offset = ind[i] & (SIZE - 1);
+ unsigned long long cond = ind[i] & SIZE;
+
+ dst[i] = cond ? src2[offset] : src1[offset];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN,) s1, s2, res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN, i_d) ind;
+ float res_ref[SIZE];
+
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ ind.a[i] = 17 * (i << 1);
+ s1.a[i] = DEFAULT_VALUE;
+ s2.a[i] = 22.5 * i;
+
+ res1.a[i] = DEFAULT_VALUE;
+ res2.a[i] = DEFAULT_VALUE;
+ res3.a[i] = DEFAULT_VALUE;
+ }
+
+ CALC (res_ref, s1.a, ind.a, s2.a);
+
+ res1.x = INTRINSIC (_permutex2var_ps) (s1.x, ind.x, s2.x);
+ res2.x = INTRINSIC (_mask_permutex2var_ps) (s1.x, mask, ind.x, s2.x);
+ res3.x =
+ INTRINSIC (_maskz_permutex2var_ps) (mask, s1.x, ind.x, s2.x);
+
+ if (UNION_CHECK (AVX512F_LEN,) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE ()(res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO ()(res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpermt2q\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpermt2q\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpermt2q\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_permutex2var_epi64 (x, x, x);
+ x = _mm512_mask_permutex2var_epi64 (x, m, x, x);
+ x = _mm512_maskz_permutex2var_epi64 (m, x, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+#include "math.h"
+#include "values.h"
+
+static void
+CALC (long long *dst, long long *src1, long long *ind, long long *src2)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ unsigned long long offset = ind[i] & (SIZE - 1);
+ unsigned long long cond = ind[i] & SIZE;
+
+ dst[i] = cond ? src2[offset] : src1[offset];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_q) s1, s2, res1, res2, res3, ind;
+ long long res_ref[SIZE];
+
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ ind.a[i] = 17 * (i << 1);
+ s1.a[i] = DEFAULT_VALUE;
+ s2.a[i] = 34 * i;
+
+ res1.a[i] = DEFAULT_VALUE;
+ res2.a[i] = DEFAULT_VALUE;
+ res3.a[i] = DEFAULT_VALUE;
+ }
+
+ CALC (res_ref, s1.a, ind.a, s2.a);
+
+ res1.x = INTRINSIC (_permutex2var_epi64) (s1.x, ind.x, s2.x);
+ res2.x =
+ INTRINSIC (_mask_permutex2var_epi64) (s1.x, mask, ind.x, s2.x);
+ res3.x =
+ INTRINSIC (_maskz_permutex2var_epi64) (mask, s1.x, ind.x, s2.x);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpexpandd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vpexpandd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+
+#include <immintrin.h>
+
+int *p;
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mask_expand_epi32 (x, m, x);
+ x = _mm512_maskz_expand_epi32 (m, x);
+
+ x = _mm512_mask_expandloadu_epi32 (x, m, p);
+ x = _mm512_maskz_expandloadu_epi32 (m, p);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (int *s, int *r, MASK_TYPE mask)
+{
+ int i, k;
+
+ for (i = 0, k = 0; i < SIZE; i++)
+ {
+ if (mask & (1 << i))
+ r[i] = s[k++];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) s1, res1, res2, res3, res4;
+ MASK_TYPE mask = MASK_VALUE;
+ int s2[SIZE];
+ int res_ref1[SIZE];
+ int res_ref2[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 12345 * (i + 200) * sign;
+ s2[i] = 67890 * (i + 300) * sign;
+ res1.a[i] = DEFAULT_VALUE;
+ res3.a[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+ res1.x = INTRINSIC (_mask_expand_epi32) (res1.x, mask, s1.x);
+ res2.x = INTRINSIC (_maskz_expand_epi32) (mask, s1.x);
+ res3.x = INTRINSIC (_mask_expandloadu_epi32) (res3.x, mask, s2);
+ res4.x = INTRINSIC (_maskz_expandloadu_epi32) (mask, s2);
+
+ CALC (s1.a, res_ref1, mask);
+ CALC (s2, res_ref2, mask);
+
+ MASK_MERGE (i_d) (res_ref1, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref1))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref1, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref1))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref2, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref2))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref2, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res4, res_ref2))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpexpandq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vpexpandq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+
+#include <immintrin.h>
+
+long long *p;
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mask_expand_epi64 (x, m, x);
+ x = _mm512_maskz_expand_epi64 (m, x);
+
+ x = _mm512_mask_expandloadu_epi64 (x, m, p);
+ x = _mm512_maskz_expandloadu_epi64 (m, p);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (long long *s, long long *r, MASK_TYPE mask)
+{
+ int i, k;
+
+ for (i = 0, k = 0; i < SIZE; i++)
+ {
+ if (mask & (1 << i))
+ r[i] = s[k++];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_q) s1, res1, res2, res3, res4;
+ MASK_TYPE mask = MASK_VALUE;
+ long long s2[SIZE];
+ long long res_ref1[SIZE];
+ long long res_ref2[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 12345 * (i + 200) * sign;
+ s2[i] = 67890 * (i + 300) * sign;
+ res1.a[i] = DEFAULT_VALUE;
+ res3.a[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+ res1.x = INTRINSIC (_mask_expand_epi64) (res1.x, mask, s1.x);
+ res2.x = INTRINSIC (_maskz_expand_epi64) (mask, s1.x);
+ res3.x = INTRINSIC (_mask_expandloadu_epi64) (res3.x, mask, s2);
+ res4.x = INTRINSIC (_maskz_expandloadu_epi64) (mask, s2);
+
+ CALC (s1.a, res_ref1, mask);
+ CALC (s2, res_ref2, mask);
+
+ MASK_MERGE (i_q) (res_ref1, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref1))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref1, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref1))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref2, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref2))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref2, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res4, res_ref2))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmaxsd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpmaxsd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmaxsd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_max_epi32 (x, x);
+ x = _mm512_mask_max_epi32 (x, m, x, x);
+ x = _mm512_maskz_max_epi32 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+
+CALC (int *src1, int *src2, int *dst)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ dst[i] = src1[i] > src2[i] ? src1[i] : src2[i];
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_d) src1, src2, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = i * sign;
+ src2.a[i] = (i + 2000) * sign;
+ sign = -sign;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_max_epi32) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_max_epi32) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_max_epi32) (mask, src1.x, src2.x);
+
+ CALC (src1.a, src2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmaxsq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpmaxsq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmaxsq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_max_epi64 (x, x);
+ x = _mm512_mask_max_epi64 (x, m, x, x);
+ x = _mm512_maskz_max_epi64 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+
+CALC (long long *src1, long long *src2, long long *dst)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ dst[i] = src1[i] > src2[i] ? src1[i] : src2[i];
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_q) src1, src2, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = i * sign;
+ src2.a[i] = (i + 2000) * sign;
+ sign = -sign;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_max_epi64) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_max_epi64) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_max_epi64) (mask, src1.x, src2.x);
+
+ CALC (src1.a, src2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmaxud\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpmaxud\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmaxud\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_max_epu32 (x, x);
+ x = _mm512_mask_max_epu32 (x, m, x, x);
+ x = _mm512_maskz_max_epu32 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+
+CALC (unsigned *src1, unsigned *src2,
+ unsigned *dst)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ dst[i] = src1[i] > src2[i] ? src1[i] : src2[i];
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_d) src1, src2, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned res_ref[SIZE];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = i;
+ src2.a[i] = (i + 2000);
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_max_epu32) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_max_epu32) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_max_epu32) (mask, src1.x, src2.x);
+
+ CALC (src1.a, src2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmaxuq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpmaxuq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmaxuq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_max_epu64 (x, x);
+ x = _mm512_mask_max_epu64 (x, m, x, x);
+ x = _mm512_maskz_max_epu64 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+
+CALC (unsigned long long *src1, unsigned long long *src2,
+ unsigned long long *dst)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ dst[i] = src1[i] > src2[i] ? src1[i] : src2[i];
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_q) src1, src2, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned long long res_ref[SIZE];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = i;
+ src2.a[i] = (i + 2000);
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_max_epu64) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_max_epu64) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_max_epu64) (mask, src1.x, src2.x);
+
+ CALC (src1.a, src2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpminsd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpminsd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpminsd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_min_epi32 (x, x);
+ x = _mm512_mask_min_epi32 (x, m, x, x);
+ x = _mm512_maskz_min_epi32 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+
+CALC (int *src1, int *src2, int *dst)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ dst[i] = src1[i] < src2[i] ? src1[i] : src2[i];
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_d) src1, src2, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = i * sign;
+ src2.a[i] = (i + 2000) * sign;
+ sign = -sign;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_min_epi32) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_min_epi32) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_min_epi32) (mask, src1.x, src2.x);
+
+ CALC (src1.a, src2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpminsq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpminsq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpminsq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_min_epi64 (x, x);
+ x = _mm512_mask_min_epi64 (x, m, x, x);
+ x = _mm512_maskz_min_epi64 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+
+CALC (long long *src1, long long *src2, long long *dst)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ dst[i] = src1[i] < src2[i] ? src1[i] : src2[i];
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_q) src1, src2, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = i * sign;
+ src2.a[i] = (i + 2000) * sign;
+ sign = -sign;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_min_epi64) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_min_epi64) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_min_epi64) (mask, src1.x, src2.x);
+
+ CALC (src1.a, src2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpminud\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpminud\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpminud\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_min_epu32 (x, x);
+ x = _mm512_mask_min_epu32 (x, m, x, x);
+ x = _mm512_maskz_min_epu32 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+
+CALC (unsigned *src1, unsigned *src2,
+ unsigned *dst)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ dst[i] = src1[i] < src2[i] ? src1[i] : src2[i];
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_d) src1, src2, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned res_ref[SIZE];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = i * i;
+ src2.a[i] = i + 20;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_min_epu32) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_min_epu32) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_min_epu32) (mask, src1.x, src2.x);
+
+ CALC (src1.a, src2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpminuq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpminuq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpminuq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_min_epu64 (x, x);
+ x = _mm512_mask_min_epu64 (x, m, x, x);
+ x = _mm512_maskz_min_epu64 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+
+CALC (unsigned long long *src1, unsigned long long *src2,
+ unsigned long long *dst)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ dst[i] = src1[i] < src2[i] ? src1[i] : src2[i];
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_q) src1, src2, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned long long res_ref[SIZE];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = i;
+ src2.a[i] = (i + 2000);
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_min_epu64) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_min_epu64) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_min_epu64) (mask, src1.x, src2.x);
+
+ CALC (src1.a, src2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovdb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovdb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovdb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i s;
+volatile __m128i res;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtepi32_epi8 (s);
+ res = _mm512_mask_cvtepi32_epi8 (res, m, s);
+ res = _mm512_maskz_cvtepi32_epi8 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (char *r, int *s)
+{
+ int i;
+ for (i = 0; i < 16; i++)
+ {
+ r[i] = (i < SIZE) ? (char) s[i] : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (128, i_b) res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN, i_d) src;
+ MASK_TYPE mask = MASK_VALUE;
+ char res_ref[16];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src.a[i] = 1 + 34 * i * sign;
+ sign = sign * -1;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtepi32_epi8) (src.x);
+ res2.x = INTRINSIC (_mask_cvtepi32_epi8) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_cvtepi32_epi8) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (128, i_b) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_b) (res_ref, mask, SIZE);
+ if (UNION_CHECK (128, i_b) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_b) (res_ref, mask, SIZE);
+ if (UNION_CHECK (128, i_b) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovdw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovdw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovdw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i s;
+volatile __m256i res;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtepi32_epi16 (s);
+ res = _mm512_mask_cvtepi32_epi16 (res, m, s);
+ res = _mm512_maskz_cvtepi32_epi16 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#define SIZE_HALF (AVX512F_LEN_HALF / 16)
+
+CALC (short *r, int *s)
+{
+ int i;
+ for (i = 0; i < SIZE_HALF; i++)
+ {
+ r[i] = (i < SIZE) ? (short) s[i] : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN_HALF, i_w) res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN, i_d) src;
+ MASK_TYPE mask = MASK_VALUE;
+ short res_ref[SIZE_HALF];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src.a[i] = 1 + 34 * i * sign;
+ sign = sign * -1;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtepi32_epi16) (src.x);
+ res2.x = INTRINSIC (_mask_cvtepi32_epi16) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_cvtepi32_epi16) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_w) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_w) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_w) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_w) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_w) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovqb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovqb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovqb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i s;
+volatile __m128i res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtepi64_epi8 (s);
+ res = _mm512_mask_cvtepi64_epi8 (res, m, s);
+ res = _mm512_maskz_cvtepi64_epi8 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (char *r, long long *s)
+{
+ int i;
+ for (i = 0; i < 16; i++)
+ {
+ r[i] = (i < SIZE) ? (char) s[i] : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (128, i_b) res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN, i_q) src;
+ MASK_TYPE mask = MASK_VALUE;
+ char res_ref[16];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src.a[i] = 1 + 34 * i * sign;
+ sign = sign * -1;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtepi64_epi8) (src.x);
+ res2.x = INTRINSIC (_mask_cvtepi64_epi8) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_cvtepi64_epi8) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (128, i_b) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_b) (res_ref, mask, SIZE);
+ if (UNION_CHECK (128, i_b) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_b) (res_ref, mask, SIZE);
+ if (UNION_CHECK (128, i_b) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovqd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovqd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovqd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i s;
+volatile __m256i res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtepi64_epi32 (s);
+ res = _mm512_mask_cvtepi64_epi32 (res, m, s);
+ res = _mm512_maskz_cvtepi64_epi32 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+#define SIZE_HALF (AVX512F_LEN_HALF / 32)
+
+CALC (int *r, long long *s)
+{
+ int i;
+ for (i = 0; i < SIZE_HALF; i++)
+ {
+ r[i] = (i < SIZE) ? (int) s[i] : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN_HALF, i_d) res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN, i_q) src;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE_HALF];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src.a[i] = 1 + 34 * i * sign;
+ sign = sign * -1;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtepi64_epi32) (src.x);
+ res2.x = INTRINSIC (_mask_cvtepi64_epi32) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_cvtepi64_epi32) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovqw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovqw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovqw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i s;
+volatile __m128i res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtepi64_epi16 (s);
+ res = _mm512_mask_cvtepi64_epi16 (res, m, s);
+ res = _mm512_maskz_cvtepi64_epi16 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (short *r, long long *s)
+{
+ int i;
+ for (i = 0; i < 8; i++)
+ {
+ r[i] = (i < SIZE) ? (short) s[i] : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (128, i_w) res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN, i_q) src;
+ MASK_TYPE mask = MASK_VALUE;
+ short res_ref[8];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src.a[i] = 1 + 34 * i * sign;
+ sign = sign * -1;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtepi64_epi16) (src.x);
+ res2.x = INTRINSIC (_mask_cvtepi64_epi16) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_cvtepi64_epi16) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (128, i_w) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_w) (res_ref, mask, SIZE);
+ if (UNION_CHECK (128, i_w) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_w) (res_ref, mask, SIZE);
+ if (UNION_CHECK (128, i_w) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovsdb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovsdb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovsdb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i s;
+volatile __m128i res;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtsepi32_epi8 (s);
+ res = _mm512_mask_cvtsepi32_epi8 (res, m, s);
+ res = _mm512_maskz_cvtsepi32_epi8 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#include <limits.h>
+
+CALC (char *r, int *s)
+{
+ int i;
+ for (i = 0; i < 16; i++)
+ {
+ if (s[i] < CHAR_MIN)
+ r[i] = CHAR_MIN;
+ else if (s[i] > CHAR_MAX)
+ r[i] = CHAR_MAX;
+ else
+ r[i] = s[i];
+ r[i] = (i < SIZE) ? r[i] : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (128, i_b) res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN, i_d) src;
+ MASK_TYPE mask = MASK_VALUE;
+ char res_ref[16];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src.a[i] = 1 + 34 * i * sign;
+ sign = sign * -1;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtsepi32_epi8) (src.x);
+ res2.x = INTRINSIC (_mask_cvtsepi32_epi8) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_cvtsepi32_epi8) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (128, i_b) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_b) (res_ref, mask, SIZE);
+ if (UNION_CHECK (128, i_b) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_b) (res_ref, mask, SIZE);
+ if (UNION_CHECK (128, i_b) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovsdw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovsdw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovsdw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i s;
+volatile __m256i res;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtsepi32_epi16 (s);
+ res = _mm512_mask_cvtsepi32_epi16 (res, m, s);
+ res = _mm512_maskz_cvtsepi32_epi16 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#define SIZE_HALF (AVX512F_LEN_HALF / 16)
+#include <limits.h>
+
+CALC (short *r, int *s)
+{
+ int i;
+ for (i = 0; i < SIZE_HALF; i++)
+ {
+ if (s[i] < SHRT_MIN)
+ r[i] = SHRT_MIN;
+ else if (s[i] > SHRT_MAX)
+ r[i] = SHRT_MAX;
+ else
+ r[i] = s[i];
+ r[i] = (i < SIZE) ? r[i] : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN_HALF, i_w) res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN, i_d) src;
+ MASK_TYPE mask = MASK_VALUE;
+ short res_ref[SIZE_HALF];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src.a[i] = 1 + 34 * i * sign;
+ sign = sign * -1;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtsepi32_epi16) (src.x);
+ res2.x = INTRINSIC (_mask_cvtsepi32_epi16) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_cvtsepi32_epi16) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_w) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_w) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_w) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_w) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_w) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovsqb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovsqb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovsqb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i s;
+volatile __m128i res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtsepi64_epi8 (s);
+ res = _mm512_mask_cvtsepi64_epi8 (res, m, s);
+ res = _mm512_maskz_cvtsepi64_epi8 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+#include <limits.h>
+
+CALC (char *r, long long *s)
+{
+ int i;
+ for (i = 0; i < 16; i++)
+ {
+ if (s[i] < CHAR_MIN)
+ r[i] = CHAR_MIN;
+ else if (s[i] > CHAR_MAX)
+ r[i] = CHAR_MAX;
+ else
+ r[i] = s[i];
+ r[i] = (i < SIZE) ? r[i] : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (128, i_b) res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN, i_q) src;
+ MASK_TYPE mask = MASK_VALUE;
+ char res_ref[16];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src.a[i] = 1 + 34 * i * sign;
+ sign = sign * -1;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtsepi64_epi8) (src.x);
+ res2.x = INTRINSIC (_mask_cvtsepi64_epi8) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_cvtsepi64_epi8) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (128, i_b) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_b) (res_ref, mask, SIZE);
+ if (UNION_CHECK (128, i_b) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_b) (res_ref, mask, SIZE);
+ if (UNION_CHECK (128, i_b) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovsqd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovsqd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovsqd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i s;
+volatile __m256i res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtsepi64_epi32 (s);
+ res = _mm512_mask_cvtsepi64_epi32 (res, m, s);
+ res = _mm512_maskz_cvtsepi64_epi32 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+#define SIZE_HALF (AVX512F_LEN_HALF / 32)
+#include <limits.h>
+
+CALC (int *r, long long *s)
+{
+ int i;
+ for (i = 0; i < SIZE_HALF; i++)
+ {
+ if (s[i] < INT_MIN)
+ r[i] = INT_MIN;
+ else if (s[i] > INT_MAX)
+ r[i] = INT_MAX;
+ else
+ r[i] = s[i];
+ r[i] = (i < SIZE) ? r[i] : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN_HALF, i_d) res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN, i_q) src;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE_HALF];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src.a[i] = 1 + 34 * i * sign;
+ sign = sign * -1;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtsepi64_epi32) (src.x);
+ res2.x = INTRINSIC (_mask_cvtsepi64_epi32) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_cvtsepi64_epi32) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovsqw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovsqw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovsqw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i s;
+volatile __m128i res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtsepi64_epi16 (s);
+ res = _mm512_mask_cvtsepi64_epi16 (res, m, s);
+ res = _mm512_maskz_cvtsepi64_epi16 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+#include <limits.h>
+
+CALC (short *r, long long *s)
+{
+ int i;
+ for (i = 0; i < 8; i++)
+ {
+ if (s[i] < SHRT_MIN)
+ r[i] = SHRT_MIN;
+ else if (s[i] > SHRT_MAX)
+ r[i] = SHRT_MAX;
+ else
+ r[i] = s[i];
+ r[i] = (i < SIZE) ? r[i] : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (128, i_w) res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN, i_q) src;
+ MASK_TYPE mask = MASK_VALUE;
+ short res_ref[8];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src.a[i] = 1 + 34 * i * sign;
+ sign = sign * -1;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtsepi64_epi16) (src.x);
+ res2.x = INTRINSIC (_mask_cvtsepi64_epi16) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_cvtsepi64_epi16) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (128, i_w) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_w) (res_ref, mask, SIZE);
+ if (UNION_CHECK (128, i_w) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_w) (res_ref, mask, SIZE);
+ if (UNION_CHECK (128, i_w) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovsxbd\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovsxbd\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovsxbd\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128i s;
+volatile __m512i res;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtepi8_epi32 (s);
+ res = _mm512_mask_cvtepi8_epi32 (res, m, s);
+ res = _mm512_maskz_cvtepi8_epi32 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (char *s, int *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = (int) s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ union128i_b s;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 8 * i * sign;
+ res2.a[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+ res1.x = INTRINSIC (_cvtepi8_epi32) (s.x);
+ res2.x = INTRINSIC (_mask_cvtepi8_epi32) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_cvtepi8_epi32) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovsxbq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovsxbq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovsxbq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128i s;
+volatile __m512i res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtepi8_epi64 (s);
+ res = _mm512_mask_cvtepi8_epi64 (res, m, s);
+ res = _mm512_maskz_cvtepi8_epi64 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (char *s, long long int *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = (long long int) s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ union128i_b s;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ long long int res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 8 * i * sign;
+ res2.a[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+ res1.x = INTRINSIC (_cvtepi8_epi64) (s.x);
+ res2.x = INTRINSIC (_mask_cvtepi8_epi64) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_cvtepi8_epi64) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovsxdq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovsxdq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovsxdq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m256i s;
+volatile __m512i res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtepi32_epi64 (s);
+ res = _mm512_mask_cvtepi32_epi64 (res, m, s);
+ res = _mm512_maskz_cvtepi32_epi64 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (int *s, long long int *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = (long long int) s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN_HALF, i_d) s;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ long long int res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 2000 * i * sign;
+ res2.a[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+ res1.x = INTRINSIC (_cvtepi32_epi64) (s.x);
+ res2.x = INTRINSIC (_mask_cvtepi32_epi64) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_cvtepi32_epi64) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovsxwd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovsxwd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovsxwd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m256i s;
+volatile __m512i res;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtepi16_epi32 (s);
+ res = _mm512_mask_cvtepi16_epi32 (res, m, s);
+ res = _mm512_maskz_cvtepi16_epi32 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (short *s, int *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = (int) s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN_HALF, i_w) s;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 2000 * i * sign;
+ res2.a[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+ res1.x = INTRINSIC (_cvtepi16_epi32) (s.x);
+ res2.x = INTRINSIC (_mask_cvtepi16_epi32) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_cvtepi16_epi32) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovsxwq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovsxwq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovsxwq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128i s;
+volatile __m512i res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtepi16_epi64 (s);
+ res = _mm512_mask_cvtepi16_epi64 (res, m, s);
+ res = _mm512_maskz_cvtepi16_epi64 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (short *s, long long int *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = (long long int) s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ union128i_w s;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ long long int res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 2000 * i * sign;
+ res2.a[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+ res1.x = INTRINSIC (_cvtepi16_epi64) (s.x);
+ res2.x = INTRINSIC (_mask_cvtepi16_epi64) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_cvtepi16_epi64) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovusdb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovusdb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovusdb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i s;
+volatile __m128i res;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtusepi32_epi8 (s);
+ res = _mm512_mask_cvtusepi32_epi8 (res, m, s);
+ res = _mm512_maskz_cvtusepi32_epi8 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#include <limits.h>
+
+CALC (unsigned char *r, unsigned int *s)
+{
+ int i;
+ for (i = 0; i < 16; i++)
+ {
+ r[i] = (s[i] > UCHAR_MAX) ? UCHAR_MAX : s[i];
+ r[i] = (i < SIZE) ? r[i] : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (128, i_b) res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN, i_d) src;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned char res_ref[16];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src.a[i] = 1 + 34 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtusepi32_epi8) (src.x);
+ res2.x = INTRINSIC (_mask_cvtusepi32_epi8) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_cvtusepi32_epi8) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (128, i_b) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_b) (res_ref, mask, SIZE);
+ if (UNION_CHECK (128, i_b) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_b) (res_ref, mask, SIZE);
+ if (UNION_CHECK (128, i_b) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovusdw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovusdw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovusdw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i s;
+volatile __m256i res;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtusepi32_epi16 (s);
+ res = _mm512_mask_cvtusepi32_epi16 (res, m, s);
+ res = _mm512_maskz_cvtusepi32_epi16 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#define SIZE_HALF (AVX512F_LEN_HALF / 16)
+#include <limits.h>
+
+CALC (unsigned short *r, unsigned int *s)
+{
+ int i;
+ for (i = 0; i < SIZE_HALF; i++)
+ {
+ r[i] = (s[i] > USHRT_MAX) ? USHRT_MAX : s[i];
+ r[i] = (i < SIZE) ? r[i] : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN_HALF, i_w) res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN, i_d) src;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned short res_ref[SIZE_HALF];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src.a[i] = 1 + 34 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtusepi32_epi16) (src.x);
+ res2.x = INTRINSIC (_mask_cvtusepi32_epi16) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_cvtusepi32_epi16) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_w) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_w) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_w) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_w) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_w) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovusqb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovusqb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovusqb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i s;
+volatile __m128i res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtusepi64_epi8 (s);
+ res = _mm512_mask_cvtusepi64_epi8 (res, m, s);
+ res = _mm512_maskz_cvtusepi64_epi8 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+#include <limits.h>
+
+CALC (unsigned char *r, unsigned long long *s)
+{
+ int i;
+ for (i = 0; i < 16; i++)
+ {
+ r[i] = (s[i] > UCHAR_MAX) ? UCHAR_MAX : s[i];
+ r[i] = (i < SIZE) ? r[i] : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (128, i_b) res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN, i_q) src;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned char res_ref[16];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src.a[i] = 1 + 34 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtusepi64_epi8) (src.x);
+ res2.x = INTRINSIC (_mask_cvtusepi64_epi8) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_cvtusepi64_epi8) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (128, i_b) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_b) (res_ref, mask, SIZE);
+ if (UNION_CHECK (128, i_b) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_b) (res_ref, mask, SIZE);
+ if (UNION_CHECK (128, i_b) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovusqd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovusqd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovusqd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i s;
+volatile __m256i res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtusepi64_epi32 (s);
+ res = _mm512_mask_cvtusepi64_epi32 (res, m, s);
+ res = _mm512_maskz_cvtusepi64_epi32 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+#define SIZE_HALF (AVX512F_LEN_HALF / 32)
+#include <limits.h>
+
+CALC (unsigned int *r, unsigned long long *s)
+{
+ int i;
+ for (i = 0; i < SIZE_HALF; i++)
+ {
+ r[i] = (s[i] > UINT_MAX) ? UINT_MAX : s[i];
+ r[i] = (i < SIZE) ? r[i] : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN_HALF, i_d) res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN, i_q) src;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned int res_ref[SIZE_HALF];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src.a[i] = 1 + 34 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtusepi64_epi32) (src.x);
+ res2.x = INTRINSIC (_mask_cvtusepi64_epi32) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_cvtusepi64_epi32) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovusqw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovusqw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovusqw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i s;
+volatile __m128i res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtusepi64_epi16 (s);
+ res = _mm512_mask_cvtusepi64_epi16 (res, m, s);
+ res = _mm512_maskz_cvtusepi64_epi16 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+#include <limits.h>
+
+CALC (unsigned short *r, unsigned long long *s)
+{
+ int i;
+ for (i = 0; i < 8; i++)
+ {
+ r[i] = (s[i] > USHRT_MAX) ? USHRT_MAX : s[i];
+ r[i] = (i < SIZE) ? r[i] : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (128, i_w) res1, res2, res3;
+ UNION_TYPE (AVX512F_LEN, i_q) src;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned short res_ref[8];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src.a[i] = 1 + 34 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtusepi64_epi16) (src.x);
+ res2.x = INTRINSIC (_mask_cvtusepi64_epi16) (res2.x, mask, src.x);
+ res3.x = INTRINSIC (_maskz_cvtusepi64_epi16) (mask, src.x);
+
+ CALC (res_ref, src.a);
+
+ if (UNION_CHECK (128, i_w) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_w) (res_ref, mask, SIZE);
+ if (UNION_CHECK (128, i_w) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_w) (res_ref, mask, SIZE);
+ if (UNION_CHECK (128, i_w) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovzxbd\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovzxbd\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovzxbd\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128i s;
+volatile __m512i res;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtepu8_epi32 (s);
+ res = _mm512_mask_cvtepu8_epi32 (res, m, s);
+ res = _mm512_maskz_cvtepu8_epi32 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (unsigned char *s, int *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ union128i_b s;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 16 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtepu8_epi32) (s.x);
+ res2.x = INTRINSIC (_mask_cvtepu8_epi32) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_cvtepu8_epi32) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovzxbq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovzxbq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovzxbq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128i s;
+volatile __m512i res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtepu8_epi64 (s);
+ res = _mm512_mask_cvtepu8_epi64 (res, m, s);
+ res = _mm512_maskz_cvtepu8_epi64 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (unsigned char *s, long long int *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ union128i_b s;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ long long int res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 16 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtepu8_epi64) (s.x);
+ res2.x = INTRINSIC (_mask_cvtepu8_epi64) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_cvtepu8_epi64) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovzxdq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovzxdq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovzxdq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m256i s;
+volatile __m512i res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtepu32_epi64 (s);
+ res = _mm512_mask_cvtepu32_epi64 (res, m, s);
+ res = _mm512_maskz_cvtepu32_epi64 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (unsigned *s, long long int *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN_HALF, i_d) s;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ long long int res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 2000 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtepu32_epi64) (s.x);
+ res2.x = INTRINSIC (_mask_cvtepu32_epi64) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_cvtepu32_epi64) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovzxwd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovzxwd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovzxwd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m256i s;
+volatile __m512i res;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtepu16_epi32 (s);
+ res = _mm512_mask_cvtepu16_epi32 (res, m, s);
+ res = _mm512_maskz_cvtepu16_epi32 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (unsigned short *s, int *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN_HALF, i_w) s;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 2000 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtepu16_epi32) (s.x);
+ res2.x = INTRINSIC (_mask_cvtepu16_epi32) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_cvtepu16_epi32) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmovzxwq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovzxwq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmovzxwq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128i s;
+volatile __m512i res;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm512_cvtepu16_epi64 (s);
+ res = _mm512_mask_cvtepu16_epi64 (res, m, s);
+ res = _mm512_maskz_cvtepu16_epi64 (m, s);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (unsigned short *s, long long int *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = (long long int) s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ union128i_w s;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ long long int res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 2000 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_cvtepu16_epi64) (s.x);
+ res2.x = INTRINSIC (_mask_cvtepu16_epi64) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_cvtepu16_epi64) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmuldq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpmuldq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmuldq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}{z}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mul_epi32 (x, x);
+ x = _mm512_mask_mul_epi32 (x, m, x, x);
+ x = _mm512_maskz_mul_epi32 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SRC_SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#define DST_SIZE (AVX512F_LEN / 64)
+
+static void
+CALC (int *s1, int *s2, long long int *r)
+{
+ int i;
+
+ for (i = 0; i < DST_SIZE; i++)
+ r[i] = s1[i * 2] * s2[i * 2];
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) s1, s2;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[DST_SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SRC_SIZE; i++)
+ {
+ s1.a[i] = i * 20;
+ s2.a[i] = i + 20;
+ }
+
+ for (i = 0; i < DST_SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ CALC (s1.a, s2.a, res_ref);
+
+ res1.x = INTRINSIC (_mul_epi32) (s1.x, s2.x);
+ res2.x = INTRINSIC (_mask_mul_epi32) (res2.x, mask, s1.x, s2.x);
+ res3.x = INTRINSIC (_maskz_mul_epi32) (mask, s1.x, s2.x);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, DST_SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, DST_SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmulld\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpmulld\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmulld\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}{z}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 mx;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mullo_epi32 (x, x);
+ x = _mm512_mask_mullo_epi32 (x, mx, x, x);
+ x = _mm512_maskz_mullo_epi32 (mx, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+
+static void
+CALC (int *src1, int *src2, int *dst)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ dst[i] = src1[i] * src2[i];
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) src1, src2, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ int dst_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = i + 50;
+ src2.a[i] = i + 100;
+ }
+
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_mullo_epi32) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_mullo_epi32) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_mullo_epi32) (mask, src1.x, src2.x);
+
+ CALC (src1.a, src2.a, dst_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, dst_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (dst_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, dst_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (dst_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, dst_ref))
+ abort ();
+
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpmuludq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpmuludq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpmuludq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}{z}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_mul_epu32 (x, x);
+ x = _mm512_mask_mul_epu32 (x, m, x, x);
+ x = _mm512_maskz_mul_epu32 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SRC_SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#define DST_SIZE (AVX512F_LEN / 64)
+
+static void
+CALC (unsigned int *s1, unsigned int *s2, unsigned long long *r)
+{
+ int i;
+
+ for (i = 0; i < DST_SIZE; i++)
+ r[i] = s1[i * 2] * s2[i * 2];
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) s1, s2;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned long long res_ref[DST_SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SRC_SIZE; i++)
+ {
+ s1.a[i] = i * 20;
+ s2.a[i] = i + 20;
+ }
+ for (i = 0; i < DST_SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ CALC (s1.a, s2.a, res_ref);
+
+ res1.x = INTRINSIC (_mul_epu32) (s1.x, s2.x);
+ res2.x = INTRINSIC (_mask_mul_epu32) (res2.x, mask, s1.x, s2.x);
+ res3.x = INTRINSIC (_maskz_mul_epu32) (mask, s1.x, s2.x);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, DST_SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, DST_SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpord\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 4 } } */
+/* { dg-final { scan-assembler-times "vpord\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpord\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_or_si512 (x, x);
+ x = _mm512_or_epi32 (x, x);
+ x = _mm512_mask_or_epi32 (x, m, x, x);
+ x = _mm512_maskz_or_epi32 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (int *s1, int *s2, int *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ r[i] = s1[i] | s2[i];
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) s1, s2, res1, res2, res3, res4;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = i * sign;
+ s2.a[i] = (i + 20) * sign;
+ sign = -sign;
+ res3.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_or_si512) (s1.x, s2.x);
+ res2.x = INTRINSIC (_or_epi32) (s1.x, s2.x);
+ res3.x = INTRINSIC (_mask_or_epi32) (res3.x, mask, s1.x, s2.x);
+ res4.x = INTRINSIC (_maskz_or_epi32) (mask, s1.x, s2.x);
+
+ CALC (s1.a, s2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res4, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vporq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vporq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vporq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_or_epi64 (x, x);
+ x = _mm512_mask_or_epi64 (x, m, x, x);
+ x = _mm512_maskz_or_epi64 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (long long *s1, long long *s2, long long *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ r[i] = s1[i] | s2[i];
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_q) s1, s2, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = i * sign;
+ s2.a[i] = (i + 20) * sign;
+ sign = -sign;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_or_epi64) (s1.x, s2.x);
+ res2.x = INTRINSIC (_mask_or_epi64) (res2.x, mask, s1.x, s2.x);
+ res3.x = INTRINSIC (_maskz_or_epi64) (mask, s1.x, s2.x);
+
+ CALC (s1.a, s2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vprold\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vprold\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vprold\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_rol_epi32 (x, 12);
+ x = _mm512_mask_rol_epi32 (x, m, x, 12);
+ x = _mm512_maskz_rol_epi32 (m, x, 12);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+#define N 0x5
+
+static void
+CALC (int *s1, int count, int *r)
+{
+ unsigned int i;
+
+ for (i = 0; i < SIZE; i++)
+ r[i] = (s1[i] << count) | (s1[i] >> sizeof (s1[i]) * 8 - count);
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) s1, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+ unsigned int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 137 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_rol_epi32) (s1.x, N);
+ res2.x = INTRINSIC (_mask_rol_epi32) (res2.x, mask, s1.x, N);
+ res3.x = INTRINSIC (_maskz_rol_epi32) (mask, s1.x, N);
+
+ CALC (s1.a, N, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vprolq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vprolq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vprolq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_rol_epi64 (x, 12);
+ x = _mm512_mask_rol_epi64 (x, m, x, 12);
+ x = _mm512_maskz_rol_epi64 (m, x, 12);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+#define N 0x5
+
+static void
+CALC (long long *s1, int count, long long *r)
+{
+ unsigned int i;
+
+ for (i = 0; i < SIZE; i++)
+ r[i] = (s1[i] << count) | (s1[i] >> sizeof (s1[i]) * 8 - count);
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_q) s1, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+ unsigned int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 137 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_rol_epi64) (s1.x, N);
+ res2.x = INTRINSIC (_mask_rol_epi64) (res2.x, mask, s1.x, N);
+ res3.x = INTRINSIC (_maskz_rol_epi64) (mask, s1.x, N);
+
+ CALC (s1.a, N, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vprolvd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vprolvd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vprolvd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_rolv_epi32 (x, x);
+ x = _mm512_mask_rolv_epi32 (x, m, x, x);
+ x = _mm512_maskz_rolv_epi32 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (int *s1, int *s2, int *r)
+{
+ unsigned int i;
+
+ for (i = 0; i < SIZE; i++)
+ r[i] = (s1[i] << s2[i]) | (s1[i] >> sizeof (s1[i]) * 8 - s2[i]);
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) s1, s2, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+ unsigned int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 137 * i;
+ s2.a[i] = (i + 7);
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_rolv_epi32) (s1.x, s2.x);
+ res2.x = INTRINSIC (_mask_rolv_epi32) (res2.x, mask, s1.x, s2.x);
+ res3.x = INTRINSIC (_maskz_rolv_epi32) (mask, s1.x, s2.x);
+
+ CALC (s1.a, s2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vprolvq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vprolvq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vprolvq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_rolv_epi64 (x, x);
+ x = _mm512_mask_rolv_epi64 (x, m, x, x);
+ x = _mm512_maskz_rolv_epi64 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (long long *s1, long long *s2, long long *r)
+{
+ unsigned int i;
+
+ for (i = 0; i < SIZE; i++)
+ r[i] = (s1[i] << s2[i]) | (s1[i] >> sizeof (s1[i]) * 8 - s2[i]);
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_q) s1, s2, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+ unsigned int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 137 * i;
+ s2.a[i] = (i + 7);
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_rolv_epi64) (s1.x, s2.x);
+ res2.x = INTRINSIC (_mask_rolv_epi64) (res2.x, mask, s1.x, s2.x);
+ res3.x = INTRINSIC (_maskz_rolv_epi64) (mask, s1.x, s2.x);
+
+ CALC (s1.a, s2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vprord\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vprord\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vprord\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_ror_epi32 (x, 12);
+ x = _mm512_mask_ror_epi32 (x, m, x, 12);
+ x = _mm512_maskz_ror_epi32 (m, x, 12);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+#define N 0x5
+
+static void
+CALC (int *s1, int count, int *r)
+{
+ unsigned int i;
+
+ for (i = 0; i < SIZE; i++)
+ r[i] = (s1[i] >> count) | (s1[i] << sizeof (s1[i]) * 8 - count);
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) s1, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+ unsigned int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 137 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_ror_epi32) (s1.x, N);
+ res2.x = INTRINSIC (_mask_ror_epi32) (res2.x, mask, s1.x, N);
+ res3.x = INTRINSIC (_maskz_ror_epi32) (mask, s1.x, N);
+
+ CALC (s1.a, N, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vprorq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vprorq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vprorq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_ror_epi64 (x, 12);
+ x = _mm512_mask_ror_epi64 (x, m, x, 12);
+ x = _mm512_maskz_ror_epi64 (m, x, 12);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+#define N 0x5
+
+static void
+CALC (long long *s1, int count, long long *r)
+{
+ unsigned int i;
+
+ for (i = 0; i < SIZE; i++)
+ r[i] = (s1[i] >> count) | (s1[i] << sizeof (s1[i]) * 8 - count);
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_q) s1, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+ unsigned int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 137 * i;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_ror_epi64) (s1.x, N);
+ res2.x = INTRINSIC (_mask_ror_epi64) (res2.x, mask, s1.x, N);
+ res3.x = INTRINSIC (_maskz_ror_epi64) (mask, s1.x, N);
+
+ CALC (s1.a, N, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler "vprorvd\[ \\t\]+\[^\n\]*%zmm\[0-9\]" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_rorv_epi32 (x, x);
+ x = _mm512_mask_rorv_epi32 (x, m, x, x);
+ x = _mm512_maskz_rorv_epi32 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (int *s1, int *s2, int *r)
+{
+ unsigned int i;
+
+ for (i = 0; i < SIZE; i++)
+ r[i] = (s1[i] >> s2[i]) | (s1[i] << sizeof (s1[i]) * 8 - s2[i]);
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) s1, s2, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+ unsigned int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 137 * i;
+ s2.a[i] = (i + 7);
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_rorv_epi32) (s1.x, s2.x);
+ res2.x = INTRINSIC (_mask_rorv_epi32) (res2.x, mask, s1.x, s2.x);
+ res3.x = INTRINSIC (_maskz_rorv_epi32) (mask, s1.x, s2.x);
+
+ CALC (s1.a, s2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vprorvq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vprorvq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vprorvq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_rorv_epi64 (x, x);
+ x = _mm512_mask_rorv_epi64 (x, m, x, x);
+ x = _mm512_maskz_rorv_epi64 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (long long *s1, long long *s2, long long *r)
+{
+ unsigned int i;
+
+ for (i = 0; i < SIZE; i++)
+ r[i] = (s1[i] >> s2[i]) | (s1[i] << sizeof (s1[i]) * 8 - s2[i]);
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_q) s1, s2, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+ unsigned int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 137 * i;
+ s2.a[i] = (i + 7);
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_rorv_epi64) (s1.x, s2.x);
+ res2.x = INTRINSIC (_mask_rorv_epi64) (res2.x, mask, s1.x, s2.x);
+ res3.x = INTRINSIC (_maskz_rorv_epi64) (mask, s1.x, s2.x);
+
+ CALC (s1.a, s2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpshufd\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpshufd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vpshufd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_shuffle_epi32 (x, _MM_PERM_AADB);
+ x = _mm512_mask_shuffle_epi32 (x, 2, x, _MM_PERM_AADB);
+ x = _mm512_maskz_shuffle_epi32 (2, x, _MM_PERM_AADB);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (int *s1, unsigned char imm, int *r)
+{
+ int i, j, offset;
+
+ for (j = 0; j < SIZE / 4; j++)
+ {
+ offset = j * 4;
+ for (i = 0; i < 4; i++)
+ r[i + offset] =
+ s1[((imm & (0x3 << (2 * i))) >> (2 * i)) + offset];
+ }
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) s1, res1, res2, res3;
+ int res_ref[SIZE];
+ int i, j, sign = 1;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (j = 0; j < SIZE; j++)
+ {
+ s1.a[j] = j * i * sign;
+ res1.a[j] = DEFAULT_VALUE;
+ res2.a[j] = DEFAULT_VALUE;
+ res3.a[j] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+ res1.x = INTRINSIC (_shuffle_epi32) (s1.x, 0xec);
+ res2.x = INTRINSIC (_mask_shuffle_epi32) (res2.x, mask, s1.x, 0xec);
+ res3.x = INTRINSIC (_maskz_shuffle_epi32) (mask, s1.x, 0xec);
+
+ CALC (s1.a, 0xec, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpslld\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpslld\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpslld\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __m128i y;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_sll_epi32 (x, y);
+ x = _mm512_mask_sll_epi32 (x, m, x, y);
+ x = _mm512_maskz_sll_epi32 (m, x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (int *r, int *s1, int* s2)
+{
+ int i;
+ int count = s2[0];
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = count < 32 ? (s1[i] << count) : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3, src1;
+ UNION_TYPE (128, i_d) src2;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+
+ long long imm;
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 2 + sign * 7 * i % 291;
+ sign = sign * -1;
+ }
+
+ for (imm = 1; imm <= 33; imm++)
+ {
+ src2.a[0] = imm;
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_sll_epi32) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_sll_epi32) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_sll_epi32) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+ }
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpslld\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpslld\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpslld\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+#define y 7
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_slli_epi32 (x, y);
+ x = _mm512_mask_slli_epi32 (x, m, x, y);
+ x = _mm512_maskz_slli_epi32 (m, x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (int *r, int *s1, int count)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = count < 32 ? (s1[i] << count) : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3, src1;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 2 + sign * 7 * i % 291;
+ sign = sign * -1;
+ }
+
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_slli_epi32) (src1.x, 2);
+ res2.x = INTRINSIC (_mask_slli_epi32) (res2.x, mask, src1.x, 2);
+ res3.x = INTRINSIC (_maskz_slli_epi32) (mask, src1.x, 2);
+
+ CALC (res_ref, src1.a, 2);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+
+
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_slli_epi32) (src1.x, 33);
+ res2.x = INTRINSIC (_mask_slli_epi32) (res2.x, mask, src1.x, 33);
+ res3.x = INTRINSIC (_maskz_slli_epi32) (mask, src1.x, 33);
+
+ CALC (res_ref, src1.a, 33);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpsllq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpsllq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpsllq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __m128i y;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_sll_epi64 (x, y);
+ x = _mm512_mask_sll_epi64 (x, m, x, y);
+ x = _mm512_maskz_sll_epi64 (m, x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (long long *r, long long *s1, long long* s2)
+{
+ int i;
+ long long count = s2[0];
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = count < 64 ? (s1[i] << count) : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3, src1;
+ UNION_TYPE (128, i_q) src2;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+
+ long long imm;
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 2 + sign * 7 * i % 291;
+ sign = sign * -1;
+ }
+
+ for (imm = 1; imm <= 65; imm++)
+ {
+ src2.a[0] = imm;
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_sll_epi64) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_sll_epi64) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_sll_epi64) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+ }
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpsllq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpsllq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpsllq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+#define y 7
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_slli_epi64 (x, y);
+ x = _mm512_mask_slli_epi64 (x, m, x, y);
+ x = _mm512_maskz_slli_epi64 (m, x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (long long *r, long long *s1, long long count)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = count < 64 ? (s1[i] << count) : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3, src1;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 2 + sign * 7 * i % 291;
+ sign = sign * -1;
+ }
+
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_slli_epi64) (src1.x, 3);
+ res2.x = INTRINSIC (_mask_slli_epi64) (res2.x, mask, src1.x, 3);
+ res3.x = INTRINSIC (_maskz_slli_epi64) (mask, src1.x, 3);
+
+ CALC (res_ref, src1.a, 3);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+
+
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_slli_epi64) (src1.x, 65);
+ res2.x = INTRINSIC (_mask_slli_epi64) (res2.x, mask, src1.x, 65);
+ res3.x = INTRINSIC (_maskz_slli_epi64) (mask, src1.x, 65);
+
+ CALC (res_ref, src1.a, 65);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpsllvd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpsllvd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpsllvd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __m512i y;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_sllv_epi32 (x, y);
+ x = _mm512_mask_sllv_epi32 (x, m, x, y);
+ x = _mm512_maskz_sllv_epi32 (m, x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (unsigned int *r, unsigned int *s1, unsigned int *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s2[i] < 32 ? (s1[i] << s2[i]) : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned int res_ref[SIZE];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 2 + 7 * i % 291;
+ src2.a[i] = 1 + 17 * i % 71;
+ }
+
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_sllv_epi32) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_sllv_epi32) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_sllv_epi32) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpsllvq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpsllvq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpsllvq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __m512i y;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_sllv_epi64 (x, y);
+ x = _mm512_mask_sllv_epi64 (x, m, x, y);
+ x = _mm512_maskz_sllv_epi64 (m, x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (unsigned long long *r, unsigned long long *s1,
+ unsigned long long *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s2[i] < 64 ? (s1[i] << s2[i]) : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned long long res_ref[SIZE];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 2 + 7 * i % 291;
+ src2.a[i] = 1 + 17 * i % 71;
+ }
+
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_sllv_epi64) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_sllv_epi64) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_sllv_epi64) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler "vpsllvq\[ \\t\]+\[^\n\]*%zmm\[0-9\]" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_sllv_epi64 (x, x);
+}
\ No newline at end of file
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include <string.h>
+#include "avx512f-check.h"
+
+static void
+compute_psllvq512 (long long int *s1, long long int *s2, long long int *r)
+{
+ int i;
+ long long int count;
+
+ for (i = 0; i < 8; ++i)
+ {
+ count = s2[i];
+ r[i] = s1[i] << count;
+ }
+}
+
+void static
+avx512f_test (void)
+{
+ union512i_q s1, s2, res;
+ long long int res_ref[8];
+ int i, j, sign = 1;
+ int fail = 0;
+
+ for (i = 0; i < 10; i++)
+ {
+ for (j = 0; j < 8; j++)
+ {
+ s1.a[j] = j * i * sign;
+ s2.a[j] = (j + i) >> 2;
+ sign = -sign;
+ }
+
+ res.x = _mm512_sllv_epi64 (s1.x, s2.x);
+
+ compute_psllvq512 (s1.a, s2.a, res_ref);
+
+ fail += check_union512i_q (res, res_ref);
+ }
+
+ if (fail != 0)
+ abort ();
+}
\ No newline at end of file
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpsrad\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpsrad\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpsrad\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __m128i y;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_sra_epi32 (x, y);
+ x = _mm512_mask_sra_epi32 (x, m, x, y);
+ x = _mm512_maskz_sra_epi32 (m, x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (int *r, int *s1, int *s2)
+{
+ int i;
+ int count = s2[0];
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] =
+ count < 32 ? (s1[i] >> count) : (s1[i] > 0 ? 0 : 0xFFFFFFFF);
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3, src1;
+ UNION_TYPE (128, i_d) src2;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+
+ long long imm;
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 2 + sign * 7 * i % 291;
+ sign = sign * -1;
+ }
+
+ for (imm = 1; imm <= 33; imm++)
+ {
+ src2.a[0] = imm;
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_sra_epi32) (src1.x, src2.x);
+ res2.x =
+ INTRINSIC (_mask_sra_epi32) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_sra_epi32) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+ }
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpsrad\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpsrad\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpsrad\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+#define y 7
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_srai_epi32 (x, y);
+ x = _mm512_mask_srai_epi32 (x, m, x, y);
+ x = _mm512_maskz_srai_epi32 (m, x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (int *r, int *s1, int count)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] =
+ count < 32 ? (s1[i] >> count) : (s1[i] > 0 ? 0 : 0xFFFFFFFF);
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3, src1;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 2 + sign * 7 * i % 291;
+ sign = sign * -1;
+ }
+
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_srai_epi32) (src1.x, 3);
+ res2.x =
+ INTRINSIC (_mask_srai_epi32) (res2.x, mask, src1.x, 3);
+ res3.x = INTRINSIC (_maskz_srai_epi32) (mask, src1.x, 3);
+
+ CALC (res_ref, src1.a, 3);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_srai_epi32) (src1.x, 33);
+ res2.x =
+ INTRINSIC (_mask_srai_epi32) (res2.x, mask, src1.x, 33);
+ res3.x = INTRINSIC (_maskz_srai_epi32) (mask, src1.x, 33);
+
+ CALC (res_ref, src1.a, 33);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpsraq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpsraq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpsraq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __m128i y;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_sra_epi64 (x, y);
+ x = _mm512_mask_sra_epi64 (x, m, x, y);
+ x = _mm512_maskz_sra_epi64 (m, x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (long long *r, long long *s1, long long *s2)
+{
+ int i;
+ long long count = s2[0];
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] =
+ count < 64 ? (s1[i] >> count) : (s1[i] >
+ 0 ? 0 : 0xFFFFFFFFFFFFFFFFLL);
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3, src1;
+ UNION_TYPE (128, i_q) src2;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+
+ long long imm;
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 2 + sign * 7 * i % 291;
+ sign = sign * -1;
+ }
+
+ for (imm = 1; imm <= 65; imm++)
+ {
+ src2.a[0] = imm;
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_sra_epi64) (src1.x, src2.x);
+ res2.x =
+ INTRINSIC (_mask_sra_epi64) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_sra_epi64) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+ }
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpsraq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpsraq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpsraq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+#define y 7
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_srai_epi64 (x, y);
+ x = _mm512_mask_srai_epi64 (x, m, x, y);
+ x = _mm512_maskz_srai_epi64 (m, x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (long long *r, long long *s1, long long count)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] =
+ count < 64 ? (s1[i] >> count) : (s1[i] >
+ 0 ? 0 : 0xFFFFFFFFFFFFFFFFLL);
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3, src1;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 2 + sign * 7 * i % 291;
+ sign = sign * -1;
+ }
+
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_srai_epi64) (src1.x, 3);
+ res2.x =
+ INTRINSIC (_mask_srai_epi64) (res2.x, mask, src1.x, 3);
+ res3.x = INTRINSIC (_maskz_srai_epi64) (mask, src1.x, 3);
+
+ CALC (res_ref, src1.a, 3);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+
+
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_srai_epi64) (src1.x, 65);
+ res2.x =
+ INTRINSIC (_mask_srai_epi64) (res2.x, mask, src1.x, 65);
+ res3.x = INTRINSIC (_maskz_srai_epi64) (mask, src1.x, 65);
+
+ CALC (res_ref, src1.a, 65);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpsravd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpsravd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpsravd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x, y;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_srav_epi32 (x, y);
+ x = _mm512_mask_srav_epi32 (x, m, x, y);
+ x = _mm512_maskz_srav_epi32 (m, x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (int *r, int *s1, int *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] =
+ s2[i] < 32 ? (s1[i] >> s2[i]) : (s1[i] > 0 ? 0 : 0xFFFFFFFF);
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 2 + sign * 7 * i % 291;
+ src2.a[i] = 1 + 17 * i % 71;
+ sign = sign * -1;
+ }
+
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_srav_epi32) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_srav_epi32) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_srav_epi32) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpsravq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpsravq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpsravq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x, y;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_srav_epi64 (x, y);
+ x = _mm512_mask_srav_epi64 (x, m, x, y);
+ x = _mm512_maskz_srav_epi64 (m, x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (long long *r, long long *s1, long long *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] =
+ s2[i] < 64 ? (s1[i] >> s2[i]) : (s1[i] >
+ 0 ? 0 : 0xFFFFFFFFFFFFFFFFLL);
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 2 + sign * 7 * i % 291;
+ src2.a[i] = 1 + 17 * i % 71;
+ sign = sign * -1;
+ }
+
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_srav_epi64) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_srav_epi64) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_srav_epi64) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler "vpsravq\[ \\t\]+\[^\n\]*%zmm\[0-9\]" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_srav_epi64 (x, x);
+}
\ No newline at end of file
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include <string.h>
+#include "avx512f-check.h"
+
+static void
+compute_psravq512 (long long int *s1, long long int *s2, long long int *r)
+{
+ int i;
+ long long int count;
+
+ for (i = 0; i < 8; ++i)
+ {
+ count = s2[i];
+ r[i] = s1[i] >> count;
+ }
+}
+
+void static
+avx512f_test (void)
+{
+ union512i_q s1, s2, res;
+ long long int res_ref[8];
+ int i, j, sign = 1;
+ int fail = 0;
+
+ for (i = 0; i < 10; i++)
+ {
+ for (j = 0; j < 8; j++)
+ {
+ s1.a[j] = j * i * sign;
+ s2.a[j] = (j + i) >> 2;
+ sign = -sign;
+ }
+
+ res.x = _mm512_srav_epi64 (s1.x, s2.x);
+
+ compute_psravq512 (s1.a, s2.a, res_ref);
+
+ fail += check_union512i_q (res, res_ref);
+ }
+
+ if (fail != 0)
+ abort ();
+}
\ No newline at end of file
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpsrld\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpsrld\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpsrld\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __m128i y;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_srl_epi32 (x, y);
+ x = _mm512_mask_srl_epi32 (x, m, x, y);
+ x = _mm512_maskz_srl_epi32 (m, x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (unsigned int *r, unsigned int *s1, unsigned int* s2)
+{
+ int i;
+ unsigned int count = s2[0];
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = count < 32 ? (s1[i] >> count) : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3, src1;
+ UNION_TYPE (128, i_d) src2;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned int res_ref[SIZE];
+
+ unsigned long long imm;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 2 + 7 * i % 291;
+ }
+
+ for (imm = 1; imm <= 33; imm++)
+ {
+ src2.a[0] = imm;
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_srl_epi32) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_srl_epi32) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_srl_epi32) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+ }
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpsrld\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpsrld\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpsrld\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+#define y 7
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_srli_epi32 (x, y);
+ x = _mm512_mask_srli_epi32 (x, m, x, y);
+ x = _mm512_maskz_srli_epi32 (m, x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (unsigned int *r, unsigned int *s1, unsigned int count)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = count < 32 ? (s1[i] >> count) : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3, src1;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned int res_ref[SIZE];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 2 + 7 * i % 291;
+ }
+
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_srli_epi32) (src1.x, 3);
+ res2.x =
+ INTRINSIC (_mask_srli_epi32) (res2.x, mask, src1.x, 3);
+ res3.x = INTRINSIC (_maskz_srli_epi32) (mask, src1.x, 3);
+
+ CALC (res_ref, src1.a, 3);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+
+
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_srli_epi32) (src1.x, 33);
+ res2.x =
+ INTRINSIC (_mask_srli_epi32) (res2.x, mask, src1.x, 33);
+ res3.x = INTRINSIC (_maskz_srli_epi32) (mask, src1.x, 33);
+
+ CALC (res_ref, src1.a, 33);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpsrlq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpsrlq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpsrlq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __m128i y;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_srl_epi64 (x, y);
+ x = _mm512_mask_srl_epi64 (x, m, x, y);
+ x = _mm512_maskz_srl_epi64 (m, x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (unsigned long long *r, unsigned long long *s1, unsigned long long* s2)
+{
+ int i;
+ unsigned long long count = s2[0];
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = count < 64 ? (s1[i] >> count) : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3, src1;
+ UNION_TYPE (128, i_q) src2;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned long long res_ref[SIZE];
+
+ unsigned long long imm;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 2 + 7 * i % 291;
+ }
+
+ for (imm = 1; imm <= 65; imm++)
+ {
+ src2.a[0] = imm;
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_srl_epi64) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_srl_epi64) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_srl_epi64) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+ }
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpsrlq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpsrlq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpsrlq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+#define y 7
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_srli_epi64 (x, y);
+ x = _mm512_mask_srli_epi64 (x, m, x, y);
+ x = _mm512_maskz_srli_epi64 (m, x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (unsigned long long *r, unsigned long long *s1,
+ unsigned long long count)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = count < 64 ? (s1[i] >> count) : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3, src1;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned long long res_ref[SIZE];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 2 + 7 * i % 291;
+ }
+
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_srli_epi64) (src1.x, 3);
+ res2.x =
+ INTRINSIC (_mask_srli_epi64) (res2.x, mask, src1.x, 3);
+ res3.x = INTRINSIC (_maskz_srli_epi64) (mask, src1.x, 3);
+
+ CALC (res_ref, src1.a, 3);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+
+
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_srli_epi64) (src1.x, 65);
+ res2.x =
+ INTRINSIC (_mask_srli_epi64) (res2.x, mask, src1.x, 65);
+ res3.x = INTRINSIC (_maskz_srli_epi64) (mask, src1.x, 65);
+
+ CALC (res_ref, src1.a, 65);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpsrlvd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpsrlvd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpsrlvd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __m512i y;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_srlv_epi32 (x, y);
+ x = _mm512_mask_srlv_epi32 (x, m, x, y);
+ x = _mm512_maskz_srlv_epi32 (m, x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (unsigned int *r, unsigned int *s1, unsigned int *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s2[i] < 32 ? (s1[i] >> s2[i]) : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned int res_ref[SIZE];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 2 + 7 * i % 291;
+ src2.a[i] = 1 + 17 * i % 71;
+ }
+
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_srlv_epi32) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_srlv_epi32) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_srlv_epi32) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpsrlvq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpsrlvq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpsrlvq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __m512i y;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_srlv_epi64 (x, y);
+ x = _mm512_mask_srlv_epi64 (x, m, x, y);
+ x = _mm512_maskz_srlv_epi64 (m, x, y);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (unsigned long long *r, unsigned long long *s1,
+ unsigned long long *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s2[i] < 64 ? (s1[i] >> s2[i]) : 0;
+ }
+}
+
+void static
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned long long res_ref[SIZE];
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 2 + 7 * i % 291;
+ src2.a[i] = 1 + 17 * i % 71;
+ }
+
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_srlv_epi64) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_srlv_epi64) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_srlv_epi64) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler "vpsrlvq\[ \\t\]+\[^\n\]*%zmm\[0-9\]" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_srlv_epi64 (x, x);
+}
\ No newline at end of file
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#include <string.h>
+#include "avx512f-check.h"
+
+static void
+compute_psrlvq512 (long long int *s1, long long int *s2, long long int *r)
+{
+ int i;
+ long long int count;
+
+ for (i = 0; i < 8; ++i)
+ {
+ count = s2[i];
+ r[i] = ((unsigned long long) s1[i]) >> count;
+ }
+}
+
+void static
+avx512f_test (void)
+{
+ union512i_q s1, s2, res;
+ long long int res_ref[8];
+ int i, j, sign = 1;
+ int fail = 0;
+
+ for (i = 0; i < 10; i++)
+ {
+ for (j = 0; j < 8; j++)
+ {
+ s1.a[j] = j * i * sign;
+ s2.a[j] = (j + i) >> 2;
+ sign = -sign;
+ }
+
+ res.x = _mm512_srlv_epi64 (s1.x, s2.x);
+
+ compute_psrlvq512 (s1.a, s2.a, res_ref);
+
+ fail += check_union512i_q (res, res_ref);
+ }
+
+ if (fail != 0)
+ abort ();
+}
\ No newline at end of file
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpsubd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpsubd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpsubd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_sub_epi32 (x, x);
+ x = _mm512_mask_sub_epi32 (x, m, x, x);
+ x = _mm512_maskz_sub_epi32 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (int *r, int *s1, int *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[i] - s2[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 1.5 + 34.67 * i * sign;
+ src2.a[i] = -22.17 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_sub_epi32) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_sub_epi32) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_sub_epi32) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vpsubq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpsubq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpsubq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_sub_epi64 (x, x);
+ x = _mm512_mask_sub_epi64 (x, m, x, x);
+ x = _mm512_maskz_sub_epi64 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (long long *r, long long *s1, long long *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[i] - s2[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 1.5 + 34.67 * i * sign;
+ src2.a[i] = -22.17 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_sub_epi64) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_sub_epi64) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_sub_epi64) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpternlogd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpternlogd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpternlogd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x, y, z;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_ternarylogic_epi32 (x, y, z, 0xF0);
+ x = _mm512_mask_ternarylogic_epi32 (x, m, y, z, 0xF0);
+ x = _mm512_maskz_ternarylogic_epi32 (m, x, y, z, 0xF0);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (int *src1, int *src2, int *src3, int imm, int *r)
+{
+ int i, j, index, res, mask, one_mask = 1;
+ int src1_bit, src2_bit, src3_bit, imm_bit;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ res = 0;
+ for (j = 0; j < 32; j++)
+ {
+ mask = one_mask << j;
+ src1_bit = ((src1[i] & mask) >> j) << 2;
+ src2_bit = ((src2[i] & mask) >> j) << 1;
+ src3_bit = ((src3[i] & mask) >> j);
+ index = src1_bit | src2_bit | src3_bit;
+ imm_bit = (imm & (one_mask << index)) >> index;
+ res = res | (imm_bit << j);
+ }
+ r[i] = res;
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) src2, src3, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+ int i, imm = 0x7D;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ res1.a[i] = DEFAULT_VALUE;
+ res2.a[i] = DEFAULT_VALUE;
+ res3.a[i] = DEFAULT_VALUE;
+ src2.a[i] = 145132 * i + 123123;
+ src3.a[i] = 1223 * i + 895;
+ }
+
+ CALC (res1.a, src2.a, src3.a, imm, res_ref);
+
+ res1.x = INTRINSIC (_ternarylogic_epi32) (res1.x, src2.x, src3.x,
+ imm);
+ res2.x = INTRINSIC (_mask_ternarylogic_epi32) (res2.x, mask, src2.x,
+ src3.x, imm);
+ res3.x = INTRINSIC (_maskz_ternarylogic_epi32) (mask, res3.x, src2.x,
+ src3.x, imm);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpternlogq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpternlogq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpternlogq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x, y, z;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_ternarylogic_epi64 (x, y, z, 0xF0);
+ x = _mm512_mask_ternarylogic_epi64 (x, m, y, z, 0xF0);
+ x = _mm512_maskz_ternarylogic_epi64 (m, x, y, z, 0xF0);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (long long *src1, long long *src2, long long *src3,
+ long long imm, long long *r)
+{
+ int i, j;
+ long long res, index, mask, one_mask = 1;
+ long long src1_bit, src2_bit, src3_bit, imm_bit;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ res = 0;
+ for (j = 0; j < 64; j++)
+ {
+ mask = one_mask << j;
+ src1_bit = ((src1[i] & mask) >> j) << 2;
+ src2_bit = ((src2[i] & mask) >> j) << 1;
+ src3_bit = ((src3[i] & mask) >> j);
+ index = src1_bit | src2_bit | src3_bit;
+ imm_bit = (imm & (one_mask << index)) >> index;
+ res = res | (imm_bit << j);
+ }
+ r[i] = res;
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_q) src2, src3, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+ int i, imm = 0x7D;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ res1.a[i] = DEFAULT_VALUE;
+ res2.a[i] = DEFAULT_VALUE;
+ res3.a[i] = DEFAULT_VALUE;
+ src2.a[i] = 145132 * i + 123123;
+ src3.a[i] = 1223 * i + 895;
+ }
+
+ CALC (res1.a, src2.a, src3.a, imm, res_ref);
+
+ res1.x = INTRINSIC (_ternarylogic_epi64) (res1.x, src2.x, src3.x,
+ imm);
+ res2.x = INTRINSIC (_mask_ternarylogic_epi64) (res2.x, mask, src2.x,
+ src3.x, imm);
+ res3.x = INTRINSIC (_maskz_ternarylogic_epi64) (mask, res3.x, src2.x,
+ src3.x, imm);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vptestmd\[ \\t\]+\[^\n\]*%zmm\[0-7\]\[^\n^k\]*k\[1-7\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vptestmd\[ \\t\]+\[^\n\]*%zmm\[0-7\]\[^\n^k\]*k\[1-7\]\{" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m16;
+
+void extern
+avx512f_test (void)
+{
+ m16 = _mm512_test_epi32_mask (x, x);
+ m16 = _mm512_mask_test_epi32_mask (3, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (MASK_TYPE *res, int *src1, int *src2)
+{
+ int i;
+ *res = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (src1[i] & src2[i])
+ *res = *res | one << i;
+}
+
+static void
+TEST (void)
+{
+ int i, sign = 1;
+ UNION_TYPE (AVX512F_LEN, i_d) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = i * i * sign;
+ src2.a[i] = i + 20;
+ sign = -sign;
+ }
+
+ res1 = INTRINSIC (_test_epi32_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_test_epi32_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res1 != res_ref)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res2 != res_ref)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vptestmq\[ \\t\]+\[^\n\]*%zmm\[0-7\]\[^\n^k\]*k\[1-7\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vptestmq\[ \\t\]+\[^\n\]*%zmm\[0-7\]\[^\n^k\]*k\[1-7\]\{" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m8;
+
+void extern
+avx512f_test (void)
+{
+ m8 = _mm512_test_epi64_mask (x, x);
+ m8 = _mm512_mask_test_epi64_mask (3, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (MASK_TYPE *res, long long *src1, long long *src2)
+{
+ int i;
+ *res = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (src1[i] & src2[i])
+ *res = *res | one << i;
+}
+
+static void
+TEST (void)
+{
+ int i, sign = 1;
+ UNION_TYPE (AVX512F_LEN, i_q) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+ res1 = 0;
+ res2 = 0;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = i * i * sign;
+ src2.a[i] = i + 20;
+ sign = -sign;
+ }
+
+ res1 = INTRINSIC (_test_epi64_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_test_epi64_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res1 != res_ref)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res2 != res_ref)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpunpckhdq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[\\n\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpunpckhdq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpunpckhdq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x, y, z;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_unpackhi_epi32 (y, z);
+ x = _mm512_mask_unpackhi_epi32 (x, m, y, z);
+ x = _mm512_maskz_unpackhi_epi32 (m, y, z);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (int *r, int *s1, int *s2)
+{
+ int i;
+ for (i = 0; i < SIZE / 4; i++)
+ {
+ r[4 * i] = s1[4 * i + 2];
+ r[4 * i + 1] = s2[4 * i + 2];
+ r[4 * i + 2] = s1[4 * i + 3];
+ r[4 * i + 3] = s2[4 * i + 3];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 34 * i * sign;
+ src1.a[i] = 179 * i;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_unpackhi_epi32) (src1.x, src2.x);
+ res2.x =
+ INTRINSIC (_mask_unpackhi_epi32) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_unpackhi_epi32) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpunpckhqdq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[\\n\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpunpckhqdq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpunpckhqdq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x, y, z;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_unpackhi_epi64 (y, z);
+ x = _mm512_mask_unpackhi_epi64 (x, m, y, z);
+ x = _mm512_maskz_unpackhi_epi64 (m, y, z);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (long long *r, long long *s1, long long *s2)
+{
+ int i;
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ r[2 * i] = s1[2 * i + 1];
+ r[2 * i + 1] = s2[2 * i + 1];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 34 * i * sign;
+ src1.a[i] = 179 * i;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_unpackhi_epi64) (src1.x, src2.x);
+ res2.x =
+ INTRINSIC (_mask_unpackhi_epi64) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_unpackhi_epi64) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpunpckldq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[\\n\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpunpckldq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpunpckldq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x, y, z;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_unpacklo_epi32 (y, z);
+ x = _mm512_mask_unpacklo_epi32 (x, m, y, z);
+ x = _mm512_maskz_unpacklo_epi32 (m, y, z);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (int *r, int *s1, int *s2)
+{
+ int i;
+ for (i = 0; i < SIZE / 4; i++)
+ {
+ r[4 * i] = s1[4 * i];
+ r[4 * i + 1] = s2[4 * i];
+ r[4 * i + 2] = s1[4 * i + 1];
+ r[4 * i + 3] = s2[4 * i + 1];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_d) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 34 * i * sign;
+ src1.a[i] = 179 * i;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_unpacklo_epi32) (src1.x, src2.x);
+ res2.x =
+ INTRINSIC (_mask_unpacklo_epi32) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_unpacklo_epi32) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpunpcklqdq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[\\n\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpunpcklqdq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpunpcklqdq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x, y, z;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_unpacklo_epi64 (y, z);
+ x = _mm512_mask_unpacklo_epi64 (x, m, y, z);
+ x = _mm512_maskz_unpacklo_epi64 (m, y, z);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (long long *r, long long *s1, long long *s2)
+{
+ int i;
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ r[2 * i] = s1[2 * i];
+ r[2 * i + 1] = s2[2 * i];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, i_q) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 34 * i * sign;
+ src1.a[i] = 179 * i;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_unpacklo_epi64) (src1.x, src2.x);
+ res2.x =
+ INTRINSIC (_mask_unpacklo_epi64) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_unpacklo_epi64) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpxord\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 4 } } */
+/* { dg-final { scan-assembler-times "vpxord\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpxord\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_xor_si512 (x, x);
+ x = _mm512_xor_epi32 (x, x);
+ x = _mm512_mask_xor_epi32 (x, m, x, x);
+ x = _mm512_maskz_xor_epi32 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (int *s1, int *s2, int *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ r[i] = s1[i] ^ s2[i];
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) s1, s2, res1, res2, res3, res4;
+ MASK_TYPE mask = MASK_VALUE;
+ int res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = i * sign;
+ s2.a[i] = (i + 20) * sign;
+ sign = -sign;
+ res3.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_xor_si512) (s1.x, s2.x);
+ res2.x = INTRINSIC (_xor_epi32) (s1.x, s2.x);
+ res3.x = INTRINSIC (_mask_xor_epi32) (res3.x, mask, s1.x, s2.x);
+ res4.x = INTRINSIC (_maskz_xor_epi32) (mask, s1.x, s2.x);
+
+ CALC (s1.a, s2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res1, res_ref))
+ abort ();
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res2, res_ref))
+ abort ();
+
+ MASK_MERGE (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res3, res_ref))
+ abort ();
+
+ MASK_ZERO (i_d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (res4, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vpxorq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vpxorq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpxorq\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_xor_epi64 (x, x);
+ x = _mm512_mask_xor_epi64 (x, m, x, x);
+ x = _mm512_maskz_xor_epi64 (m, x, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (long long *s1, long long *s2, long long *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ r[i] = s1[i] ^ s2[i];
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_q) s1, s2, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ long long res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = i * sign;
+ s2.a[i] = (i + 20) * sign;
+ sign = -sign;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_xor_epi64) (s1.x, s2.x);
+ res2.x = INTRINSIC (_mask_xor_epi64) (res2.x, mask, s1.x, s2.x);
+ res3.x = INTRINSIC (_maskz_xor_epi64) (mask, s1.x, s2.x);
+
+ CALC (s1.a, s2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (i_q) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vrcp14pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vrcp14pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vrcp14pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_rcp14_pd (x);
+ x = _mm512_mask_rcp14_pd (x, m, x);
+ x = _mm512_maskz_rcp14_pd (m, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (double *s, double *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = 1.0 / s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 123.456 * (i + 2000) * sign;
+ res2.a[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+ res1.x = INTRINSIC (_rcp14_pd) (s.x);
+ res2.x = INTRINSIC (_mask_rcp14_pd) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_rcp14_pd) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res1, res_ref, 0.0001))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res2, res_ref, 0.0001))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res3, res_ref, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vrcp14ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vrcp14ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vrcp14ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_rcp14_ps (x);
+ x = _mm512_mask_rcp14_ps (x, m, x);
+ x = _mm512_maskz_rcp14_ps (m, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (float *s, float *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = 1.0 / s[i];
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN,) s, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+ int i, sign = 1;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 123.456 * (i + 2000) * sign;
+ res2.a[i] = DEFAULT_VALUE;
+ sign = -sign;
+ }
+
+ res1.x = INTRINSIC (_rcp14_ps) (s.x);
+ res2.x = INTRINSIC (_mask_rcp14_ps) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_rcp14_ps) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_ROUGH_CHECK (AVX512F_LEN,) (res1, res_ref, 0.0001))
+ abort ();
+
+ MASK_MERGE () (res_ref, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN,) (res2, res_ref, 0.0001))
+ abort ();
+
+ MASK_ZERO () (res_ref, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN,) (res3, res_ref, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vrndscalepd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\{\]" 6} } */
+/* { dg-final { scan-assembler-times "vrndscalepd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 9} } */
+/* { dg-final { scan-assembler-times "vrndscalepd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 3} } */
+/* { dg-final { scan-assembler-times "vrndscalepd\[ \\t\]+\\S*,\[ \\t\]+\{sae\}\[^\n\]*%zmm\[0-9\]\[^\n\]*%zmm\[0-9\]\[^\{\]" 3} } */
+/* { dg-final { scan-assembler-times "vrndscalepd\[ \\t\]+\\S*,\[ \\t\]+\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 6} } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_roundscale_pd (x, 0x42);
+ x = _mm512_ceil_pd (x);
+ x = _mm512_floor_pd (x);
+ x = _mm512_mask_roundscale_pd (x, 2, x, 0x42);
+ x = _mm512_mask_ceil_pd (x, 2, x);
+ x = _mm512_mask_floor_pd (x, 2, x);
+ x = _mm512_maskz_roundscale_pd (2, x, 0x42);
+ x = _mm512_maskz_ceil_pd (2, x);
+ x = _mm512_maskz_floor_pd (2, x);
+
+ x = _mm512_roundscale_round_pd (x, 0x42, _MM_FROUND_NO_EXC);
+ x = _mm512_ceil_round_pd (x, _MM_FROUND_NO_EXC);
+ x = _mm512_floor_round_pd (x, _MM_FROUND_NO_EXC);
+ x = _mm512_mask_roundscale_round_pd (x, 2, x, 0x42, _MM_FROUND_NO_EXC);
+ x = _mm512_mask_ceil_round_pd (x, 2, x, _MM_FROUND_NO_EXC);
+ x = _mm512_mask_floor_round_pd (x, 2, x, _MM_FROUND_NO_EXC);
+ x = _mm512_maskz_roundscale_round_pd (2, x, 0x42, _MM_FROUND_NO_EXC);
+ x = _mm512_maskz_ceil_round_pd (2, x, _MM_FROUND_NO_EXC);
+ x = _mm512_maskz_floor_round_pd (2, x, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+#include "math.h"
+
+static void
+CALC (double *s, double *r, int imm)
+{
+ int i = 0, rc, m;
+ rc = imm & 0xf;
+ m = imm >> 4;
+ for (i = 0; i < SIZE; i++)
+ switch (rc)
+ {
+ case _MM_FROUND_FLOOR:
+ r[i] = floor (s[i] * pow (2, m)) / pow (2, m);
+ break;
+ case _MM_FROUND_CEIL:
+ r[i] = ceil (s[i] * pow (2, m)) / pow (2, m);
+ break;
+ default:
+ abort ();
+ break;
+ }
+}
+
+void static
+TEST (void)
+{
+ int imm, i, j;
+ UNION_TYPE (AVX512F_LEN, d) res1,res2,res3,s;
+ double res_ref[SIZE];
+ double res_ref_mask[SIZE];
+
+ MASK_TYPE mask = 6 ^ (0xff >> SIZE);
+
+ imm = _MM_FROUND_FLOOR | (7 << 4);
+
+ for (i = 0; i < 3; i++)
+ {
+
+ for (j = 0; j < SIZE; j++)
+ {
+ s.a[j] = j * (j + 12.0231);
+ res1.a[j] = DEFAULT_VALUE;
+ res2.a[j] = DEFAULT_VALUE;
+ res3.a[j] = DEFAULT_VALUE;
+ }
+
+ switch (i)
+ {
+ case 0:
+ imm = _MM_FROUND_FLOOR | (7 << 4);
+ res1.x = INTRINSIC (_roundscale_pd) (s.x, imm);
+ res2.x = INTRINSIC (_mask_roundscale_pd) (res2.x, mask, s.x, imm);
+ res3.x = INTRINSIC (_maskz_roundscale_pd) (mask, s.x, imm);
+ break;
+ case 1:
+ imm = _MM_FROUND_FLOOR;
+ res1.x = INTRINSIC (_floor_pd) (s.x);
+ res2.x = INTRINSIC (_mask_floor_pd) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_floor_pd) (mask, s.x);
+ break;
+ case 2:
+ imm = _MM_FROUND_CEIL;
+ res1.x = INTRINSIC (_ceil_pd) (s.x);
+ res2.x = INTRINSIC (_mask_ceil_pd) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_ceil_pd) (mask, s.x);
+ break;
+ }
+
+ CALC (s.a, res_ref, imm);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE(d) (res_ref,mask,SIZE );
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO(d) (res_ref,mask,SIZE );
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+
+ }
+}
+
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vrndscaleps\[ \\t\]+\[^\n\]*%zmm\[0-9\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6} } */
+/* { dg-final { scan-assembler-times "vrndscaleps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 9} } */
+/* { dg-final { scan-assembler-times "vrndscaleps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 3} } */
+/* { dg-final { scan-assembler-times "vrndscaleps\[ \\t\]+\\S*,\[ \\t\]+\{sae\}\[^\n\]*%zmm\[0-9\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3} } */
+/* { dg-final { scan-assembler-times "vrndscaleps\[ \\t\]+\\S*,\[ \\t\]+\{sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 6} } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_roundscale_ps (x, 0x42);
+ x = _mm512_ceil_ps (x);
+ x = _mm512_floor_ps (x);
+ x = _mm512_mask_roundscale_ps (x, 2, x, 0x42);
+ x = _mm512_mask_ceil_ps (x, 2, x);
+ x = _mm512_mask_floor_ps (x, 2, x);
+ x = _mm512_maskz_roundscale_ps (2, x, 0x42);
+ x = _mm512_maskz_ceil_ps (2, x);
+ x = _mm512_maskz_floor_ps (2, x);
+
+ x = _mm512_roundscale_round_ps (x, 0x42, _MM_FROUND_NO_EXC);
+ x = _mm512_ceil_round_ps (x, _MM_FROUND_NO_EXC);
+ x = _mm512_floor_round_ps (x, _MM_FROUND_NO_EXC);
+ x = _mm512_mask_roundscale_round_ps (x, 2, x, 0x42, _MM_FROUND_NO_EXC);
+ x = _mm512_mask_ceil_round_ps (x, 2, x, _MM_FROUND_NO_EXC);
+ x = _mm512_mask_floor_round_ps (x, 2, x, _MM_FROUND_NO_EXC);
+ x = _mm512_maskz_roundscale_round_ps (2, x, 0x42, _MM_FROUND_NO_EXC);
+ x = _mm512_maskz_ceil_round_ps (2, x, _MM_FROUND_NO_EXC);
+ x = _mm512_maskz_floor_round_ps (2, x, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#include "math.h"
+
+static void
+CALC (float *s, float *r, int imm)
+{
+ int i = 0, rc, m;
+ rc = imm & 0xf;
+ m = imm >> 4;
+ for (i = 0; i < SIZE; i++)
+ switch (rc)
+ {
+ case _MM_FROUND_FLOOR:
+ r[i] = floor (s[i] * pow (2, m)) / pow (2, m);
+ break;
+ case _MM_FROUND_CEIL:
+ r[i] = ceil (s[i] * pow (2, m)) / pow (2, m);
+ break;
+ default:
+ abort ();
+ break;
+ }
+}
+
+void static
+TEST (void)
+{
+ int imm, i, j;
+ UNION_TYPE (AVX512F_LEN,) res1, res2, res3, s;
+ float res_ref[SIZE];
+
+ MASK_TYPE mask = 6 ^ (0xffff >> SIZE);
+
+ imm = _MM_FROUND_FLOOR | (7 << 4);
+
+ for (i = 0; i < 3; i++)
+ {
+
+ for (j = 0; j < SIZE; j++)
+ {
+ s.a[j] = j * (j + 12.0231);
+ res1.a[j] = DEFAULT_VALUE;
+ res2.a[j] = DEFAULT_VALUE;
+ res3.a[j] = DEFAULT_VALUE;
+ }
+
+ switch (i)
+ {
+ case 0:
+ imm = _MM_FROUND_FLOOR | (7 << 4);
+ res1.x = INTRINSIC (_roundscale_ps) (s.x, imm);
+ res2.x = INTRINSIC (_mask_roundscale_ps) (res2.x, mask, s.x, imm);
+ res3.x = INTRINSIC (_maskz_roundscale_ps) (mask, s.x, imm);
+ break;
+ case 1:
+ imm = _MM_FROUND_FLOOR;
+ res1.x = INTRINSIC (_floor_ps) (s.x);
+ res2.x = INTRINSIC (_mask_floor_ps) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_floor_ps) (mask, s.x);
+ break;
+ case 2:
+ imm = _MM_FROUND_CEIL;
+ res1.x = INTRINSIC (_ceil_ps) (s.x);
+ res2.x = INTRINSIC (_mask_ceil_ps) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_ceil_ps) (mask, s.x);
+ break;
+ }
+
+ CALC (s.a, res_ref, imm);
+
+ if (UNION_CHECK (AVX512F_LEN,) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE ()(res_ref, mask, SIZE);
+
+ if (UNION_CHECK (AVX512F_LEN,) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO ()(res_ref, mask, SIZE);
+
+ if (UNION_CHECK (AVX512F_LEN,) (res3, res_ref))
+ abort ();
+
+ }
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vrsqrt14pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vrsqrt14pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vrsqrt14pd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_rsqrt14_pd (x);
+ x = _mm512_mask_rsqrt14_pd (x, m, x);
+ x = _mm512_maskz_rsqrt14_pd (m, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#include <math.h>
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (double *s, double *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = 1.0 / sqrt(s[i]);
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 123.456 * (i + 2000);
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_rsqrt14_pd) (s.x);
+ res2.x = INTRINSIC (_mask_rsqrt14_pd) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_rsqrt14_pd) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res1, res_ref, 0.0001))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res2, res_ref, 0.0001))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN, d) (res3, res_ref, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vrsqrt14ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 3 } } */
+/* { dg-final { scan-assembler-times "vrsqrt14ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vrsqrt14ps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_rsqrt14_ps (x);
+ x = _mm512_mask_rsqrt14_ps (x, m, x);
+ x = _mm512_maskz_rsqrt14_ps (m, x);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#include <math.h>
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (float *s, float *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = 1.0 / sqrt(s[i]);
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN,) s, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 123.456 * (i + 2000);
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_rsqrt14_ps) (s.x);
+ res2.x = INTRINSIC (_mask_rsqrt14_ps) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_rsqrt14_ps) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_ROUGH_CHECK (AVX512F_LEN,) (res1, res_ref, 0.0001))
+ abort ();
+
+ MASK_MERGE () (res_ref, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN,) (res2, res_ref, 0.0001))
+ abort ();
+
+ MASK_ZERO () (res_ref, mask, SIZE);
+ if (UNION_ROUGH_CHECK (AVX512F_LEN,) (res3, res_ref, 0.0001))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vscalefpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vscalefpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vscalefpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vscalefpd\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vscalefpd\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vscalefpd\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_scalef_pd (x, x);
+ x = _mm512_mask_scalef_pd (x, m, x, x);
+ x = _mm512_maskz_scalef_pd (m, x, x);
+ x = _mm512_scalef_round_pd (x, x, _MM_FROUND_TO_NEAREST_INT);
+ x = _mm512_mask_scalef_round_pd (x, m, x, x, _MM_FROUND_TO_NEG_INF);
+ x = _mm512_maskz_scalef_round_pd (m, x, x, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+#include <math.h>
+
+CALC (double *r, double *s1, double *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = ldexp (s1[i], floor (s2[i]));
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, d) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 1.5 + 34.67 * i * sign;
+ src2.a[i] = -22.17 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_scalef_pd) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_scalef_pd) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_scalef_pd) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vscalefps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vscalefps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vscalefps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vscalefps\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vscalefps\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vscalefps\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_scalef_ps (x, x);
+ x = _mm512_mask_scalef_ps (x, m, x, x);
+ x = _mm512_maskz_scalef_ps (m, x, x);
+ x = _mm512_scalef_round_ps (x, x, _MM_FROUND_TO_NEAREST_INT);
+ x = _mm512_mask_scalef_round_ps (x, m, x, x, _MM_FROUND_TO_POS_INF);
+ x = _mm512_maskz_scalef_round_ps (m, x, x, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#include <math.h>
+
+CALC (float *r, float *s1, float *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = ldexp (s1[i], floor (s2[i]));
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, ) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 1.5 + 34.67 * i * sign;
+ src2.a[i] = -22.17 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_scalef_ps) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_scalef_ps) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_scalef_ps) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, ) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vshuff32x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vshuff32x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vshuff32x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+__m512 x;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_shuffle_f32x4 (x, x, 56);
+ x = _mm512_mask_shuffle_f32x4 (x, 4, x, x, 56);
+ x = _mm512_maskz_shuffle_f32x4 (6, x, x, 56);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#include "string.h"
+
+void
+CALC (float *e, UNION_TYPE (AVX512F_LEN,) s1, UNION_TYPE (AVX512F_LEN,) s2,
+ int imm)
+{
+ int i, offset, selector;
+ float *source;
+ for (i = 0; i < SIZE / 4; i++)
+ {
+
+#if AVX512F_LEN == 512
+ selector = (imm >> i * 2) & 0x3;
+#else
+ selector = (imm >> i) & 0x1;
+#endif
+
+ offset = i * 4;
+ source = i * 4 * 32 < AVX512F_LEN / 2 ? s1.a : s2.a;
+ memcpy (e + offset, source + selector * 4, 16);
+ }
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN,) u1, u2, u3, s1, s2;
+ MASK_TYPE mask = MASK_VALUE;
+ float e[SIZE];
+ int i;
+ int imm = 203;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 1.2 / (i + 0.378);
+ s1.a[i] = 91.02 / (i + 4.3578);
+ u1.a[i] = DEFAULT_VALUE;
+ u2.a[i] = DEFAULT_VALUE;
+ u3.a[i] = DEFAULT_VALUE;
+ }
+
+ u1.x = INTRINSIC (_shuffle_f32x4) (s1.x, s2.x, imm);
+ u2.x = INTRINSIC (_mask_shuffle_f32x4) (u2.x, mask, s1.x, s2.x, imm);
+ u3.x = INTRINSIC (_maskz_shuffle_f32x4) (mask, s1.x, s2.x, imm);
+
+ CALC (e, s1, s2, imm);
+
+ if (UNION_CHECK (AVX512F_LEN,) (u1, e))
+ abort ();
+
+ MASK_MERGE ()(e, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (u2, e))
+ abort ();
+
+ MASK_ZERO ()(e, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (u3, e))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vshuff64x2\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vshuff64x2\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vshuff64x2\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+__m512d x;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_shuffle_f64x2 (x, x, 56);
+ x = _mm512_maskz_shuffle_f64x2 (3, x, x, 56);
+ x = _mm512_mask_shuffle_f64x2 (x, 3, x, x, 56);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+#include "string.h"
+
+void
+CALC (double *e, UNION_TYPE (AVX512F_LEN, d) s1,
+ UNION_TYPE (AVX512F_LEN, d) s2, int imm)
+{
+ int i, offset, selector;
+ double *source;
+ for (i = 0; i < SIZE / 2; i++)
+ {
+
+#if AVX512F_LEN == 512
+ selector = (imm >> i * 2) & 0x3;
+#else
+ selector = (imm >> i) & 0x1;
+#endif
+
+ offset = i * 2;
+ source = i * 2 * 64 < AVX512F_LEN / 2 ? s1.a : s2.a;
+ memcpy (e + offset, source + selector * 2, 16);
+ }
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) u1, u2, u3, s1, s2;
+ MASK_TYPE mask = MASK_VALUE;
+ double e[SIZE];
+ int i;
+ int imm = 203;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 1.2 / (i + 0.378);
+ s1.a[i] = 91.02 / (i + 4.3578);
+ u1.a[i] = DEFAULT_VALUE;
+ u2.a[i] = DEFAULT_VALUE;
+ u3.a[i] = DEFAULT_VALUE;
+ }
+
+ u1.x = INTRINSIC (_shuffle_f64x2) (s1.x, s2.x, imm);
+ u2.x = INTRINSIC (_mask_shuffle_f64x2) (u2.x, mask, s1.x, s2.x, imm);
+ u3.x = INTRINSIC (_maskz_shuffle_f64x2) (mask, s1.x, s2.x, imm);
+
+ CALC (e, s1, s2, imm);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (u1, e))
+ abort ();
+
+ MASK_MERGE (d) (e, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (u2, e))
+ abort ();
+
+ MASK_ZERO (d) (e, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (u3, e))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vshufi32x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vshufi32x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vshufi32x4\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+__m512i x;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_shuffle_i32x4 (x, x, 56);
+ x = _mm512_mask_shuffle_i32x4 (x, 8, x, x, 56);
+ x = _mm512_maskz_shuffle_i32x4 (8, x, x, 56);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+#include "string.h"
+
+void
+CALC (int *e, UNION_TYPE (AVX512F_LEN, i_d) s1,
+ UNION_TYPE (AVX512F_LEN, i_d) s2, int imm)
+{
+ int i, offset, selector;
+ int *source;
+ for (i = 0; i < SIZE / 4; i++)
+ {
+
+#if AVX512F_LEN == 512
+ selector = (imm >> i * 2) & 0x3;
+#else
+ selector = (imm >> i) & 0x1;
+#endif
+
+ offset = i * 4;
+ source = i * 4 * 32 < AVX512F_LEN / 2 ? s1.a : s2.a;
+ memcpy (e + offset, source + selector * 4, 16);
+ }
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_d) u1, u2, u3, s1, s2;
+ MASK_TYPE mask = MASK_VALUE;
+ int e[SIZE];
+ int i;
+ int imm = 203;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 1.2 / (i + 0.378);
+ s1.a[i] = 91.02 / (i + 4.3578);
+ u1.a[i] = DEFAULT_VALUE;
+ u2.a[i] = DEFAULT_VALUE;
+ u3.a[i] = DEFAULT_VALUE;
+ }
+
+ u1.x = INTRINSIC (_shuffle_i32x4) (s1.x, s2.x, imm);
+ u2.x = INTRINSIC (_mask_shuffle_i32x4) (u2.x, mask, s1.x, s2.x, imm);
+ u3.x = INTRINSIC (_maskz_shuffle_i32x4) (mask, s1.x, s2.x, imm);
+
+ CALC (e, s1, s2, imm);
+
+ if (UNION_CHECK (AVX512F_LEN, i_d) (u1, e))
+ abort ();
+
+ MASK_MERGE (i_d) (e, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (u2, e))
+ abort ();
+
+ MASK_ZERO (i_d) (e, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_d) (u3, e))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vshufi64x2\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vshufi64x2\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vshufi64x2\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+__m512i x;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_shuffle_i64x2 (x, x, 56);
+ x = _mm512_mask_shuffle_i64x2 (x, 3, x, x, 56);
+ x = _mm512_maskz_shuffle_i64x2 (2, x, x, 56);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+#include "string.h"
+
+void
+CALC (long long *e, UNION_TYPE (AVX512F_LEN, i_q) s1,
+ UNION_TYPE (AVX512F_LEN, i_q) s2, int imm)
+{
+ int i, offset, selector;
+ long long *source;
+ for (i = 0; i < SIZE / 2; i++)
+ {
+
+#if AVX512F_LEN == 512
+ selector = (imm >> i * 2) & 0x3;
+#else
+ selector = (imm >> i) & 0x1;
+#endif
+
+ offset = i * 2;
+ source = i * 2 * 64 < AVX512F_LEN / 2 ? s1.a : s2.a;
+ memcpy (e + offset, source + selector * 2, 16);
+ }
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, i_q) u1, u2, u3, s1, s2;
+ MASK_TYPE mask = MASK_VALUE;
+ long long e[SIZE];
+ int i;
+ int imm = 203;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 1.2 / (i + 0.378);
+ s1.a[i] = 91.02 / (i + 4.3578);
+ u1.a[i] = DEFAULT_VALUE;
+ u2.a[i] = DEFAULT_VALUE;
+ u3.a[i] = DEFAULT_VALUE;
+ }
+
+ u1.x = INTRINSIC (_shuffle_i64x2) (s1.x, s2.x, imm);
+ u2.x = INTRINSIC (_mask_shuffle_i64x2) (u2.x, mask, s1.x, s2.x, imm);
+ u3.x = INTRINSIC (_maskz_shuffle_i64x2) (mask, s1.x, s2.x, imm);
+
+ CALC (e, s1, s2, imm);
+
+ if (UNION_CHECK (AVX512F_LEN, i_q) (u1, e))
+ abort ();
+
+ MASK_MERGE (i_q) (e, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (u2, e))
+ abort ();
+
+ MASK_ZERO (i_q) (e, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, i_q) (u3, e))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vshufpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vshufpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vshufpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+__m512d x;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_shuffle_pd (x, x, 56);
+ x = _mm512_mask_shuffle_pd (x, 2, x, x, 56);
+ x = _mm512_maskz_shuffle_pd (2, x, x, 56);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+void static
+CALC (double *e, UNION_TYPE (AVX512F_LEN, d) s1,
+ UNION_TYPE (AVX512F_LEN, d) s2, int imm)
+{
+ e[0] = (imm & (1 << 0)) ? s1.a[1] : s1.a[0];
+ e[1] = (imm & (1 << 1)) ? s2.a[1] : s2.a[0];
+#if AVX512F_LEN > 128
+ e[2] = (imm & (1 << 2)) ? s1.a[3] : s1.a[2];
+ e[3] = (imm & (1 << 3)) ? s2.a[3] : s2.a[2];
+#if AVX512F_LEN > 256
+ e[4] = (imm & (1 << 4)) ? s1.a[5] : s1.a[4];
+ e[5] = (imm & (1 << 5)) ? s2.a[5] : s2.a[4];
+ e[6] = (imm & (1 << 6)) ? s1.a[7] : s1.a[6];
+ e[7] = (imm & (1 << 7)) ? s2.a[7] : s2.a[6];
+#endif
+#endif
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) u1, u2, u3, s1, s2;
+ double e[SIZE];
+ MASK_TYPE mask = MASK_VALUE;
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 2134.3343 * i + 54846.4641;
+ s2.a[i] = 856.43576 * i + 1124.209;
+ u1.a[i] = DEFAULT_VALUE;
+ u2.a[i] = DEFAULT_VALUE;
+ u3.a[i] = DEFAULT_VALUE;
+ }
+
+ u1.x = INTRINSIC (_shuffle_pd) (s1.x, s2.x, 120);
+ u2.x = INTRINSIC (_mask_shuffle_pd) (u2.x, mask, s1.x, s2.x, 120);
+ u3.x = INTRINSIC (_maskz_shuffle_pd) (mask, s1.x, s2.x, 120);
+ CALC (e, s1, s2, 120);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (u1, e))
+ abort ();
+
+ MASK_MERGE (d) (e, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (u2, e))
+ abort ();
+
+ MASK_ZERO (d) (e, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (u3, e))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vshufps\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vshufps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vshufps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+__m512 x;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_shuffle_ps (x, x, 56);
+ x = _mm512_mask_shuffle_ps (x, 2, x, x, 56);
+ x = _mm512_maskz_shuffle_ps (2, x, x, 56);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+void
+CALC (float *e, UNION_TYPE (AVX512F_LEN,) s1, UNION_TYPE (AVX512F_LEN,) s2,
+ int imm)
+{
+ e[0] = s1.a[(imm >> 0) & 0x3];
+ e[1] = s1.a[(imm >> 2) & 0x3];
+ e[2] = s2.a[(imm >> 4) & 0x3];
+ e[3] = s2.a[(imm >> 6) & 0x3];
+#if AVX512F_LEN > 128
+ e[4] = s1.a[4 + ((imm >> 0) & 0x3)];
+ e[5] = s1.a[4 + ((imm >> 2) & 0x3)];
+ e[6] = s2.a[4 + ((imm >> 4) & 0x3)];
+ e[7] = s2.a[4 + ((imm >> 6) & 0x3)];
+#if AVX512F_LEN > 256
+ e[8] = s1.a[8 + ((imm >> 0) & 0x3)];
+ e[9] = s1.a[8 + ((imm >> 2) & 0x3)];
+ e[10] = s2.a[8 + ((imm >> 4) & 0x3)];
+ e[11] = s2.a[8 + ((imm >> 6) & 0x3)];
+ e[12] = s1.a[12 + ((imm >> 0) & 0x3)];
+ e[13] = s1.a[12 + ((imm >> 2) & 0x3)];
+ e[14] = s2.a[12 + ((imm >> 4) & 0x3)];
+ e[15] = s2.a[12 + ((imm >> 6) & 0x3)];
+#endif
+#endif
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN,) u1, u2, u3, s1, s2;
+ float e[SIZE];
+ int i, sign;
+ MASK_TYPE mask = MASK_VALUE;
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = 1.5 + 34.67 * i * sign;
+ s2.a[i] = -22.17 * i * sign;
+ u1.a[i] = DEFAULT_VALUE;
+ u2.a[i] = DEFAULT_VALUE;
+ u3.a[i] = DEFAULT_VALUE;
+ sign = sign * -1;
+ }
+
+
+ u1.x = INTRINSIC (_shuffle_ps) (s1.x, s2.x, 203);
+ u2.x = INTRINSIC (_mask_shuffle_ps) (u2.x, mask, s1.x, s2.x, 203);
+ u3.x = INTRINSIC (_maskz_shuffle_ps) (mask, s1.x, s2.x, 203);
+
+ CALC (e, s1, s2, 203);
+
+ if (UNION_CHECK (AVX512F_LEN,) (u1, e))
+ abort ();
+
+ MASK_MERGE ()(e, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (u2, e))
+ abort ();
+
+ MASK_ZERO ()(e, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (u3, e))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vsqrtpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vsqrtpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vsqrtpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vsqrtpd\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vsqrtpd\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vsqrtpd\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_sqrt_pd (x);
+ x = _mm512_mask_sqrt_pd (x, m, x);
+ x = _mm512_maskz_sqrt_pd (m, x);
+ x = _mm512_sqrt_round_pd (x, _MM_FROUND_TO_NEAREST_INT);
+ x = _mm512_mask_sqrt_round_pd (x, m, x, _MM_FROUND_TO_NEG_INF);
+ x = _mm512_maskz_sqrt_round_pd (m, x, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#include <math.h>
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (double *s, double *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = sqrt(s[i]);
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 123.456 * (i + 2000);
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_sqrt_pd) (s.x);
+ res2.x = INTRINSIC (_mask_sqrt_pd) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_sqrt_pd) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_FP_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (UNION_FP_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (UNION_FP_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vsqrtps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vsqrtps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vsqrtps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vsqrtps\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vsqrtps\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vsqrtps\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_sqrt_ps (x);
+ x = _mm512_mask_sqrt_ps (x, m, x);
+ x = _mm512_maskz_sqrt_ps (m, x);
+ x = _mm512_sqrt_round_ps (x, _MM_FROUND_TO_NEAREST_INT);
+ x = _mm512_mask_sqrt_round_ps (x, m, x, _MM_FROUND_TO_POS_INF);
+ x = _mm512_maskz_sqrt_round_ps (m, x, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#include <math.h>
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+static void
+CALC (float *s, float *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = sqrt(s[i]);
+ }
+}
+
+static void
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN,) s, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s.a[i] = 123.456 * (i + 2000);
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_sqrt_ps) (s.x);
+ res2.x = INTRINSIC (_mask_sqrt_ps) (res2.x, mask, s.x);
+ res3.x = INTRINSIC (_maskz_sqrt_ps) (mask, s.x);
+
+ CALC (s.a, res_ref);
+
+ if (UNION_FP_CHECK (AVX512F_LEN,) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE () (res_ref, mask, SIZE);
+ if (UNION_FP_CHECK (AVX512F_LEN,) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO () (res_ref, mask, SIZE);
+ if (UNION_FP_CHECK (AVX512F_LEN,) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vsubpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vsubpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vsubpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vsubpd\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vsubpd\[ \\t\]+\[^\n\]*\{rd-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vsubpd\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_sub_pd (x, x);
+ x = _mm512_mask_sub_pd (x, m, x, x);
+ x = _mm512_maskz_sub_pd (m, x, x);
+ x = _mm512_sub_round_pd (x, x, _MM_FROUND_TO_NEAREST_INT);
+ x = _mm512_mask_sub_round_pd (x, m, x, x, _MM_FROUND_TO_NEG_INF);
+ x = _mm512_maskz_sub_round_pd (m, x, x, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+CALC (double *r, double *s1, double *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[i] - s2[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN, d) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 1.5 + 34.67 * i * sign;
+ src2.a[i] = -22.17 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_sub_pd) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_sub_pd) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_sub_pd) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-final { scan-assembler-times "vsubps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\{\]" 6 } } */
+/* { dg-final { scan-assembler-times "vsubps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 2 } } */
+/* { dg-final { scan-assembler-times "vsubps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 2 } } */
+/* { dg-final { scan-assembler-times "vsubps\[ \\t\]+\[^\n\]*\{rn-sae\}\[^\n\]*%zmm\[0-9\]" 1 } } */
+/* { dg-final { scan-assembler-times "vsubps\[ \\t\]+\[^\n\]*\{ru-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vsubps\[ \\t\]+\[^\n\]*\{rz-sae\}\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x;
+volatile __mmask16 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_sub_ps (x, x);
+ x = _mm512_mask_sub_ps (x, m, x, x);
+ x = _mm512_maskz_sub_ps (m, x, x);
+ x = _mm512_sub_round_ps (x, x, _MM_FROUND_TO_NEAREST_INT);
+ x = _mm512_mask_sub_round_ps (x, m, x, x, _MM_FROUND_TO_POS_INF);
+ x = _mm512_maskz_sub_round_ps (m, x, x, _MM_FROUND_TO_ZERO);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+CALC (float *r, float *s1, float *s2)
+{
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ r[i] = s1[i] - s2[i];
+ }
+}
+
+void static
+TEST (void)
+{
+ int i, sign;
+ UNION_TYPE (AVX512F_LEN,) res1, res2, res3, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+
+ sign = -1;
+ for (i = 0; i < SIZE; i++)
+ {
+ src1.a[i] = 1.5 + 34.67 * i * sign;
+ src2.a[i] = -22.17 * i * sign;
+ sign = sign * -1;
+ }
+ for (i = 0; i < SIZE; i++)
+ res2.a[i] = DEFAULT_VALUE;
+
+ res1.x = INTRINSIC (_sub_ps) (src1.x, src2.x);
+ res2.x = INTRINSIC (_mask_sub_ps) (res2.x, mask, src1.x, src2.x);
+ res3.x = INTRINSIC (_maskz_sub_ps) (mask, src1.x, src2.x);
+
+ CALC (res_ref, src1.a, src2.a);
+
+ if (UNION_CHECK (AVX512F_LEN,) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN,) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler "vucomisd\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm" } } */
+
+#include <immintrin.h>
+
+volatile __m128d x;
+volatile int res;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm_comi_round_sd (x, x, _CMP_NLE_UQ, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler "vucomiss\[ \\t\]+\[^\n\]*\{sae\}\[^\n\]*%xmm" } } */
+
+#include <immintrin.h>
+
+volatile __m128 x;
+volatile int res;
+
+void extern
+avx512f_test (void)
+{
+ res = _mm_comi_round_ss (x, x, _CMP_EQ_OQ, _MM_FROUND_NO_EXC);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vunpckhpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vunpckhpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vunpckhpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x, y, z;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_unpackhi_pd (y, z);
+ x = _mm512_mask_unpackhi_pd (x, m, y, z);
+ x = _mm512_maskz_unpackhi_pd (m, y, z);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+void static
+CALC (double *s1, double *s2, double *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ r[2 * i] = s1[2 * i + 1];
+ r[2 * i + 1] = s2[2 * i + 1];
+ }
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s1, s2, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = i * 123.2 + 32.6;
+ s2.a[i] = i + 2.5;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_unpackhi_pd) (s1.x, s2.x);
+ res2.x = INTRINSIC (_mask_unpackhi_pd) (res2.x, mask, s1.x, s2.x);
+ res3.x = INTRINSIC (_maskz_unpackhi_pd) (mask, s1.x, s2.x);
+
+ CALC (s1.a, s2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vunpckhps\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vunpckhps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vunpckhps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x, y, z;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_unpackhi_ps (y, z);
+ x = _mm512_mask_unpackhi_ps (x, m, y, z);
+ x = _mm512_maskz_unpackhi_ps (m, y, z);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+void static
+CALC (float *s1, float *s2, float *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE / 4; i++)
+ {
+ r[4 * i] = s1[4 * i + 2];
+ r[4 * i + 1] = s2[4 * i + 2];
+ r[4 * i + 2] = s1[4 * i + 3];
+ r[4 * i + 3] = s2[4 * i + 3];
+ }
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, ) s1, s2, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ float res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = i * 123.2 + 32.6;
+ s2.a[i] = i + 2.5;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_unpackhi_ps) (s1.x, s2.x);
+ res2.x = INTRINSIC (_mask_unpackhi_ps) (res2.x, mask, s1.x, s2.x);
+ res3.x = INTRINSIC (_maskz_unpackhi_ps) (mask, s1.x, s2.x);
+
+ CALC (s1.a, s2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, ) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO () (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vunpcklpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vunpcklpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vunpcklpd\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512d x, y, z;
+volatile __mmask8 m;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_unpacklo_pd (y, z);
+ x = _mm512_mask_unpacklo_pd (x, m, y, z);
+ x = _mm512_maskz_unpacklo_pd (m, y, z);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+void static
+CALC (double *s1, double *s2, double *r)
+{
+ int i;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ r[2 * i] = s1[2 * i];
+ r[2 * i + 1] = s2[2 * i];
+ }
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN, d) s1, s2, res1, res2, res3;
+ MASK_TYPE mask = MASK_VALUE;
+ double res_ref[SIZE];
+ int i;
+
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = i * 123.2 + 32.6;
+ s2.a[i] = i + 2.5;
+ res2.a[i] = DEFAULT_VALUE;
+ }
+
+ res1.x = INTRINSIC (_unpacklo_pd) (s1.x, s2.x);
+ res2.x = INTRINSIC (_mask_unpacklo_pd) (res2.x, mask, s1.x, s2.x);
+ res3.x = INTRINSIC (_maskz_unpacklo_pd) (mask, s1.x, s2.x);
+
+ CALC (s1.a, s2.a, res_ref);
+
+ if (UNION_CHECK (AVX512F_LEN, d) (res1, res_ref))
+ abort ();
+
+ MASK_MERGE (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res2, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, d) (res3, res_ref))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vunpcklps\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 3 } } */
+/* { dg-final { scan-assembler-times "vunpcklps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\{z\}" 1 } } */
+/* { dg-final { scan-assembler-times "vunpcklps\[ \\t\]+\[^\n\]*%zmm\[0-9\]\{%k\[1-7\]\}\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512 x, y, z;
+
+void extern
+avx512f_test (void)
+{
+ x = _mm512_unpacklo_ps (y, z);
+ x = _mm512_mask_unpacklo_ps (x, 2, y, z);
+ x = _mm512_maskz_unpacklo_ps (2, y, z);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f" } */
+/* { dg-require-effective-target avx512f } */
+
+#define AVX512F
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+#include "avx512f-mask-type.h"
+
+void static
+CALC (float *e, float *s1, float *s2)
+{
+ int i;
+ for (i = 0; i < SIZE / 4; i++)
+ {
+ e[4 * i] = s1[4 * i];
+ e[4 * i + 1] = s2[4 * i];
+ e[4 * i + 2] = s1[4 * i + 1];
+ e[4 * i + 3] = s2[4 * i + 1];
+ }
+}
+
+void static
+TEST (void)
+{
+ UNION_TYPE (AVX512F_LEN,) s1, s2, u1, u2, u3;
+ MASK_TYPE mask = MASK_VALUE;
+ float e[SIZE];
+ int i;
+ for (i = 0; i < SIZE; i++)
+ {
+ s1.a[i] = i * 123.2 + 32.6;
+ s2.a[i] = i + 2.5;
+ u1.a[i]= DEFAULT_VALUE;
+ u2.a[i]= DEFAULT_VALUE;
+ u3.a[i]= DEFAULT_VALUE;
+ }
+
+ u1.x = INTRINSIC (_unpacklo_ps) (s1.x, s2.x);
+ u2.x = INTRINSIC (_mask_unpacklo_ps) (u2.x, mask, s1.x, s2.x);
+ u3.x = INTRINSIC (_maskz_unpacklo_ps) (mask, s1.x, s2.x);
+
+ CALC (e, s1.a, s2.a);
+
+ if (UNION_CHECK (AVX512F_LEN,) (u1, e))
+ abort ();
+
+ MASK_MERGE () (e, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (u2, e))
+ abort ();
+
+ MASK_ZERO () (e, mask, SIZE);
+ if (UNION_CHECK (AVX512F_LEN, ) (u3, e))
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O3 -mavx512f" } */
+/* { dg-final { scan-assembler "(vpblendmd|vmovdqa32)" } } */
+
+unsigned int x[128];
+unsigned int y[128];
+
+void
+foo ()
+{
+ int i;
+ for (i = 0; i < 128; i++)
+ x[i] = y[i] > 3 ? 2 : 0;
+}
--- /dev/null
+/* { dg-do compile { target { ! { ia32 } } } } */
+/* { dg-options "-O2 -mavx512f -ffixed-xmm0 -ffixed-xmm1 -ffixed-xmm2 -ffixed-xmm3 -ffixed-xmm4 -ffixed-xmm5 -ffixed-xmm6 -ffixed-xmm7 -ffixed-xmm8 -ffixed-xmm9 -ffixed-xmm10 -ffixed-xmm11 -ffixed-xmm12 -ffixed-xmm13 -ffixed-xmm14 -ffixed-xmm15" } */
+
+volatile float a,b,c,d;
+
+void foo()
+{
+ __asm__ __volatile__( "vcmpss $1,%1, %2,%3;" : "=x"(c) : "x"(a),"x"(b),"x"(d) );/* { dg-error "inconsistent operand constraints" } */
+}
--- /dev/null
+/* { dg-do compile { target { ! { ia32 } } } } */
+/* { dg-options "-O2 -mavx512f -ffixed-xmm0 -ffixed-xmm1 -ffixed-xmm2 -ffixed-xmm3 -ffixed-xmm4 -ffixed-xmm5 -ffixed-xmm6 -ffixed-xmm7 -ffixed-xmm8 -ffixed-xmm9 -ffixed-xmm10 -ffixed-xmm11 -ffixed-xmm12 -ffixed-xmm13 -ffixed-xmm14 -ffixed-xmm15" } */
+/* { dg-final { scan-assembler "vaddss\[ \\t\]+\[^\n\]*%xmm(1\[6-9\]|2\[0-9\]|3\[0-1\])\[^\{\]" } } */
+
+volatile float a, b, c, d;
+
+void foo()
+{
+ __asm__ __volatile__( "vaddss %1, %2, %3;" : "=v"(c) : "v"(a),"v"(b),"v"(d) );
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512pf -O2" } */
+/* { dg-final { scan-assembler-times "vgatherpf0dps\[ \\t\]+\[^\n\]*\{%k\[1-7\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i idx;
+volatile __mmask16 m16;
+int *base;
+
+void extern
+avx512pf_test (void)
+{
+ _mm512_mask_prefetch_i32gather_ps (idx, m16, base, 8, 0);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512pf -O2" } */
+/* { dg-final { scan-assembler-times "vgatherpf0qps\[ \\t\]+\[^\n\]*\{%k\[1-7\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i idx;
+volatile __mmask8 m8;
+int *base;
+
+void extern
+avx512pf_test (void)
+{
+ _mm512_mask_prefetch_i64gather_ps (idx, m8, base, 8, 0);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512pf -O2" } */
+/* { dg-final { scan-assembler-times "vgatherpf1dps\[ \\t\]+\[^\n\]*\{%k\[1-7\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i idx;
+volatile __mmask16 m16;
+int *base;
+
+void extern
+avx512pf_test (void)
+{
+ _mm512_mask_prefetch_i32gather_ps (idx, m16, base, 8, 1);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512pf -O2" } */
+/* { dg-final { scan-assembler-times "vgatherpf1qps\[ \\t\]+\[^\n\]*\{%k\[1-7\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i idx;
+volatile __mmask8 m8;
+int *base;
+
+void extern
+avx512pf_test (void)
+{
+ _mm512_mask_prefetch_i64gather_ps (idx, m8, base, 8, 1);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512pf -O2" } */
+/* { dg-final { scan-assembler-times "vscatterpf0dps\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 2 } } */
+/* { dg-final { scan-assembler-times "vscatterpf0dps\[ \\t\]+\[^\n\]*\{%k\[1-7\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i idx;
+volatile __mmask16 m16;
+int *base;
+
+void extern
+avx512pf_test (void)
+{
+ _mm512_prefetch_i32scatter_ps (base, idx, 8, 0);
+ _mm512_mask_prefetch_i32scatter_ps (base, m16, idx, 8, 0);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512pf -O2" } */
+/* { dg-final { scan-assembler-times "vscatterpf0qps\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 2 } } */
+/* { dg-final { scan-assembler-times "vscatterpf0qps\[ \\t\]+\[^\n\]*\{%k\[1-7\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i idx;
+volatile __mmask8 m8;
+int *base;
+
+void extern
+avx512pf_test (void)
+{
+ _mm512_prefetch_i64scatter_ps (base, idx, 8, 0);
+ _mm512_mask_prefetch_i64scatter_ps (base, m8, idx, 8, 0);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512pf -O2" } */
+/* { dg-final { scan-assembler-times "vscatterpf1dps\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 2 } } */
+/* { dg-final { scan-assembler-times "vscatterpf1dps\[ \\t\]+\[^\n\]*\{%k\[1-7\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i idx;
+volatile __mmask16 m16;
+int *base;
+
+void extern
+avx512pf_test (void)
+{
+ _mm512_prefetch_i32scatter_ps (base, idx, 8, 1);
+ _mm512_mask_prefetch_i32scatter_ps (base, m16, idx, 8, 1);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512pf -O2" } */
+/* { dg-final { scan-assembler-times "vscatterpf1qps\[ \\t\]+\[^\n\]*%zmm\[0-9\]" 2 } } */
+/* { dg-final { scan-assembler-times "vscatterpf1qps\[ \\t\]+\[^\n\]*\{%k\[1-7\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m512i idx;
+volatile __mmask8 m8;
+int *base;
+
+void extern
+avx512pf_test (void)
+{
+ _mm512_prefetch_i64scatter_ps (base, idx, 8, 1);
+ _mm512_mask_prefetch_i64scatter_ps (base, m8, idx, 8, 1);
+}
} "-mrtm" ]
}
+# Return 1 if avx512f instructions can be compiled.
+proc check_effective_target_avx512f { } {
+ return [check_no_compiler_messages avx512f object {
+ typedef long long __v8di __attribute__ ((__vector_size__ (64)));
+ __v8di
+ mm512_and_epi64 (__v8di __X, __v8di __Y)
+ {
+ __v8di __W;
+ return __builtin_ia32_pandq512_mask (__X, __Y, __W, -1);
+ }
+ } "-mavx512f" ]
+}
+
+# Return 1 if avx512cd instructions can be compiled.
+proc check_effective_target_avx512cd { } {
+ return [check_no_compiler_messages avx512cd_trans object {
+ typedef long long __v8di __attribute__ ((__vector_size__ (64)));
+ __v8di
+ _mm512_conflict_epi64 (__v8di __W, __v8di __A)
+ {
+ return (__v8di) __builtin_ia32_vpconflictdi_512_mask ((__v8di) __A,
+ (__v8di) __W,
+ -1);
+ }
+ } "-Wno-psabi -mavx512cd" ]
+}
+
+# Return 1 if avx512er instructions can be compiled.
+proc check_effective_target_avx512er { } {
+ return [check_no_compiler_messages avx512er_trans object {
+ typedef float __v16sf __attribute__ ((__vector_size__ (64)));
+ __v16sf
+ mm512_exp2a23_ps (__v16sf __X)
+ {
+ __v16sf __W;
+ return __builtin_ia32_exp2ps_mask (__X, __W, -1, 4);
+ }
+ } "-Wno-psabi -mavx512er" ]
+}
+
# If the linker used understands -M <mapfile>, pass it to clear hardware
# capabilities set by the Sun assembler.
# Try mapfile syntax v2 first which is the only way to clear hwcap_2 flags.
} bits __attribute__((packed));
};
#endif
+
+#define CHECK_FP_EXP(UINON_TYPE, VALUE_TYPE, ESP, FMT) \
+static int \
+__attribute__((noinline, unused)) \
+check_fp_##UINON_TYPE (UINON_TYPE u, const VALUE_TYPE *v) \
+{ \
+ int i; \
+ int err = 0; \
+ \
+ for (i = 0; i < ARRAY_SIZE (u.a); i++) \
+ if (u.a[i] > (v[i] + (ESP)) || u.a[i] < (v[i] - (ESP))) \
+ { \
+ err++; \
+ PRINTF ("%i: " FMT " != " FMT "\n", \
+ i, v[i], u.a[i]); \
+ } \
+ return err; \
+}
+
+CHECK_FP_EXP (union128, float, ESP_FLOAT, "%f")
+#ifdef __SSE2__
+CHECK_FP_EXP (union128d, double, ESP_DOUBLE, "%f")
+#endif
--- /dev/null
+#include <immintrin.h>
+#include "m256-check.h"
+
+typedef union
+{
+ __m512i x;
+ char a[64];
+} union512i_b;
+
+typedef union
+{
+ __m512i x;
+ short a[32];
+} union512i_w;
+
+typedef union
+{
+ __m512i x;
+ int a[16];
+} union512i_d;
+
+typedef union
+{
+ __m512i x;
+ long long a[8];
+} union512i_q;
+
+typedef union
+{
+ __m512 x;
+ float a[16];
+} union512;
+
+typedef union
+{
+ __m512d x;
+ double a[8];
+} union512d;
+
+CHECK_EXP (union512i_b, char, "%d")
+CHECK_EXP (union512i_w, short, "%d")
+CHECK_EXP (union512i_d, int, "0x%x")
+CHECK_EXP (union512i_q, long long, "0x%llx")
+CHECK_EXP (union512, float, "%f")
+CHECK_EXP (union512d, double, "%f")
+
+CHECK_FP_EXP (union512, float, ESP_FLOAT, "%f")
+CHECK_FP_EXP (union512d, double, ESP_DOUBLE, "%f")
+
+#define CHECK_ROUGH_EXP(UINON_TYPE, VALUE_TYPE, FMT) \
+static int \
+__attribute__((noinline, unused)) \
+check_rough_##UINON_TYPE (UINON_TYPE u, const VALUE_TYPE *v, \
+ VALUE_TYPE eps) \
+{ \
+ int i; \
+ int err = 0; \
+ \
+ for (i = 0; i < ARRAY_SIZE (u.a); i++) \
+ { \
+ VALUE_TYPE rel_err = (u.a[i] - v[i]) / v[i]; \
+ if (((rel_err < 0) ? -rel_err : rel_err) > eps) \
+ { \
+ err++; \
+ PRINTF ("%i: " FMT " != " FMT "\n", \
+ i, v[i], u.a[i]); \
+ } \
+ } \
+ return err; \
+}
+
+CHECK_ROUGH_EXP (union512, float, "%f")
+CHECK_ROUGH_EXP (union512d, double, "%f")
popcntintrin.h and mm_malloc.h are usable
with -O -std=c89 -pedantic-errors. */
/* { dg-do compile } */
-/* { dg-options "-O -std=c89 -pedantic-errors -march=k8 -msse4a -m3dnow -mavx -mavx2 -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi -mbmi2 -mtbm -mlwp -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -mprfchw -madx -mfxsr -mxsaveopt" } */
+/* { dg-options "-O -std=c89 -pedantic-errors -march=k8 -msse4a -m3dnow -mavx -mavx2 -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi -mbmi2 -mtbm -mlwp -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -mprfchw -madx -mfxsr -mxsaveopt -mavx512f" } */
#include <x86intrin.h>
/* { dg-do compile } */
-/* { dg-options "-O2 -Werror-implicit-function-declaration -march=k8 -msse4a -m3dnow -mavx -mavx2 -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi -mbmi2 -mtbm -mlwp -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -mprfchw -madx -mfxsr -mxsaveopt" } */
+/* { dg-options "-O2 -Werror-implicit-function-declaration -march=k8 -msse4a -m3dnow -mavx -mavx2 -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi -mbmi2 -mtbm -mlwp -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -mprfchw -madx -mfxsr -mxsaveopt -mavx512f" } */
#include <mm_malloc.h>
#define __builtin_ia32_vcvtps2ph(A, I) __builtin_ia32_vcvtps2ph(A, 1)
#define __builtin_ia32_vcvtps2ph256(A, I) __builtin_ia32_vcvtps2ph256(A, 1)
+/* avx512pfintrin.h */
+#define __builtin_ia32_gatherpfdps(A, B, C, D, E) __builtin_ia32_gatherpfdps (A, B, C, 1, 1)
+#define __builtin_ia32_gatherpfqps(A, B, C, D, E) __builtin_ia32_gatherpfqps (A, B, C, 1, 1)
+#define __builtin_ia32_scatterpfdps(A, B, C, D, E) __builtin_ia32_scatterpfdps (A, B, C, 1, 1)
+#define __builtin_ia32_scatterpfqps(A, B, C, D, E) __builtin_ia32_scatterpfqps (A, B, C, 1, 1)
+
+/* avx512erintrin.h */
+#define __builtin_ia32_exp2pd_mask(A, B, C, D) __builtin_ia32_exp2pd_mask (A, B, C, 1)
+#define __builtin_ia32_exp2ps_mask(A, B, C, D) __builtin_ia32_exp2ps_mask (A, B, C, 1)
+#define __builtin_ia32_rcp28pd_mask(A, B, C, D) __builtin_ia32_rcp28pd_mask (A, B, C, 1)
+#define __builtin_ia32_rcp28ps_mask(A, B, C, D) __builtin_ia32_rcp28ps_mask (A, B, C, 1)
+#define __builtin_ia32_rsqrt28pd_mask(A, B, C, D) __builtin_ia32_rsqrt28pd_mask (A, B, C, 1)
+#define __builtin_ia32_rsqrt28ps_mask(A, B, C, D) __builtin_ia32_rsqrt28ps_mask (A, B, C, 1)
+
/* wmmintrin.h */
#define __builtin_ia32_aeskeygenassist128(X, C) __builtin_ia32_aeskeygenassist128(X, 1)
#define __builtin_ia32_pclmulqdq128(X, Y, I) __builtin_ia32_pclmulqdq128(X, Y, 1)
/* rtmintrin.h */
#define __builtin_ia32_xabort (N) __builtin_ia32_xabort (1)
+
+/* avx512fintrin.h */
+#define __builtin_ia32_addpd512_mask(A, B, C, D, E) __builtin_ia32_addpd512_mask(A, B, C, D, 1)
+#define __builtin_ia32_addps512_mask(A, B, C, D, E) __builtin_ia32_addps512_mask(A, B, C, D, 1)
+#define __builtin_ia32_addsd_mask(A, B, C, D, E) __builtin_ia32_addsd_mask(A, B, C, D, 1)
+#define __builtin_ia32_addss_mask(A, B, C, D, E) __builtin_ia32_addss_mask(A, B, C, D, 1)
+#define __builtin_ia32_alignd512_mask(A, B, F, D, E) __builtin_ia32_alignd512_mask(A, B, 1, D, E)
+#define __builtin_ia32_alignq512_mask(A, B, F, D, E) __builtin_ia32_alignq512_mask(A, B, 1, D, E)
+#define __builtin_ia32_cmpd512_mask(A, B, E, D) __builtin_ia32_cmpd512_mask(A, B, 1, D)
+#define __builtin_ia32_cmppd512_mask(A, B, F, D, E) __builtin_ia32_cmppd512_mask(A, B, 1, D, 5)
+#define __builtin_ia32_cmpps512_mask(A, B, F, D, E) __builtin_ia32_cmpps512_mask(A, B, 1, D, 5)
+#define __builtin_ia32_cmpq512_mask(A, B, E, D) __builtin_ia32_cmpq512_mask(A, B, 1, D)
+#define __builtin_ia32_cmpsd_mask(A, B, F, D, E) __builtin_ia32_cmpsd_mask(A, B, 1, D, 5)
+#define __builtin_ia32_cmpss_mask(A, B, F, D, E) __builtin_ia32_cmpss_mask(A, B, 1, D, 5)
+#define __builtin_ia32_cvtdq2ps512_mask(A, B, C, D) __builtin_ia32_cvtdq2ps512_mask(A, B, C, 1)
+#define __builtin_ia32_cvtpd2dq512_mask(A, B, C, D) __builtin_ia32_cvtpd2dq512_mask(A, B, C, 1)
+#define __builtin_ia32_cvtpd2ps512_mask(A, B, C, D) __builtin_ia32_cvtpd2ps512_mask(A, B, C, 1)
+#define __builtin_ia32_cvtpd2udq512_mask(A, B, C, D) __builtin_ia32_cvtpd2udq512_mask(A, B, C, 1)
+#define __builtin_ia32_cvtps2dq512_mask(A, B, C, D) __builtin_ia32_cvtps2dq512_mask(A, B, C, 1)
+#define __builtin_ia32_cvtps2pd512_mask(A, B, C, D) __builtin_ia32_cvtps2pd512_mask(A, B, C, 5)
+#define __builtin_ia32_cvtps2udq512_mask(A, B, C, D) __builtin_ia32_cvtps2udq512_mask(A, B, C, 1)
+#define __builtin_ia32_cvtsd2ss_mask(A, B, C, D, E) __builtin_ia32_cvtsd2ss_mask(A, B, C, D, 1)
+#define __builtin_ia32_cvtsi2sd64(A, B, C) __builtin_ia32_cvtsi2sd64(A, B, 1)
+#define __builtin_ia32_cvtsi2ss32(A, B, C) __builtin_ia32_cvtsi2ss32(A, B, 1)
+#define __builtin_ia32_cvtsi2ss64(A, B, C) __builtin_ia32_cvtsi2ss64(A, B, 1)
+#define __builtin_ia32_cvtss2sd_mask(A, B, C, D, E) __builtin_ia32_cvtss2sd_mask(A, B, C, D, 5)
+#define __builtin_ia32_cvttpd2dq512_mask(A, B, C, D) __builtin_ia32_cvttpd2dq512_mask(A, B, C, 5)
+#define __builtin_ia32_cvttpd2udq512_mask(A, B, C, D) __builtin_ia32_cvttpd2udq512_mask(A, B, C, 5)
+#define __builtin_ia32_cvttps2dq512_mask(A, B, C, D) __builtin_ia32_cvttps2dq512_mask(A, B, C, 5)
+#define __builtin_ia32_cvttps2udq512_mask(A, B, C, D) __builtin_ia32_cvttps2udq512_mask(A, B, C, 5)
+#define __builtin_ia32_cvtudq2ps512_mask(A, B, C, D) __builtin_ia32_cvtudq2ps512_mask(A, B, C, 1)
+#define __builtin_ia32_cvtusi2sd64(A, B, C) __builtin_ia32_cvtusi2sd64(A, B, 1)
+#define __builtin_ia32_cvtusi2ss32(A, B, C) __builtin_ia32_cvtusi2ss32(A, B, 1)
+#define __builtin_ia32_cvtusi2ss64(A, B, C) __builtin_ia32_cvtusi2ss64(A, B, 1)
+#define __builtin_ia32_divpd512_mask(A, B, C, D, E) __builtin_ia32_divpd512_mask(A, B, C, D, 1)
+#define __builtin_ia32_divps512_mask(A, B, C, D, E) __builtin_ia32_divps512_mask(A, B, C, D, 1)
+#define __builtin_ia32_divsd_mask(A, B, C, D, E) __builtin_ia32_divsd_mask(A, B, C, D, 1)
+#define __builtin_ia32_divss_mask(A, B, C, D, E) __builtin_ia32_divss_mask(A, B, C, D, 1)
+#define __builtin_ia32_extractf32x4_mask(A, E, C, D) __builtin_ia32_extractf32x4_mask(A, 1, C, D)
+#define __builtin_ia32_extractf64x4_mask(A, E, C, D) __builtin_ia32_extractf64x4_mask(A, 1, C, D)
+#define __builtin_ia32_extracti32x4_mask(A, E, C, D) __builtin_ia32_extracti32x4_mask(A, 1, C, D)
+#define __builtin_ia32_extracti64x4_mask(A, E, C, D) __builtin_ia32_extracti64x4_mask(A, 1, C, D)
+#define __builtin_ia32_fixupimmpd512_mask(A, B, C, I, E, F) __builtin_ia32_fixupimmpd512_mask(A, B, C, 1, E, 5)
+#define __builtin_ia32_fixupimmpd512_maskz(A, B, C, I, E, F) __builtin_ia32_fixupimmpd512_maskz(A, B, C, 1, E, 5)
+#define __builtin_ia32_fixupimmps512_mask(A, B, C, I, E, F) __builtin_ia32_fixupimmps512_mask(A, B, C, 1, E, 5)
+#define __builtin_ia32_fixupimmps512_maskz(A, B, C, I, E, F) __builtin_ia32_fixupimmps512_maskz(A, B, C, 1, E, 5)
+#define __builtin_ia32_fixupimmsd_mask(A, B, C, I, E, F) __builtin_ia32_fixupimmsd_mask(A, B, C, 1, E, 5)
+#define __builtin_ia32_fixupimmsd_maskz(A, B, C, I, E, F) __builtin_ia32_fixupimmsd_maskz(A, B, C, 1, E, 5)
+#define __builtin_ia32_fixupimmss_mask(A, B, C, I, E, F) __builtin_ia32_fixupimmss_mask(A, B, C, 1, E, 5)
+#define __builtin_ia32_fixupimmss_maskz(A, B, C, I, E, F) __builtin_ia32_fixupimmss_maskz(A, B, C, 1, E, 5)
+#define __builtin_ia32_gatherdiv8df(A, B, C, D, F) __builtin_ia32_gatherdiv8df(A, B, C, D, 1)
+#define __builtin_ia32_gatherdiv8di(A, B, C, D, F) __builtin_ia32_gatherdiv8di(A, B, C, D, 1)
+#define __builtin_ia32_gatherdiv16sf(A, B, C, D, F) __builtin_ia32_gatherdiv16sf(A, B, C, D, 1)
+#define __builtin_ia32_gatherdiv16si(A, B, C, D, F) __builtin_ia32_gatherdiv16si(A, B, C, D, 1)
+#define __builtin_ia32_gathersiv16sf(A, B, C, D, F) __builtin_ia32_gathersiv16sf(A, B, C, D, 1)
+#define __builtin_ia32_gathersiv16si(A, B, C, D, F) __builtin_ia32_gathersiv16si(A, B, C, D, 1)
+#define __builtin_ia32_gathersiv8df(A, B, C, D, F) __builtin_ia32_gathersiv8df(A, B, C, D, 1)
+#define __builtin_ia32_gathersiv8di(A, B, C, D, F) __builtin_ia32_gathersiv8di(A, B, C, D, 1)
+#define __builtin_ia32_getexppd512_mask(A, B, C, D) __builtin_ia32_getexppd512_mask(A, B, C, 5)
+#define __builtin_ia32_getexpps512_mask(A, B, C, D) __builtin_ia32_getexpps512_mask(A, B, C, 5)
+#define __builtin_ia32_getexpsd128_mask(A, B, C, D, E) __builtin_ia32_getexpsd128_mask(A, B, C, D, 5)
+#define __builtin_ia32_getexpss128_mask(A, B, C, D, E) __builtin_ia32_getexpss128_mask(A, B, C, D, 5)
+#define __builtin_ia32_getmantpd512_mask(A, F, C, D, E) __builtin_ia32_getmantpd512_mask(A, 1, C, D, 5)
+#define __builtin_ia32_getmantps512_mask(A, F, C, D, E) __builtin_ia32_getmantps512_mask(A, 1, C, D, 5)
+#define __builtin_ia32_getmantsd_mask(A, B, I, D, E, F) __builtin_ia32_getmantsd_mask(A, B, 1, D, E, 5)
+#define __builtin_ia32_getmantss_mask(A, B, I, D, E, F) __builtin_ia32_getmantss_mask(A, B, 1, D, E, 5)
+#define __builtin_ia32_insertf32x4_mask(A, B, F, D, E) __builtin_ia32_insertf32x4_mask(A, B, 1, D, E)
+#define __builtin_ia32_insertf64x4_mask(A, B, F, D, E) __builtin_ia32_insertf64x4_mask(A, B, 1, D, E)
+#define __builtin_ia32_inserti32x4_mask(A, B, F, D, E) __builtin_ia32_inserti32x4_mask(A, B, 1, D, E)
+#define __builtin_ia32_inserti64x4_mask(A, B, F, D, E) __builtin_ia32_inserti64x4_mask(A, B, 1, D, E)
+#define __builtin_ia32_maxpd512_mask(A, B, C, D, E) __builtin_ia32_maxpd512_mask(A, B, C, D, 5)
+#define __builtin_ia32_maxps512_mask(A, B, C, D, E) __builtin_ia32_maxps512_mask(A, B, C, D, 5)
+#define __builtin_ia32_maxsd_mask(A, B, C, D, E) __builtin_ia32_maxsd_mask(A, B, C, D, 5)
+#define __builtin_ia32_maxss_mask(A, B, C, D, E) __builtin_ia32_maxss_mask(A, B, C, D, 5)
+#define __builtin_ia32_minpd512_mask(A, B, C, D, E) __builtin_ia32_minpd512_mask(A, B, C, D, 5)
+#define __builtin_ia32_minps512_mask(A, B, C, D, E) __builtin_ia32_minps512_mask(A, B, C, D, 5)
+#define __builtin_ia32_minsd_mask(A, B, C, D, E) __builtin_ia32_minsd_mask(A, B, C, D, 5)
+#define __builtin_ia32_minss_mask(A, B, C, D, E) __builtin_ia32_minss_mask(A, B, C, D, 5)
+#define __builtin_ia32_mulpd512_mask(A, B, C, D, E) __builtin_ia32_mulpd512_mask(A, B, C, D, 1)
+#define __builtin_ia32_mulps512_mask(A, B, C, D, E) __builtin_ia32_mulps512_mask(A, B, C, D, 1)
+#define __builtin_ia32_mulsd_mask(A, B, C, D, E) __builtin_ia32_mulsd_mask(A, B, C, D, 1)
+#define __builtin_ia32_mulss_mask(A, B, C, D, E) __builtin_ia32_mulss_mask(A, B, C, D, 1)
+#define __builtin_ia32_permdf512_mask(A, E, C, D) __builtin_ia32_permdf512_mask(A, 1, C, D)
+#define __builtin_ia32_permdi512_mask(A, E, C, D) __builtin_ia32_permdi512_mask(A, 1, C, D)
+#define __builtin_ia32_prold512_mask(A, E, C, D) __builtin_ia32_prold512_mask(A, 1, C, D)
+#define __builtin_ia32_prolq512_mask(A, E, C, D) __builtin_ia32_prolq512_mask(A, 1, C, D)
+#define __builtin_ia32_prord512_mask(A, E, C, D) __builtin_ia32_prord512_mask(A, 1, C, D)
+#define __builtin_ia32_prorq512_mask(A, E, C, D) __builtin_ia32_prorq512_mask(A, 1, C, D)
+#define __builtin_ia32_pshufd512_mask(A, E, C, D) __builtin_ia32_pshufd512_mask(A, 1, C, D)
+#define __builtin_ia32_pslldi512_mask(A, E, C, D) __builtin_ia32_pslldi512_mask(A, 1, C, D)
+#define __builtin_ia32_psllqi512_mask(A, E, C, D) __builtin_ia32_psllqi512_mask(A, 1, C, D)
+#define __builtin_ia32_psradi512_mask(A, E, C, D) __builtin_ia32_psradi512_mask(A, 1, C, D)
+#define __builtin_ia32_psraqi512_mask(A, E, C, D) __builtin_ia32_psraqi512_mask(A, 1, C, D)
+#define __builtin_ia32_psrldi512_mask(A, E, C, D) __builtin_ia32_psrldi512_mask(A, 1, C, D)
+#define __builtin_ia32_psrlqi512_mask(A, E, C, D) __builtin_ia32_psrlqi512_mask(A, 1, C, D)
+#define __builtin_ia32_pternlogd512_mask(A, B, C, F, E) __builtin_ia32_pternlogd512_mask(A, B, C, 1, E)
+#define __builtin_ia32_pternlogd512_maskz(A, B, C, F, E) __builtin_ia32_pternlogd512_maskz(A, B, C, 1, E)
+#define __builtin_ia32_pternlogq512_mask(A, B, C, F, E) __builtin_ia32_pternlogq512_mask(A, B, C, 1, E)
+#define __builtin_ia32_pternlogq512_maskz(A, B, C, F, E) __builtin_ia32_pternlogq512_maskz(A, B, C, 1, E)
+#define __builtin_ia32_rndscalepd_mask(A, F, C, D, E) __builtin_ia32_rndscalepd_mask(A, 1, C, D, 5)
+#define __builtin_ia32_rndscaleps_mask(A, F, C, D, E) __builtin_ia32_rndscaleps_mask(A, 1, C, D, 5)
+#define __builtin_ia32_rndscalesd_mask(A, B, I, D, E, F) __builtin_ia32_rndscalesd_mask(A, B, 1, D, E, 5)
+#define __builtin_ia32_rndscaless_mask(A, B, I, D, E, F) __builtin_ia32_rndscaless_mask(A, B, 1, D, E, 5)
+#define __builtin_ia32_scalefpd512_mask(A, B, C, D, E) __builtin_ia32_scalefpd512_mask(A, B, C, D, 1)
+#define __builtin_ia32_scalefps512_mask(A, B, C, D, E) __builtin_ia32_scalefps512_mask(A, B, C, D, 1)
+#define __builtin_ia32_scalefsd_mask(A, B, C, D, E) __builtin_ia32_scalefsd_mask(A, B, C, D, 1)
+#define __builtin_ia32_scalefss_mask(A, B, C, D, E) __builtin_ia32_scalefss_mask(A, B, C, D, 1)
+#define __builtin_ia32_scatterdiv8df(A, B, C, D, F) __builtin_ia32_scatterdiv8df(A, B, C, D, 1)
+#define __builtin_ia32_scatterdiv8di(A, B, C, D, F) __builtin_ia32_scatterdiv8di(A, B, C, D, 1)
+#define __builtin_ia32_scatterdiv16sf(A, B, C, D, F) __builtin_ia32_scatterdiv16sf(A, B, C, D, 1)
+#define __builtin_ia32_scatterdiv16si(A, B, C, D, F) __builtin_ia32_scatterdiv16si(A, B, C, D, 1)
+#define __builtin_ia32_scattersiv16sf(A, B, C, D, F) __builtin_ia32_scattersiv16sf(A, B, C, D, 1)
+#define __builtin_ia32_scattersiv16si(A, B, C, D, F) __builtin_ia32_scattersiv16si(A, B, C, D, 1)
+#define __builtin_ia32_scattersiv8df(A, B, C, D, F) __builtin_ia32_scattersiv8df(A, B, C, D, 1)
+#define __builtin_ia32_scattersiv8di(A, B, C, D, F) __builtin_ia32_scattersiv8di(A, B, C, D, 1)
+#define __builtin_ia32_shuf_f32x4_mask(A, B, F, D, E) __builtin_ia32_shuf_f32x4_mask(A, B, 1, D, E)
+#define __builtin_ia32_shuf_f64x2_mask(A, B, F, D, E) __builtin_ia32_shuf_f64x2_mask(A, B, 1, D, E)
+#define __builtin_ia32_shuf_i32x4_mask(A, B, F, D, E) __builtin_ia32_shuf_i32x4_mask(A, B, 1, D, E)
+#define __builtin_ia32_shuf_i64x2_mask(A, B, F, D, E) __builtin_ia32_shuf_i64x2_mask(A, B, 1, D, E)
+#define __builtin_ia32_shufpd512_mask(A, B, F, D, E) __builtin_ia32_shufpd512_mask(A, B, 1, D, E)
+#define __builtin_ia32_shufps512_mask(A, B, F, D, E) __builtin_ia32_shufps512_mask(A, B, 1, D, E)
+#define __builtin_ia32_sqrtpd512_mask(A, B, C, D) __builtin_ia32_sqrtpd512_mask(A, B, C, 1)
+#define __builtin_ia32_sqrtps512_mask(A, B, C, D) __builtin_ia32_sqrtps512_mask(A, B, C, 1)
+#define __builtin_ia32_sqrtsd_mask(A, B, C, D, E) __builtin_ia32_sqrtsd_mask(A, B, C, D, 1)
+#define __builtin_ia32_sqrtss_mask(A, B, C, D, E) __builtin_ia32_sqrtss_mask(A, B, C, D, 1)
+#define __builtin_ia32_subpd512_mask(A, B, C, D, E) __builtin_ia32_subpd512_mask(A, B, C, D, 1)
+#define __builtin_ia32_subps512_mask(A, B, C, D, E) __builtin_ia32_subps512_mask(A, B, C, D, 1)
+#define __builtin_ia32_subsd_mask(A, B, C, D, E) __builtin_ia32_subsd_mask(A, B, C, D, 1)
+#define __builtin_ia32_subss_mask(A, B, C, D, E) __builtin_ia32_subss_mask(A, B, C, D, 1)
+#define __builtin_ia32_ucmpd512_mask(A, B, E, D) __builtin_ia32_ucmpd512_mask(A, B, 1, D)
+#define __builtin_ia32_ucmpq512_mask(A, B, E, D) __builtin_ia32_ucmpq512_mask(A, B, 1, D)
+#define __builtin_ia32_vcomisd(A, B, C, D) __builtin_ia32_vcomisd(A, B, 1, 5)
+#define __builtin_ia32_vcomiss(A, B, C, D) __builtin_ia32_vcomiss(A, B, 1, 5)
+#define __builtin_ia32_vcvtph2ps512_mask(A, B, C, D) __builtin_ia32_vcvtph2ps512_mask(A, B, C, 5)
+#define __builtin_ia32_vcvtps2ph512_mask(A, E, C, D) __builtin_ia32_vcvtps2ph512_mask(A, 1, C, D)
+#define __builtin_ia32_vcvtsd2si32(A, B) __builtin_ia32_vcvtsd2si32(A, 1)
+#define __builtin_ia32_vcvtsd2si64(A, B) __builtin_ia32_vcvtsd2si64(A, 1)
+#define __builtin_ia32_vcvtsd2usi32(A, B) __builtin_ia32_vcvtsd2usi32(A, 1)
+#define __builtin_ia32_vcvtsd2usi64(A, B) __builtin_ia32_vcvtsd2usi64(A, 1)
+#define __builtin_ia32_vcvtss2si32(A, B) __builtin_ia32_vcvtss2si32(A, 1)
+#define __builtin_ia32_vcvtss2si64(A, B) __builtin_ia32_vcvtss2si64(A, 1)
+#define __builtin_ia32_vcvtss2usi32(A, B) __builtin_ia32_vcvtss2usi32(A, 1)
+#define __builtin_ia32_vcvtss2usi64(A, B) __builtin_ia32_vcvtss2usi64(A, 1)
+#define __builtin_ia32_vcvttsd2si32(A, B) __builtin_ia32_vcvttsd2si32(A, 5)
+#define __builtin_ia32_vcvttsd2si64(A, B) __builtin_ia32_vcvttsd2si64(A, 5)
+#define __builtin_ia32_vcvttsd2usi32(A, B) __builtin_ia32_vcvttsd2usi32(A, 5)
+#define __builtin_ia32_vcvttsd2usi64(A, B) __builtin_ia32_vcvttsd2usi64(A, 5)
+#define __builtin_ia32_vcvttss2si32(A, B) __builtin_ia32_vcvttss2si32(A, 5)
+#define __builtin_ia32_vcvttss2si64(A, B) __builtin_ia32_vcvttss2si64(A, 5)
+#define __builtin_ia32_vcvttss2usi32(A, B) __builtin_ia32_vcvttss2usi32(A, 5)
+#define __builtin_ia32_vcvttss2usi64(A, B) __builtin_ia32_vcvttss2usi64(A, 5)
+#define __builtin_ia32_vfmaddpd512_mask(A, B, C, D, E) __builtin_ia32_vfmaddpd512_mask(A, B, C, D, 1)
+#define __builtin_ia32_vfmaddpd512_mask3(A, B, C, D, E) __builtin_ia32_vfmaddpd512_mask3(A, B, C, D, 1)
+#define __builtin_ia32_vfmaddpd512_maskz(A, B, C, D, E) __builtin_ia32_vfmaddpd512_maskz(A, B, C, D, 1)
+#define __builtin_ia32_vfmaddps512_mask(A, B, C, D, E) __builtin_ia32_vfmaddps512_mask(A, B, C, D, 1)
+#define __builtin_ia32_vfmaddps512_mask3(A, B, C, D, E) __builtin_ia32_vfmaddps512_mask3(A, B, C, D, 1)
+#define __builtin_ia32_vfmaddps512_maskz(A, B, C, D, E) __builtin_ia32_vfmaddps512_maskz(A, B, C, D, 1)
+#define __builtin_ia32_vfmaddsd3_mask(A, B, C, D, E) __builtin_ia32_vfmaddsd3_mask(A, B, C, D, 1)
+#define __builtin_ia32_vfmaddsd3_mask3(A, B, C, D, E) __builtin_ia32_vfmaddsd3_mask3(A, B, C, D, 1)
+#define __builtin_ia32_vfmaddsd3_maskz(A, B, C, D, E) __builtin_ia32_vfmaddsd3_maskz(A, B, C, D, 1)
+#define __builtin_ia32_vfmaddss3_mask(A, B, C, D, E) __builtin_ia32_vfmaddss3_mask(A, B, C, D, 1)
+#define __builtin_ia32_vfmaddss3_mask3(A, B, C, D, E) __builtin_ia32_vfmaddss3_mask3(A, B, C, D, 1)
+#define __builtin_ia32_vfmaddss3_maskz(A, B, C, D, E) __builtin_ia32_vfmaddss3_maskz(A, B, C, D, 1)
+#define __builtin_ia32_vfmaddsubpd512_mask(A, B, C, D, E) __builtin_ia32_vfmaddsubpd512_mask(A, B, C, D, 1)
+#define __builtin_ia32_vfmaddsubpd512_mask3(A, B, C, D, E) __builtin_ia32_vfmaddsubpd512_mask3(A, B, C, D, 1)
+#define __builtin_ia32_vfmaddsubpd512_maskz(A, B, C, D, E) __builtin_ia32_vfmaddsubpd512_maskz(A, B, C, D, 1)
+#define __builtin_ia32_vfmaddsubps512_mask(A, B, C, D, E) __builtin_ia32_vfmaddsubps512_mask(A, B, C, D, 1)
+#define __builtin_ia32_vfmaddsubps512_mask3(A, B, C, D, E) __builtin_ia32_vfmaddsubps512_mask3(A, B, C, D, 1)
+#define __builtin_ia32_vfmaddsubps512_maskz(A, B, C, D, E) __builtin_ia32_vfmaddsubps512_maskz(A, B, C, D, 1)
+#define __builtin_ia32_vfmsubaddpd512_mask3(A, B, C, D, E) __builtin_ia32_vfmsubaddpd512_mask3(A, B, C, D, 1)
+#define __builtin_ia32_vfmsubaddps512_mask3(A, B, C, D, E) __builtin_ia32_vfmsubaddps512_mask3(A, B, C, D, 1)
+#define __builtin_ia32_vfmsubpd512_mask3(A, B, C, D, E) __builtin_ia32_vfmsubpd512_mask3(A, B, C, D, 1)
+#define __builtin_ia32_vfmsubps512_mask3(A, B, C, D, E) __builtin_ia32_vfmsubps512_mask3(A, B, C, D, 1)
+#define __builtin_ia32_vfmsubsd3_mask3(A, B, C, D, E) __builtin_ia32_vfmsubsd3_mask3(A, B, C, D, 1)
+#define __builtin_ia32_vfmsubss3_mask3(A, B, C, D, E) __builtin_ia32_vfmsubss3_mask3(A, B, C, D, 1)
+#define __builtin_ia32_vfnmaddpd512_mask(A, B, C, D, E) __builtin_ia32_vfnmaddpd512_mask(A, B, C, D, 1)
+#define __builtin_ia32_vfnmaddps512_mask(A, B, C, D, E) __builtin_ia32_vfnmaddps512_mask(A, B, C, D, 1)
+#define __builtin_ia32_vfnmsubpd512_mask(A, B, C, D, E) __builtin_ia32_vfnmsubpd512_mask(A, B, C, D, 1)
+#define __builtin_ia32_vfnmsubpd512_mask3(A, B, C, D, E) __builtin_ia32_vfnmsubpd512_mask3(A, B, C, D, 1)
+#define __builtin_ia32_vfnmsubps512_mask(A, B, C, D, E) __builtin_ia32_vfnmsubps512_mask(A, B, C, D, 1)
+#define __builtin_ia32_vfnmsubps512_mask3(A, B, C, D, E) __builtin_ia32_vfnmsubps512_mask3(A, B, C, D, 1)
+#define __builtin_ia32_vpermilpd512_mask(A, E, C, D) __builtin_ia32_vpermilpd512_mask(A, 1, C, D)
+#define __builtin_ia32_vpermilps512_mask(A, E, C, D) __builtin_ia32_vpermilps512_mask(A, 1, C, D)
/* { dg-do compile } */
-/* { dg-options "-O0 -Werror-implicit-function-declaration -march=k8 -msse4a -m3dnow -mavx -mavx2 -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi -mbmi2 -mtbm -mlwp -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -mprfchw -madx -mfxsr -mxsaveopt" } */
-
+/* { dg-options "-O0 -Werror-implicit-function-declaration -march=k8 -msse4a -m3dnow -mavx -mavx2 -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi -mbmi2 -mtbm -mlwp -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -mprfchw -madx -mfxsr -mxsaveopt -mavx512f -mavx512er -mavx512pf -mavx512cd" } */
#include <mm_malloc.h>
/* Test that the intrinsics compile without optimization. All of them are
type _CONCAT(_,func) (op1_type A, int const I, int const L) \
{ return func (A, imm1, imm2); }
+#define test_1y(func, type, op1_type, imm1, imm2, imm3) \
+ type _CONCAT(_,func) (op1_type A, int const I, int const L, int const R)\
+ { return func (A, imm1, imm2, imm3); }
+
#define test_2(func, type, op1_type, op2_type, imm) \
type _CONCAT(_,func) (op1_type A, op2_type B, int const I) \
{ return func (A, B, imm); }
type _CONCAT(_,func) (op1_type A, op2_type B, int const I, int const L) \
{ return func (A, B, imm1, imm2); }
+#define test_2y(func, type, op1_type, op2_type, imm1, imm2, imm3) \
+ type _CONCAT(_,func) (op1_type A, op2_type B, int const I, int const L,\
+ int const R) \
+ { return func (A, B, imm1, imm2, imm3); }
+
+#define test_2vx(func, op1_type, op2_type, imm1, imm2) \
+ _CONCAT(_,func) (op1_type A, op2_type B, int const I, int const L) \
+ { func (A, B, imm1, imm2); }
+
#define test_3(func, type, op1_type, op2_type, op3_type, imm) \
type _CONCAT(_,func) (op1_type A, op2_type B, \
op3_type C, int const I) \
{ return func (A, B, C, imm); }
+#define test_3x(func, type, op1_type, op2_type, op3_type, imm1, imm2) \
+ type _CONCAT(_,func) (op1_type A, op2_type B, \
+ op3_type C, int const I, int const L) \
+ { return func (A, B, C, imm1, imm2); }
+
+#define test_3y(func, type, op1_type, op2_type, op3_type, imm1, imm2, imm3) \
+ type _CONCAT(_,func) (op1_type A, op2_type B, \
+ op3_type C, int const I, int const L, int const R) \
+ { return func (A, B, C, imm1, imm2, imm3); }
+
+#define test_3v(func, op1_type, op2_type, op3_type, imm) \
+ _CONCAT(_,func) (op1_type A, op2_type B, \
+ op3_type C, int const I) \
+ { func (A, B, C, imm); }
+
+#define test_3vx(func, op1_type, op2_type, op3_type, imm1, imm2) \
+ _CONCAT(_,func) (op1_type A, op2_type B, \
+ op3_type C, int const I, int const L) \
+ { func (A, B, C, imm1, imm2); }
+
#define test_4(func, type, op1_type, op2_type, op3_type, op4_type, imm) \
type _CONCAT(_,func) (op1_type A, op2_type B, \
op3_type C, op4_type D, int const I) \
{ return func (A, B, C, D, imm); }
+#define test_4x(func, type, op1_type, op2_type, op3_type, op4_type, imm1, imm2) \
+ type _CONCAT(_,func) (op1_type A, op2_type B, \
+ op3_type C, op4_type D, int const I, int const L) \
+ { return func (A, B, C, D, imm1, imm2); }
+
+#define test_4y(func, type, op1_type, op2_type, op3_type, op4_type, imm1, imm2, imm3) \
+ type _CONCAT(_,func) (op1_type A, op2_type B, op3_type C, \
+ op4_type D, int const I, int const L, int const R) \
+ { return func (A, B, C, D, imm1, imm2, imm3); }
+
+#define test_4v(func, op1_type, op2_type, op3_type, op4_type, imm) \
+ _CONCAT(_,func) (op1_type A, op2_type B, \
+ op3_type C, op4_type D, int const I) \
+ { func (A, B, C, D, imm); }
+
/* Following intrinsics require immediate arguments. They
are defined as macros for non-optimized compilations. */
test_1 (_mm_cvtps_ph, __m128i, __m128, 1)
test_1 (_mm256_cvtps_ph, __m128i, __m256, 1)
test_0 (_xabort, void, 1)
+test_1 (_mm512_cvt_roundepi32_ps, __m512, __m512i, 1)
+test_1 (_mm512_cvt_roundepu32_ps, __m512, __m512i, 1)
+test_1 (_mm512_cvt_roundpd_epi32, __m256i, __m512d, 1)
+test_1 (_mm512_cvt_roundpd_epu32, __m256i, __m512d, 1)
+test_1 (_mm512_cvt_roundpd_ps, __m256, __m512d, 1)
+test_1 (_mm512_cvt_roundph_ps, __m512, __m256i, 5)
+test_1 (_mm512_cvt_roundps_epi32, __m512i, __m512, 1)
+test_1 (_mm512_cvt_roundps_epu32, __m512i, __m512, 1)
+test_1 (_mm512_cvt_roundps_pd, __m512d, __m256, 5)
+test_1 (_mm512_cvtps_ph, __m256i, __m512, 1)
+test_1 (_mm512_cvtt_roundpd_epi32, __m256i, __m512d, 5)
+test_1 (_mm512_cvtt_roundpd_epu32, __m256i, __m512d, 5)
+test_1 (_mm512_cvtt_roundps_epi32, __m512i, __m512, 5)
+test_1 (_mm512_cvtt_roundps_epu32, __m512i, __m512, 5)
+test_1 (_mm512_extractf32x4_ps, __m128, __m512, 1)
+test_1 (_mm512_extractf64x4_pd, __m256d, __m512d, 1)
+test_1 (_mm512_extracti32x4_epi32, __m128i, __m512i, 1)
+test_1 (_mm512_extracti64x4_epi64, __m256i, __m512i, 1)
+test_1 (_mm512_getexp_round_pd, __m512d, __m512d, 5)
+test_1 (_mm512_getexp_round_ps, __m512, __m512, 5)
+test_1y (_mm512_getmant_round_pd, __m512d, __m512d, 1, 1, 5)
+test_1y (_mm512_getmant_round_ps, __m512, __m512, 1, 1, 5)
+test_1 (_mm512_permute_pd, __m512d, __m512d, 1)
+test_1 (_mm512_permute_ps, __m512, __m512, 1)
+test_1 (_mm512_permutex_epi64, __m512i, __m512i, 1)
+test_1 (_mm512_permutex_pd, __m512d, __m512d, 1)
+test_1 (_mm512_rol_epi32, __m512i, __m512i, 1)
+test_1 (_mm512_rol_epi64, __m512i, __m512i, 1)
+test_1 (_mm512_ror_epi32, __m512i, __m512i, 1)
+test_1 (_mm512_ror_epi64, __m512i, __m512i, 1)
+test_1 (_mm512_shuffle_epi32, __m512i, __m512i, 1)
+test_1 (_mm512_slli_epi32, __m512i, __m512i, 1)
+test_1 (_mm512_slli_epi64, __m512i, __m512i, 1)
+test_1 (_mm512_sqrt_round_pd, __m512d, __m512d, 1)
+test_1 (_mm512_sqrt_round_ps, __m512, __m512, 1)
+test_1 (_mm512_srai_epi32, __m512i, __m512i, 1)
+test_1 (_mm512_srai_epi64, __m512i, __m512i, 1)
+test_1 (_mm512_srli_epi32, __m512i, __m512i, 1)
+test_1 (_mm512_srli_epi64, __m512i, __m512i, 1)
+test_1 (_mm_cvt_roundsd_i32, int, __m128d, 1)
+test_1 (_mm_cvt_roundsd_u32, unsigned, __m128d, 1)
+test_1 (_mm_cvt_roundss_i32, int, __m128, 1)
+test_1 (_mm_cvt_roundss_u32, unsigned, __m128, 1)
+test_1 (_mm_cvtt_roundsd_i32, int, __m128d, 5)
+test_1 (_mm_cvtt_roundsd_u32, unsigned, __m128d, 5)
+test_1 (_mm_cvtt_roundss_i32, int, __m128, 5)
+test_1 (_mm_cvtt_roundss_u32, unsigned, __m128, 5)
+test_1x (_mm512_getmant_pd, __m512d, __m512d, 1, 1)
+test_1x (_mm512_getmant_ps, __m512, __m512, 1, 1)
+test_1x (_mm512_roundscale_round_pd, __m512d, __m512d, 1, 5)
+test_1x (_mm512_roundscale_round_ps, __m512, __m512, 1, 5)
+test_2 (_mm512_add_round_pd, __m512d, __m512d, __m512d, 1)
+test_2 (_mm512_add_round_ps, __m512, __m512, __m512, 1)
+test_2 (_mm512_alignr_epi32, __m512i, __m512i, __m512i, 1)
+test_2 (_mm512_alignr_epi64, __m512i, __m512i, __m512i, 1)
+test_2 (_mm512_cmp_epi32_mask, __mmask16, __m512i, __m512i, 1)
+test_2 (_mm512_cmp_epi64_mask, __mmask8, __m512i, __m512i, 1)
+test_2 (_mm512_cmp_epu32_mask, __mmask16, __m512i, __m512i, 1)
+test_2 (_mm512_cmp_epu64_mask, __mmask8, __m512i, __m512i, 1)
+test_2 (_mm512_cmp_pd_mask, __mmask8, __m512d, __m512d, 1)
+test_2 (_mm512_cmp_ps_mask, __mmask16, __m512, __m512, 1)
+test_2 (_mm512_div_round_pd, __m512d, __m512d, __m512d, 1)
+test_2 (_mm512_div_round_ps, __m512, __m512, __m512, 1)
+test_2 (_mm512_i32gather_epi32, __m512i, __m512i, void const *, 1)
+test_2 (_mm512_i32gather_epi64, __m512i, __m256i, void const *, 1)
+test_2 (_mm512_i32gather_pd, __m512d, __m256i, void const *, 1)
+test_2 (_mm512_i32gather_ps, __m512, __m512i, void const *, 1)
+test_2 (_mm512_i64gather_epi32, __m256i, __m512i, void const *, 1)
+test_2 (_mm512_i64gather_epi64, __m512i, __m512i, void const *, 1)
+test_2 (_mm512_i64gather_pd, __m512d, __m512i, void const *, 1)
+test_2 (_mm512_i64gather_ps, __m256, __m512i, void const *, 1)
+test_2 (_mm512_insertf32x4, __m512, __m512, __m128, 1)
+test_2 (_mm512_insertf64x4, __m512d, __m512d, __m256d, 1)
+test_2 (_mm512_inserti32x4, __m512i, __m512i, __m128i, 1)
+test_2 (_mm512_inserti64x4, __m512i, __m512i, __m256i, 1)
+test_2 (_mm512_maskz_cvt_roundepi32_ps, __m512, __mmask16, __m512i, 1)
+test_2 (_mm512_maskz_cvt_roundepu32_ps, __m512, __mmask16, __m512i, 1)
+test_2 (_mm512_maskz_cvt_roundpd_epi32, __m256i, __mmask8, __m512d, 1)
+test_2 (_mm512_maskz_cvt_roundpd_epu32, __m256i, __mmask8, __m512d, 1)
+test_2 (_mm512_maskz_cvt_roundpd_ps, __m256, __mmask8, __m512d, 1)
+test_2 (_mm512_maskz_cvt_roundph_ps, __m512, __mmask16, __m256i, 5)
+test_2 (_mm512_maskz_cvt_roundps_epi32, __m512i, __mmask16, __m512, 1)
+test_2 (_mm512_maskz_cvt_roundps_epu32, __m512i, __mmask16, __m512, 1)
+test_2 (_mm512_maskz_cvt_roundps_pd, __m512d, __mmask8, __m256, 5)
+test_2 (_mm512_maskz_cvtps_ph, __m256i, __mmask16, __m512, 1)
+test_2 (_mm512_maskz_cvtt_roundpd_epi32, __m256i, __mmask8, __m512d, 5)
+test_2 (_mm512_maskz_cvtt_roundpd_epu32, __m256i, __mmask8, __m512d, 5)
+test_2 (_mm512_maskz_cvtt_roundps_epi32, __m512i, __mmask16, __m512, 5)
+test_2 (_mm512_maskz_cvtt_roundps_epu32, __m512i, __mmask16, __m512, 5)
+test_2 (_mm512_maskz_extractf32x4_ps, __m128, __mmask8, __m512, 1)
+test_2 (_mm512_maskz_extractf64x4_pd, __m256d, __mmask8, __m512d, 1)
+test_2 (_mm512_maskz_extracti32x4_epi32, __m128i, __mmask8, __m512i, 1)
+test_2 (_mm512_maskz_extracti64x4_epi64, __m256i, __mmask8, __m512i, 1)
+test_2 (_mm512_maskz_getexp_round_pd, __m512d, __mmask8, __m512d, 5)
+test_2 (_mm512_maskz_getexp_round_ps, __m512, __mmask16, __m512, 5)
+test_2y (_mm512_maskz_getmant_round_pd, __m512d, __mmask8, __m512d, 1, 1, 5)
+test_2y (_mm512_maskz_getmant_round_ps, __m512, __mmask16, __m512, 1, 1, 5)
+test_2 (_mm512_maskz_permute_pd, __m512d, __mmask8, __m512d, 1)
+test_2 (_mm512_maskz_permute_ps, __m512, __mmask16, __m512, 1)
+test_2 (_mm512_maskz_permutex_epi64, __m512i, __mmask8, __m512i, 1)
+test_2 (_mm512_maskz_permutex_pd, __m512d, __mmask8, __m512d, 1)
+test_2 (_mm512_maskz_rol_epi32, __m512i, __mmask16, __m512i, 1)
+test_2 (_mm512_maskz_rol_epi64, __m512i, __mmask8, __m512i, 1)
+test_2 (_mm512_maskz_ror_epi32, __m512i, __mmask16, __m512i, 1)
+test_2 (_mm512_maskz_ror_epi64, __m512i, __mmask8, __m512i, 1)
+test_2 (_mm512_maskz_shuffle_epi32, __m512i, __mmask16, __m512i, 1)
+test_2 (_mm512_maskz_slli_epi32, __m512i, __mmask16, __m512i, 1)
+test_2 (_mm512_maskz_slli_epi64, __m512i, __mmask8, __m512i, 1)
+test_2 (_mm512_maskz_sqrt_round_pd, __m512d, __mmask8, __m512d, 1)
+test_2 (_mm512_maskz_sqrt_round_ps, __m512, __mmask16, __m512, 1)
+test_2 (_mm512_maskz_srai_epi32, __m512i, __mmask16, __m512i, 1)
+test_2 (_mm512_maskz_srai_epi64, __m512i, __mmask8, __m512i, 1)
+test_2 (_mm512_maskz_srli_epi32, __m512i, __mmask16, __m512i, 1)
+test_2 (_mm512_maskz_srli_epi64, __m512i, __mmask8, __m512i, 1)
+test_2 (_mm512_max_round_pd, __m512d, __m512d, __m512d, 5)
+test_2 (_mm512_max_round_ps, __m512, __m512, __m512, 5)
+test_2 (_mm512_min_round_pd, __m512d, __m512d, __m512d, 5)
+test_2 (_mm512_min_round_ps, __m512, __m512, __m512, 5)
+test_2 (_mm512_mul_round_pd, __m512d, __m512d, __m512d, 1)
+test_2 (_mm512_mul_round_ps, __m512, __m512, __m512, 1)
+test_2 (_mm512_scalef_round_pd, __m512d, __m512d, __m512d, 1)
+test_2 (_mm512_scalef_round_ps, __m512, __m512, __m512, 1)
+test_2 (_mm512_shuffle_f32x4, __m512, __m512, __m512, 1)
+test_2 (_mm512_shuffle_f64x2, __m512d, __m512d, __m512d, 1)
+test_2 (_mm512_shuffle_i32x4, __m512i, __m512i, __m512i, 1)
+test_2 (_mm512_shuffle_i64x2, __m512i, __m512i, __m512i, 1)
+test_2 (_mm512_shuffle_pd, __m512d, __m512d, __m512d, 1)
+test_2 (_mm512_shuffle_ps, __m512, __m512, __m512, 1)
+test_2 (_mm512_sub_round_pd, __m512d, __m512d, __m512d, 1)
+test_2 (_mm512_sub_round_ps, __m512, __m512, __m512, 1)
+test_2 (_mm_cmp_sd_mask, __mmask8, __m128d, __m128d, 1)
+test_2 (_mm_cmp_ss_mask, __mmask8, __m128, __m128, 1)
+#ifdef __x86_64__
+#endif
+#ifdef __x86_64__
+#endif
+test_2x (_mm512_cmp_round_pd_mask, __mmask8, __m512d, __m512d, 1, 5)
+test_2x (_mm512_cmp_round_ps_mask, __mmask16, __m512, __m512, 1, 5)
+test_2x (_mm512_maskz_roundscale_round_pd, __m512d, __mmask8, __m512d, 1, 5)
+test_2x (_mm512_maskz_roundscale_round_ps, __m512, __mmask16, __m512, 1, 5)
+test_3 (_mm512_fmadd_round_pd, __m512d, __m512d, __m512d, __m512d, 1)
+test_3 (_mm512_fmadd_round_ps, __m512, __m512, __m512, __m512, 1)
+test_3 (_mm512_fmaddsub_round_pd, __m512d, __m512d, __m512d, __m512d, 1)
+test_3 (_mm512_fmaddsub_round_ps, __m512, __m512, __m512, __m512, 1)
+test_3 (_mm512_fmsub_round_pd, __m512d, __m512d, __m512d, __m512d, 1)
+test_3 (_mm512_fmsub_round_ps, __m512, __m512, __m512, __m512, 1)
+test_3 (_mm512_fmsubadd_round_pd, __m512d, __m512d, __m512d, __m512d, 1)
+test_3 (_mm512_fmsubadd_round_ps, __m512, __m512, __m512, __m512, 1)
+test_3 (_mm512_fnmadd_round_pd, __m512d, __m512d, __m512d, __m512d, 1)
+test_3 (_mm512_fnmadd_round_ps, __m512, __m512, __m512, __m512, 1)
+test_3 (_mm512_fnmsub_round_pd, __m512d, __m512d, __m512d, __m512d, 1)
+test_3 (_mm512_fnmsub_round_ps, __m512, __m512, __m512, __m512, 1)
+test_3 (_mm512_mask_cmp_epi32_mask, __mmask16, __mmask16, __m512i, __m512i, 1)
+test_3 (_mm512_mask_cmp_epi64_mask, __mmask8, __mmask8, __m512i, __m512i, 1)
+test_3 (_mm512_mask_cmp_epu32_mask, __mmask16, __mmask16, __m512i, __m512i, 1)
+test_3 (_mm512_mask_cmp_epu64_mask, __mmask8, __mmask8, __m512i, __m512i, 1)
+test_3 (_mm512_mask_cmp_pd_mask, __mmask8, __mmask8, __m512d, __m512d, 1)
+test_3 (_mm512_mask_cmp_ps_mask, __mmask16, __mmask16, __m512, __m512, 1)
+test_3 (_mm512_mask_cvt_roundepi32_ps, __m512, __m512, __mmask16, __m512i, 1)
+test_3 (_mm512_mask_cvt_roundepu32_ps, __m512, __m512, __mmask16, __m512i, 1)
+test_3 (_mm512_mask_cvt_roundpd_epi32, __m256i, __m256i, __mmask8, __m512d, 1)
+test_3 (_mm512_mask_cvt_roundpd_epu32, __m256i, __m256i, __mmask8, __m512d, 1)
+test_3 (_mm512_mask_cvt_roundpd_ps, __m256, __m256, __mmask8, __m512d, 1)
+test_3 (_mm512_mask_cvt_roundph_ps, __m512, __m512, __mmask16, __m256i, 5)
+test_3 (_mm512_mask_cvt_roundps_epi32, __m512i, __m512i, __mmask16, __m512, 1)
+test_3 (_mm512_mask_cvt_roundps_epu32, __m512i, __m512i, __mmask16, __m512, 1)
+test_3 (_mm512_mask_cvt_roundps_pd, __m512d, __m512d, __mmask8, __m256, 5)
+test_3 (_mm512_mask_cvtps_ph, __m256i, __m256i, __mmask16, __m512, 1)
+test_3 (_mm512_mask_cvtt_roundpd_epi32, __m256i, __m256i, __mmask8, __m512d, 5)
+test_3 (_mm512_mask_cvtt_roundpd_epu32, __m256i, __m256i, __mmask8, __m512d, 5)
+test_3 (_mm512_mask_cvtt_roundps_epi32, __m512i, __m512i, __mmask16, __m512, 5)
+test_3 (_mm512_mask_cvtt_roundps_epu32, __m512i, __m512i, __mmask16, __m512, 5)
+test_3 (_mm512_mask_extractf32x4_ps, __m128, __m128, __mmask8, __m512, 1)
+test_3 (_mm512_mask_extractf64x4_pd, __m256d, __m256d, __mmask8, __m512d, 1)
+test_3 (_mm512_mask_extracti32x4_epi32, __m128i, __m128i, __mmask8, __m512i, 1)
+test_3 (_mm512_mask_extracti64x4_epi64, __m256i, __m256i, __mmask8, __m512i, 1)
+test_3 (_mm512_mask_getexp_round_pd, __m512d, __m512d, __mmask8, __m512d, 5)
+test_3 (_mm512_mask_getexp_round_ps, __m512, __m512, __mmask16, __m512, 5)
+test_3y (_mm512_mask_getmant_round_pd, __m512d, __m512d, __mmask8, __m512d, 1, 1, 5)
+test_3y (_mm512_mask_getmant_round_ps, __m512, __m512, __mmask16, __m512, 1, 1, 5)
+test_3 (_mm512_mask_permute_pd, __m512d, __m512d, __mmask8, __m512d, 1)
+test_3 (_mm512_mask_permute_ps, __m512, __m512, __mmask16, __m512, 1)
+test_3 (_mm512_mask_permutex_epi64, __m512i, __m512i, __mmask8, __m512i, 1)
+test_3 (_mm512_mask_permutex_pd, __m512d, __m512d, __mmask8, __m512d, 1)
+test_3 (_mm512_mask_rol_epi32, __m512i, __m512i, __mmask16, __m512i, 1)
+test_3 (_mm512_mask_rol_epi64, __m512i, __m512i, __mmask8, __m512i, 1)
+test_3 (_mm512_mask_ror_epi32, __m512i, __m512i, __mmask16, __m512i, 1)
+test_3 (_mm512_mask_ror_epi64, __m512i, __m512i, __mmask8, __m512i, 1)
+test_3 (_mm512_mask_shuffle_epi32, __m512i, __m512i, __mmask16, __m512i, 1)
+test_3 (_mm512_mask_slli_epi32, __m512i, __m512i, __mmask16, __m512i, 1)
+test_3 (_mm512_mask_slli_epi64, __m512i, __m512i, __mmask8, __m512i, 1)
+test_3 (_mm512_mask_sqrt_round_pd, __m512d, __m512d, __mmask8, __m512d, 1)
+test_3 (_mm512_mask_sqrt_round_ps, __m512, __m512, __mmask16, __m512, 1)
+test_3 (_mm512_mask_srai_epi32, __m512i, __m512i, __mmask16, __m512i, 1)
+test_3 (_mm512_mask_srai_epi64, __m512i, __m512i, __mmask8, __m512i, 1)
+test_3 (_mm512_mask_srli_epi32, __m512i, __m512i, __mmask16, __m512i, 1)
+test_3 (_mm512_mask_srli_epi64, __m512i, __m512i, __mmask8, __m512i, 1)
+test_3 (_mm512_maskz_add_round_pd, __m512d, __mmask8, __m512d, __m512d, 1)
+test_3 (_mm512_maskz_add_round_ps, __m512, __mmask16, __m512, __m512, 1)
+test_3 (_mm512_maskz_alignr_epi32, __m512i, __mmask16, __m512i, __m512i, 1)
+test_3 (_mm512_maskz_alignr_epi64, __m512i, __mmask8, __m512i, __m512i, 1)
+test_3 (_mm512_maskz_div_round_pd, __m512d, __mmask8, __m512d, __m512d, 1)
+test_3 (_mm512_maskz_div_round_ps, __m512, __mmask16, __m512, __m512, 1)
+test_3 (_mm512_maskz_insertf32x4, __m512, __mmask16, __m512, __m128, 1)
+test_3 (_mm512_maskz_insertf64x4, __m512d, __mmask8, __m512d, __m256d, 1)
+test_3 (_mm512_maskz_inserti32x4, __m512i, __mmask16, __m512i, __m128i, 1)
+test_3 (_mm512_maskz_inserti64x4, __m512i, __mmask8, __m512i, __m256i, 1)
+test_3 (_mm512_maskz_max_round_pd, __m512d, __mmask8, __m512d, __m512d, 5)
+test_3 (_mm512_maskz_max_round_ps, __m512, __mmask16, __m512, __m512, 5)
+test_3 (_mm512_maskz_min_round_pd, __m512d, __mmask8, __m512d, __m512d, 5)
+test_3 (_mm512_maskz_min_round_ps, __m512, __mmask16, __m512, __m512, 5)
+test_3 (_mm512_maskz_mul_round_pd, __m512d, __mmask8, __m512d, __m512d, 1)
+test_3 (_mm512_maskz_mul_round_ps, __m512, __mmask16, __m512, __m512, 1)
+test_3 (_mm512_maskz_scalef_round_pd, __m512d, __mmask8, __m512d, __m512d, 1)
+test_3 (_mm512_maskz_scalef_round_ps, __m512, __mmask16, __m512, __m512, 1)
+test_3 (_mm512_maskz_shuffle_f32x4, __m512, __mmask16, __m512, __m512, 1)
+test_3 (_mm512_maskz_shuffle_f64x2, __m512d, __mmask8, __m512d, __m512d, 1)
+test_3 (_mm512_maskz_shuffle_i32x4, __m512i, __mmask16, __m512i, __m512i, 1)
+test_3 (_mm512_maskz_shuffle_i64x2, __m512i, __mmask8, __m512i, __m512i, 1)
+test_3 (_mm512_maskz_shuffle_pd, __m512d, __mmask8, __m512d, __m512d, 1)
+test_3 (_mm512_maskz_shuffle_ps, __m512, __mmask16, __m512, __m512, 1)
+test_3 (_mm512_maskz_sub_round_pd, __m512d, __mmask8, __m512d, __m512d, 1)
+test_3 (_mm512_maskz_sub_round_ps, __m512, __mmask16, __m512, __m512, 1)
+test_3 (_mm512_ternarylogic_epi32, __m512i, __m512i, __m512i, __m512i, 1)
+test_3 (_mm512_ternarylogic_epi64, __m512i, __m512i, __m512i, __m512i, 1)
+test_3 (_mm_mask_cmp_sd_mask, __mmask8, __mmask8, __m128d, __m128d, 1)
+test_3 (_mm_mask_cmp_ss_mask, __mmask8, __mmask8, __m128, __m128, 1)
+test_3v (_mm512_i32scatter_epi32, void *, __m512i, __m512i, 1)
+test_3v (_mm512_i32scatter_epi64, void *, __m256i, __m512i, 1)
+test_3v (_mm512_i32scatter_pd, void *, __m256i, __m512d, 1)
+test_3v (_mm512_i32scatter_ps, void *, __m512i, __m512, 1)
+test_3v (_mm512_i64scatter_epi32, void *, __m512i, __m256i, 1)
+test_3v (_mm512_i64scatter_epi64, void *, __m512i, __m512i, 1)
+test_3v (_mm512_i64scatter_pd, void *, __m512i, __m512d, 1)
+test_3v (_mm512_i64scatter_ps, void *, __m512i, __m256, 1)
+test_3x (_mm512_mask_roundscale_round_pd, __m512d, __m512d, __mmask8, __m512d, 1, 5)
+test_3x (_mm512_mask_roundscale_round_ps, __m512, __m512, __mmask16, __m512, 1, 5)
+test_4 (_mm512_mask3_fmadd_round_pd, __m512d, __m512d, __m512d, __m512d, __mmask8, 1)
+test_4 (_mm512_mask3_fmadd_round_ps, __m512, __m512, __m512, __m512, __mmask16, 1)
+test_4 (_mm512_mask3_fmaddsub_round_pd, __m512d, __m512d, __m512d, __m512d, __mmask8, 1)
+test_4 (_mm512_mask3_fmaddsub_round_ps, __m512, __m512, __m512, __m512, __mmask16, 1)
+test_4 (_mm512_mask3_fmsub_round_pd, __m512d, __m512d, __m512d, __m512d, __mmask8, 1)
+test_4 (_mm512_mask3_fmsub_round_ps, __m512, __m512, __m512, __m512, __mmask16, 1)
+test_4 (_mm512_mask3_fmsubadd_round_pd, __m512d, __m512d, __m512d, __m512d, __mmask8, 1)
+test_4 (_mm512_mask3_fmsubadd_round_ps, __m512, __m512, __m512, __m512, __mmask16, 1)
+test_4 (_mm512_mask3_fnmadd_round_pd, __m512d, __m512d, __m512d, __m512d, __mmask8, 1)
+test_4 (_mm512_mask3_fnmadd_round_ps, __m512, __m512, __m512, __m512, __mmask16, 1)
+test_4 (_mm512_mask3_fnmsub_round_pd, __m512d, __m512d, __m512d, __m512d, __mmask8, 1)
+test_4 (_mm512_mask3_fnmsub_round_ps, __m512, __m512, __m512, __m512, __mmask16, 1)
+test_4 (_mm512_mask_add_round_pd, __m512d, __m512d, __mmask8, __m512d, __m512d, 1)
+test_4 (_mm512_mask_add_round_ps, __m512, __m512, __mmask16, __m512, __m512, 1)
+test_4 (_mm512_mask_alignr_epi32, __m512i, __m512i, __mmask16, __m512i, __m512i, 1)
+test_4 (_mm512_mask_alignr_epi64, __m512i, __m512i, __mmask8, __m512i, __m512i, 1)
+test_4 (_mm512_mask_div_round_pd, __m512d, __m512d, __mmask8, __m512d, __m512d, 1)
+test_4 (_mm512_mask_div_round_ps, __m512, __m512, __mmask16, __m512, __m512, 1)
+test_4 (_mm512_mask_fmadd_round_pd, __m512d, __m512d, __mmask8, __m512d, __m512d, 1)
+test_4 (_mm512_mask_fmadd_round_ps, __m512, __m512, __mmask16, __m512, __m512, 1)
+test_4 (_mm512_mask_fmaddsub_round_pd, __m512d, __m512d, __mmask8, __m512d, __m512d, 1)
+test_4 (_mm512_mask_fmaddsub_round_ps, __m512, __m512, __mmask16, __m512, __m512, 1)
+test_4 (_mm512_mask_fmsub_round_pd, __m512d, __m512d, __mmask8, __m512d, __m512d, 1)
+test_4 (_mm512_mask_fmsub_round_ps, __m512, __m512, __mmask16, __m512, __m512, 1)
+test_4 (_mm512_mask_fmsubadd_round_pd, __m512d, __m512d, __mmask8, __m512d, __m512d, 1)
+test_4 (_mm512_mask_fmsubadd_round_ps, __m512, __m512, __mmask16, __m512, __m512, 1)
+test_4 (_mm512_mask_fnmadd_round_pd, __m512d, __m512d, __mmask8, __m512d, __m512d, 1)
+test_4 (_mm512_mask_fnmadd_round_ps, __m512, __m512, __mmask16, __m512, __m512, 1)
+test_4 (_mm512_mask_fnmsub_round_pd, __m512d, __m512d, __mmask8, __m512d, __m512d, 1)
+test_4 (_mm512_mask_fnmsub_round_ps, __m512, __m512, __mmask16, __m512, __m512, 1)
+test_4 (_mm512_mask_i32gather_epi32, __m512i, __m512i, __mmask16, __m512i, void const *, 1)
+test_4 (_mm512_mask_i32gather_epi64, __m512i, __m512i, __mmask8, __m256i, void const *, 1)
+test_4 (_mm512_mask_i32gather_pd, __m512d, __m512d, __mmask8, __m256i, void const *, 1)
+test_4 (_mm512_mask_i32gather_ps, __m512, __m512, __mmask16, __m512i, void const *, 1)
+test_4 (_mm512_mask_i64gather_epi32, __m256i, __m256i, __mmask8, __m512i, void const *, 1)
+test_4 (_mm512_mask_i64gather_epi64, __m512i, __m512i, __mmask8, __m512i, void const *, 1)
+test_4 (_mm512_mask_i64gather_pd, __m512d, __m512d, __mmask8, __m512i, void const *, 1)
+test_4 (_mm512_mask_i64gather_ps, __m256, __m256, __mmask8, __m512i, void const *, 1)
+test_4 (_mm512_mask_insertf32x4, __m512, __m512, __mmask16, __m512, __m128, 1)
+test_4 (_mm512_mask_insertf64x4, __m512d, __m512d, __mmask8, __m512d, __m256d, 1)
+test_4 (_mm512_mask_inserti32x4, __m512i, __m512i, __mmask16, __m512i, __m128i, 1)
+test_4 (_mm512_mask_inserti64x4, __m512i, __m512i, __mmask8, __m512i, __m256i, 1)
+test_4 (_mm512_mask_max_round_pd, __m512d, __m512d, __mmask8, __m512d, __m512d, 5)
+test_4 (_mm512_mask_max_round_ps, __m512, __m512, __mmask16, __m512, __m512, 5)
+test_4 (_mm512_mask_min_round_pd, __m512d, __m512d, __mmask8, __m512d, __m512d, 5)
+test_4 (_mm512_mask_min_round_ps, __m512, __m512, __mmask16, __m512, __m512, 5)
+test_4 (_mm512_mask_mul_round_pd, __m512d, __m512d, __mmask8, __m512d, __m512d, 1)
+test_4 (_mm512_mask_mul_round_ps, __m512, __m512, __mmask16, __m512, __m512, 1)
+test_4 (_mm512_mask_scalef_round_pd, __m512d, __m512d, __mmask8, __m512d, __m512d, 1)
+test_4 (_mm512_mask_scalef_round_ps, __m512, __m512, __mmask16, __m512, __m512, 1)
+test_4 (_mm512_mask_shuffle_f32x4, __m512, __m512, __mmask16, __m512, __m512, 1)
+test_4 (_mm512_mask_shuffle_f64x2, __m512d, __m512d, __mmask8, __m512d, __m512d, 1)
+test_4 (_mm512_mask_shuffle_i32x4, __m512i, __m512i, __mmask16, __m512i, __m512i, 1)
+test_4 (_mm512_mask_shuffle_i64x2, __m512i, __m512i, __mmask8, __m512i, __m512i, 1)
+test_4 (_mm512_mask_shuffle_pd, __m512d, __m512d, __mmask8, __m512d, __m512d, 1)
+test_4 (_mm512_mask_shuffle_ps, __m512, __m512, __mmask16, __m512, __m512, 1)
+test_4 (_mm512_mask_sub_round_pd, __m512d, __m512d, __mmask8, __m512d, __m512d, 1)
+test_4 (_mm512_mask_sub_round_ps, __m512, __m512, __mmask16, __m512, __m512, 1)
+test_4 (_mm512_mask_ternarylogic_epi32, __m512i, __m512i, __mmask16, __m512i, __m512i, 1)
+test_4 (_mm512_mask_ternarylogic_epi64, __m512i, __m512i, __mmask8, __m512i, __m512i, 1)
+test_4 (_mm512_maskz_fmadd_round_pd, __m512d, __mmask8, __m512d, __m512d, __m512d, 1)
+test_4 (_mm512_maskz_fmadd_round_ps, __m512, __mmask16, __m512, __m512, __m512, 1)
+test_4 (_mm512_maskz_fmaddsub_round_pd, __m512d, __mmask8, __m512d, __m512d, __m512d, 1)
+test_4 (_mm512_maskz_fmaddsub_round_ps, __m512, __mmask16, __m512, __m512, __m512, 1)
+test_4 (_mm512_maskz_fmsub_round_pd, __m512d, __mmask8, __m512d, __m512d, __m512d, 1)
+test_4 (_mm512_maskz_fmsub_round_ps, __m512, __mmask16, __m512, __m512, __m512, 1)
+test_4 (_mm512_maskz_fmsubadd_round_pd, __m512d, __mmask8, __m512d, __m512d, __m512d, 1)
+test_4 (_mm512_maskz_fmsubadd_round_ps, __m512, __mmask16, __m512, __m512, __m512, 1)
+test_4 (_mm512_maskz_fnmadd_round_pd, __m512d, __mmask8, __m512d, __m512d, __m512d, 1)
+test_4 (_mm512_maskz_fnmadd_round_ps, __m512, __mmask16, __m512, __m512, __m512, 1)
+test_4 (_mm512_maskz_fnmsub_round_pd, __m512d, __mmask8, __m512d, __m512d, __m512d, 1)
+test_4 (_mm512_maskz_fnmsub_round_ps, __m512, __mmask16, __m512, __m512, __m512, 1)
+test_4 (_mm512_maskz_ternarylogic_epi32, __m512i, __mmask16, __m512i, __m512i, __m512i, 1)
+test_4 (_mm512_maskz_ternarylogic_epi64, __m512i, __mmask8, __m512i, __m512i, __m512i, 1)
+test_4v (_mm512_mask_i32scatter_epi32, void *, __mmask16, __m512i, __m512i, 1)
+test_4v (_mm512_mask_i32scatter_epi64, void *, __mmask8, __m256i, __m512i, 1)
+test_4v (_mm512_mask_i32scatter_pd, void *, __mmask8, __m256i, __m512d, 1)
+test_4v (_mm512_mask_i32scatter_ps, void *, __mmask16, __m512i, __m512, 1)
+test_4v (_mm512_mask_i64scatter_epi32, void *, __mmask8, __m512i, __m256i, 1)
+test_4v (_mm512_mask_i64scatter_epi64, void *, __mmask8, __m512i, __m512i, 1)
+test_4v (_mm512_mask_i64scatter_pd, void *, __mmask8, __m512i, __m512d, 1)
+test_4v (_mm512_mask_i64scatter_ps, void *, __mmask8, __m512i, __m256, 1)
+test_4x (_mm512_mask_fixupimm_round_pd, __m512d, __m512d, __mmask8, __m512d, __m512i, 1, 5)
+test_4x (_mm512_mask_fixupimm_round_ps, __m512, __m512, __mmask16, __m512, __m512i, 1, 5)
+test_4x (_mm512_maskz_fixupimm_round_pd, __m512d, __mmask8, __m512d, __m512d, __m512i, 1, 5)
+test_4x (_mm512_maskz_fixupimm_round_ps, __m512, __mmask16, __m512, __m512, __m512i, 1, 5)
+
+/* avx512pfintrin.h */
+test_3vx (_mm512_mask_prefetch_i32gather_ps, __m512i, __mmask16, void const *, 1, 1)
+test_3vx (_mm512_mask_prefetch_i32scatter_ps, void const *, __mmask16, __m512i, 1, 1)
+test_3vx (_mm512_mask_prefetch_i64gather_ps, __m512i, __mmask8, void const *, 1, 1)
+test_3vx (_mm512_mask_prefetch_i64scatter_ps, void const *, __mmask8, __m512i, 1, 1)
+
+/* avx512erintrin.h */
+test_1 (_mm512_exp2a23_round_pd, __m512d, __m512d, 1)
+test_1 (_mm512_exp2a23_round_ps, __m512, __m512, 1)
+test_1 (_mm512_rcp28_round_pd, __m512d, __m512d, 1)
+test_1 (_mm512_rcp28_round_ps, __m512, __m512, 1)
+test_1 (_mm512_rsqrt28_round_pd, __m512d, __m512d, 1)
+test_1 (_mm512_rsqrt28_round_ps, __m512, __m512, 1)
+test_2 (_mm512_maskz_exp2a23_round_pd, __m512d, __mmask8, __m512d, 1)
+test_2 (_mm512_maskz_exp2a23_round_ps, __m512, __mmask16, __m512, 1)
+test_2 (_mm512_maskz_rcp28_round_pd, __m512d, __mmask8, __m512d, 1)
+test_2 (_mm512_maskz_rcp28_round_ps, __m512, __mmask16, __m512, 1)
+test_2 (_mm512_maskz_rsqrt28_round_pd, __m512d, __mmask8, __m512d, 1)
+test_2 (_mm512_maskz_rsqrt28_round_ps, __m512, __mmask16, __m512, 1)
+test_3 (_mm512_mask_exp2a23_round_pd, __m512d, __m512d, __mmask8, __m512d, 1)
+test_3 (_mm512_mask_exp2a23_round_ps, __m512, __m512, __mmask16, __m512, 1)
+test_3 (_mm512_mask_rcp28_round_pd, __m512d, __m512d, __mmask8, __m512d, 1)
+test_3 (_mm512_mask_rcp28_round_ps, __m512, __m512, __mmask16, __m512, 1)
+test_3 (_mm512_mask_rsqrt28_round_pd, __m512d, __m512d, __mmask8, __m512d, 1)
+test_3 (_mm512_mask_rsqrt28_round_ps, __m512, __m512, __mmask16, __m512, 1)
/* wmmintrin.h */
test_1 (_mm_aeskeygenassist_si128, __m128i, __m128i, 1)
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O0 -mavx512f" } */
+
+#include <x86intrin.h>
+
+__m512i m512i;
+__m512d m512d;
+__m512 m512;
+__m256i m256i;
+__m256d m256d;
+__m256 m256;
+__m128i m128i;
+__m128d m128d;
+__m128 m128;
+__mmask8 mmask8;
+__mmask16 mmask16;
+
+void
+test8bit (void)
+{
+ m512i = _mm512_permutex_epi64 (m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_mask_permutex_epi64 (m512i, mmask8, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_maskz_permutex_epi64 (mmask8, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+
+ m512i = _mm512_ternarylogic_epi64 (m512i, m512i, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_mask_ternarylogic_epi64 (m512i, mmask8, m512i, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_maskz_ternarylogic_epi64 (mmask8, m512i, m512i, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_ternarylogic_epi32 (m512i, m512i, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_mask_ternarylogic_epi32 (m512i, mmask16, m512i, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_maskz_ternarylogic_epi32 (mmask16, m512i, m512i, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+
+ m512i = _mm512_shuffle_epi32 (m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_mask_shuffle_epi32 (m512i, mmask16, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_maskz_shuffle_epi32 (mmask16, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+
+ m512i = _mm512_shuffle_i64x2 (m512i, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_mask_shuffle_i64x2 (m512i, mmask8, m512i, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_maskz_shuffle_i64x2 (mmask8, m512i, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+
+ m512i = _mm512_shuffle_i32x4 (m512i, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_mask_shuffle_i32x4 (m512i, mmask16, m512i, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_maskz_shuffle_i32x4 (mmask16, m512i, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+
+ m512d = _mm512_shuffle_f64x2 (m512d, m512d, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512d = _mm512_mask_shuffle_f64x2 (m512d, mmask8, m512d, m512d, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512d = _mm512_maskz_shuffle_f64x2 (mmask8, m512d, m512d, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+
+ m512 = _mm512_shuffle_f32x4 (m512, m512, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512 = _mm512_mask_shuffle_f32x4 (m512, mmask16, m512, m512, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512 = _mm512_maskz_shuffle_f32x4 (mmask16, m512, m512, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+
+ m512d = _mm512_permutex_pd (m512d, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512d = _mm512_mask_permutex_pd (m512d, mmask8, m512d, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512d = _mm512_maskz_permutex_pd (mmask8, m512d, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+
+ m512d = _mm512_permute_pd (m512d, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512d = _mm512_mask_permute_pd (m512d, mmask8, m512d, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512d = _mm512_maskz_permute_pd (mmask8, m512d, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+
+ m512 = _mm512_permute_ps (m512, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512 = _mm512_mask_permute_ps (m512, mmask16, m512, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512 = _mm512_maskz_permute_ps (mmask16, m512, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+
+ m512d = _mm512_shuffle_pd (m512d, m512d, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512d = _mm512_mask_shuffle_pd (m512d, mmask8, m512d, m512d, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512d = _mm512_maskz_shuffle_pd (mmask8, m512d, m512d, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+
+ m512 = _mm512_shuffle_ps (m512, m512, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512 = _mm512_mask_shuffle_ps (m512, mmask16, m512, m512, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512 = _mm512_maskz_shuffle_ps (mmask16, m512, m512, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+
+ m512d = _mm512_fixupimm_pd (m512d, m512d, m512i, 256); /* { dg-error "the immediate argument must be an 8-bit immediate" } */
+ m512d = _mm512_mask_fixupimm_pd (m512d, mmask8, m512d, m512i, 256); /* { dg-error "the immediate argument must be an 8-bit immediate" } */
+ m512d = _mm512_maskz_fixupimm_pd (mmask8, m512d, m512d, m512i, 256); /* { dg-error "the immediate argument must be an 8-bit immediate" } */
+
+ m512 = _mm512_fixupimm_ps (m512, m512, m512i, 256); /* { dg-error "the immediate argument must be an 8-bit immediate" } */
+ m512 = _mm512_mask_fixupimm_ps (m512, mmask16, m512, m512i, 256); /* { dg-error "the immediate argument must be an 8-bit immediate" } */
+ m512 = _mm512_maskz_fixupimm_ps (mmask16, m512, m512, m512i, 256); /* { dg-error "the immediate argument must be an 8-bit immediate" } */
+
+
+
+ m512i = _mm512_rol_epi32 (m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_mask_rol_epi32 (m512i, mmask16, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_maskz_rol_epi32 (mmask16, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+
+ m512i = _mm512_ror_epi32 (m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_mask_ror_epi32 (m512i, mmask16, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_maskz_ror_epi32 (mmask16, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+
+ m512i = _mm512_rol_epi64 (m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_mask_rol_epi64 (m512i, mmask8, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_maskz_rol_epi64 (mmask8, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+
+ m512i = _mm512_ror_epi64 (m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_mask_ror_epi64 (m512i, mmask8, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_maskz_ror_epi64 (mmask8, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+
+ m256i = _mm512_cvtps_ph (m512, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m256i = _mm512_mask_cvtps_ph (m256i, mmask16, m512, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m256i = _mm512_maskz_cvtps_ph (mmask16, m512, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+
+ m512d = _mm512_roundscale_pd (m512d, 256); /* { dg-error "the immediate argument must be an 8-bit immediate" } */
+ m512d = _mm512_mask_roundscale_pd (m512d, mmask8, m512d, 256); /* { dg-error "the immediate argument must be an 8-bit immediate" } */
+ m512d = _mm512_maskz_roundscale_pd (mmask8, m512d, 256); /* { dg-error "the immediate argument must be an 8-bit immediate" } */
+
+ m512 = _mm512_roundscale_ps (m512, 256); /* { dg-error "the immediate argument must be an 8-bit immediate" } */
+ m512 = _mm512_mask_roundscale_ps (m512, mmask16, m512, 256); /* { dg-error "the immediate argument must be an 8-bit immediate" } */
+ m512 = _mm512_maskz_roundscale_ps (mmask16, m512, 256); /* { dg-error "the immediate argument must be an 8-bit immediate" } */
+
+
+ m512i = _mm512_alignr_epi32 (m512i, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_mask_alignr_epi32 (m512i, mmask16, m512i, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_maskz_alignr_epi32 (mmask16, m512i, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_alignr_epi64 (m512i, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_mask_alignr_epi64 (m512i, mmask8, m512i, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+ m512i = _mm512_maskz_alignr_epi64 (mmask8, m512i, m512i, 256); /* { dg-error "the last argument must be an 8-bit immediate" } */
+
+ mmask8 = _mm512_cmp_epi64_mask (m512i, m512i, 256); /* { dg-error "the last argument must be a 3-bit immediate" } */
+ mmask8 = _mm512_cmp_epi32_mask (m512i, m512i, 256); /* { dg-error "the last argument must be a 3-bit immediate" } */
+ mmask8 = _mm512_cmp_epu64_mask (m512i, m512i, 256); /* { dg-error "the last argument must be a 3-bit immediate" } */
+ mmask8 = _mm512_cmp_epu32_mask (m512i, m512i, 256); /* { dg-error "the last argument must be a 3-bit immediate" } */
+ mmask8 = _mm512_cmp_pd_mask (m512d, m512d, 256); /* { dg-error "the immediate argument must be a 5-bit immediate" } */
+ mmask8 = _mm512_cmp_ps_mask (m512, m512, 256); /* { dg-error "the immediate argument must be a 5-bit immediate" } */
+ mmask8 = _mm512_mask_cmp_epi64_mask (2, m512i, m512i, 256); /* { dg-error "the last argument must be a 3-bit immediate" } */
+ mmask8 = _mm512_mask_cmp_epi32_mask (2, m512i, m512i, 256); /* { dg-error "the last argument must be a 3-bit immediate" } */
+ mmask8 = _mm512_mask_cmp_epu64_mask (2, m512i, m512i, 256); /* { dg-error "the last argument must be a 3-bit immediate" } */
+ mmask8 = _mm512_mask_cmp_epu32_mask (2, m512i, m512i, 256); /* { dg-error "the last argument must be a 3-bit immediate" } */
+ mmask8 = _mm512_mask_cmp_pd_mask (2, m512d, m512d, 256); /* { dg-error "the immediate argument must be a 5-bit immediate" } */
+ mmask8 = _mm512_mask_cmp_ps_mask (2, m512, m512, 256); /* { dg-error "the immediate argument must be a 5-bit immediate" } */
+ mmask8 = _mm_cmp_sd_mask (m128d, m128d, 256); /* { dg-error "the immediate argument must be a 5-bit immediate" } */
+ mmask8 = _mm_cmp_ss_mask (m128, m128, 256); /* { dg-error "the immediate argument must be a 5-bit immediate" } */
+ mmask8 = _mm_mask_cmp_sd_mask (1, m128d, m128d, 256); /* { dg-error "the immediate argument must be a 5-bit immediate" } */
+ mmask8 = _mm_mask_cmp_ss_mask (1, m128, m128, 256); /* { dg-error "the immediate argument must be a 5-bit immediate" } */
+
+}
+
+test1bit (void) {
+ m256d = _mm512_extractf64x4_pd (m512d, 256); /* { dg-error "the last argument must be a 1-bit immediate" } */
+ m256d = _mm512_mask_extractf64x4_pd (m256d, mmask8, m512d, 256); /* { dg-error "the last argument must be a 1-bit immediate" } */
+ m256d = _mm512_maskz_extractf64x4_pd (mmask8, m512d, 256); /* { dg-error "the last argument must be a 1-bit immediate" } */
+
+ m256i = _mm512_extracti64x4_epi64 (m512i, 256); /* { dg-error "the last argument must be a 1-bit immediate" } */
+ m256i = _mm512_mask_extracti64x4_epi64 (m256i, mmask8, m512i, 256); /* { dg-error "the last argument must be a 1-bit immediate" } */
+ m256i = _mm512_maskz_extracti64x4_epi64 (mmask8, m512i, 256); /* { dg-error "the last argument must be a 1-bit immediate" } */
+
+ m512d = _mm512_insertf64x4 (m512d, m256d, 256); /* { dg-error "the last argument must be a 1-bit immediate" } */
+ m512d = _mm512_mask_insertf64x4 (m512d, mmask8, m512d, m256d, 256); /* { dg-error "the last argument must be a 1-bit immediate" } */
+ m512d = _mm512_maskz_insertf64x4 (mmask8, m512d, m256d, 256); /* { dg-error "the last argument must be a 1-bit immediate" } */
+
+ m512i = _mm512_inserti64x4 (m512i, m256i, 256); /* { dg-error "the last argument must be a 1-bit immediate" } */
+ m512i = _mm512_mask_inserti64x4 (m512i, mmask8, m512i, m256i, 256); /* { dg-error "the last argument must be a 1-bit immediate" } */
+ m512i = _mm512_maskz_inserti64x4 (mmask8, m512i, m256i, 256); /* { dg-error "the last argument must be a 1-bit immediate" } */
+}
+
+test2bit (void) {
+ m128 = _mm512_extractf32x4_ps(m512, 256); /* { dg-error "the last argument must be a 2-bit immediate" } */
+ m128 = _mm512_mask_extractf32x4_ps(m128, mmask8, m512, 256); /* { dg-error "the last argument must be a 2-bit immediate" } */
+ m128 = _mm512_maskz_extractf32x4_ps(mmask8, m512, 256); /* { dg-error "the last argument must be a 2-bit immediate" } */
+
+ m128i = _mm512_extracti32x4_epi32 (m512i, 256); /* { dg-error "the last argument must be a 2-bit immediate" } */
+ m128i = _mm512_mask_extracti32x4_epi32 (m128i, mmask8, m512i, 256); /* { dg-error "the last argument must be a 2-bit immediate" } */
+ m128i = _mm512_maskz_extracti32x4_epi32 (mmask8, m512i, 256); /* { dg-error "the last argument must be a 2-bit immediate" } */
+
+ m512 = _mm512_insertf32x4 (m512, m128, 256); /* { dg-error "the last argument must be a 2-bit immediate" } */
+ m512 = _mm512_mask_insertf32x4 (m512, mmask16, m512, m128, 256); /* { dg-error "the last argument must be a 2-bit immediate" } */
+ m512 = _mm512_maskz_insertf32x4 (mmask16, m512, m128, 256); /* { dg-error "the last argument must be a 2-bit immediate" } */
+
+ m512i = _mm512_inserti32x4 (m512i, m128i, 256); /* { dg-error "the last argument must be a 2-bit immediate" } */
+ m512i = _mm512_mask_inserti32x4 (m512i, mmask16, m512i, m128i, 256); /* { dg-error "the last argument must be a 2-bit immediate" } */
+ m512i = _mm512_maskz_inserti32x4 (mmask16, m512i, m128i, 256); /* { dg-error "the last argument must be a 2-bit immediate" } */
+}
+
+test4bit (void) {
+ m512d = _mm512_getmant_pd (m512d, 1, 64); /* { dg-error "the immediate argument must be a 4-bit immediate" } */
+ m512d = _mm512_mask_getmant_pd (m512d, mmask8, m512d, 1, 64); /* { dg-error "the immediate argument must be a 4-bit immediate" } */
+ m512d = _mm512_maskz_getmant_pd (mmask8, m512d, 1, 64); /* { dg-error "the immediate argument must be a 4-bit immediate" } */
+
+ m512 = _mm512_getmant_ps (m512, 1, 64); /* { dg-error "the immediate argument must be a 4-bit immediate" } */
+ m512 = _mm512_mask_getmant_ps (m512, mmask16, m512, 1, 64); /* { dg-error "the immediate argument must be a 4-bit immediate" } */
+ m512 = _mm512_maskz_getmant_ps (mmask16, m512, 1, 64); /* { dg-error "the immediate argument must be a 4-bit immediate" } */
+
+
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-O0 -mavx512f" } */
+
+#include <x86intrin.h>
+
+int i;
+unsigned int ui;
+__m512i m512i;
+__m512d m512d;
+__m512 m512;
+__m256i m256i;
+__m256 m256;
+__m128i m128i;
+__m128d m128d;
+__m128 m128;
+__mmask8 mmask8;
+__mmask16 mmask16;
+
+void
+test_round (void)
+{
+ m512d = _mm512_sqrt_round_pd (m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_sqrt_round_pd (m512d, mmask8, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_sqrt_round_pd (mmask8, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_sqrt_round_ps (m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_sqrt_round_ps (m512, mmask16, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_sqrt_round_ps (mmask16, m512, 7); /* { dg-error "incorrect rounding operand" } */
+
+ m512d = _mm512_add_round_pd (m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_add_round_pd (m512d, mmask8, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_add_round_pd (mmask8, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_add_round_ps (m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_add_round_ps (m512, mmask16, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_add_round_ps (mmask16, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_sub_round_pd (m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_sub_round_pd (m512d, mmask8, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_sub_round_pd (mmask8, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_sub_round_ps (m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_sub_round_ps (m512, mmask16, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_sub_round_ps (mmask16, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+
+ m512d = _mm512_mul_round_pd (m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_mul_round_pd (m512d, mmask8, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_mul_round_pd (mmask8, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mul_round_ps (m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_mul_round_ps (m512, mmask16, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_mul_round_ps (mmask16, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_div_round_pd (m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_div_round_pd (m512d, mmask8, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_div_round_pd (mmask8, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_div_round_ps (m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_div_round_ps (m512, mmask16, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_div_round_ps (mmask16, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+
+ m512d = _mm512_scalef_round_pd(m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_scalef_round_pd(m512d, mmask8, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_scalef_round_pd(mmask8, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_scalef_round_ps(m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_scalef_round_ps(m512, mmask16, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_scalef_round_ps(mmask16, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+
+ m512d = _mm512_fmadd_round_pd (m512d, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_fmadd_round_pd (m512d, mmask8, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask3_fmadd_round_pd (m512d, m512d, m512d, mmask8, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_fmadd_round_pd (mmask8, m512d, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_fmadd_round_ps (m512, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_fmadd_round_ps (m512, mmask16, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask3_fmadd_round_ps (m512, m512, m512, mmask16, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_fmadd_round_ps (mmask16, m512, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_fmsub_round_pd (m512d, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_fmsub_round_pd (m512d, mmask8, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask3_fmsub_round_pd (m512d, m512d, m512d, mmask8, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_fmsub_round_pd (mmask8, m512d, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_fmsub_round_ps (m512, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_fmsub_round_ps (m512, mmask16, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask3_fmsub_round_ps (m512, m512, m512, mmask16, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_fmsub_round_ps (mmask16, m512, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_fmaddsub_round_pd (m512d, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_fmaddsub_round_pd (m512d, mmask8, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask3_fmaddsub_round_pd (m512d, m512d, m512d, mmask8, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_fmaddsub_round_pd (mmask8, m512d, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_fmaddsub_round_ps (m512, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_fmaddsub_round_ps (m512, mmask16, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask3_fmaddsub_round_ps (m512, m512, m512, mmask16, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_fmaddsub_round_ps (mmask16, m512, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_fmsubadd_round_pd (m512d, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_fmsubadd_round_pd (m512d, mmask8, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask3_fmsubadd_round_pd (m512d, m512d, m512d, mmask8, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_fmsubadd_round_pd (mmask8, m512d, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_fmsubadd_round_ps (m512, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_fmsubadd_round_ps (m512, mmask16, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask3_fmsubadd_round_ps (m512, m512, m512, mmask16, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_fmsubadd_round_ps (mmask16, m512, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_fnmadd_round_pd (m512d, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_fnmadd_round_pd (m512d, mmask8, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask3_fnmadd_round_pd (m512d, m512d, m512d, mmask8, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_fnmadd_round_pd (mmask8, m512d, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_fnmadd_round_ps (m512, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_fnmadd_round_ps (m512, mmask16, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask3_fnmadd_round_ps (m512, m512, m512, mmask16, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_fnmadd_round_ps (mmask16, m512, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_fnmsub_round_pd (m512d, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_fnmsub_round_pd (m512d, mmask8, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask3_fnmsub_round_pd (m512d, m512d, m512d, mmask8, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_fnmsub_round_pd (mmask8, m512d, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_fnmsub_round_ps (m512, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_fnmsub_round_ps (m512, mmask16, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask3_fnmsub_round_ps (m512, m512, m512, mmask16, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_fnmsub_round_ps (mmask16, m512, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+
+ m256i = _mm512_cvt_roundpd_epi32 (m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m256i = _mm512_mask_cvt_roundpd_epi32 (m256i, mmask8, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m256i = _mm512_maskz_cvt_roundpd_epi32 (mmask8, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m256i = _mm512_cvt_roundpd_epu32 (m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m256i = _mm512_mask_cvt_roundpd_epu32 (m256i, mmask8, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m256i = _mm512_maskz_cvt_roundpd_epu32 (mmask8, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+
+ m512i = _mm512_cvt_roundps_epi32 (m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512i = _mm512_mask_cvt_roundps_epi32 (m512i, mmask16, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512i = _mm512_maskz_cvt_roundps_epi32 (mmask16, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512i = _mm512_cvt_roundps_epu32 (m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512i = _mm512_mask_cvt_roundps_epu32 (m512i, mmask16, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512i = _mm512_maskz_cvt_roundps_epu32 (mmask16, m512, 7); /* { dg-error "incorrect rounding operand" } */
+
+ m128 = _mm_cvt_roundu32_ss (m128, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ m128 = _mm_cvt_roundi32_ss (m128, 4, 7); /* { dg-error "incorrect rounding operand" } */
+
+ m512 = _mm512_cvt_roundepi32_ps (m512i, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_cvt_roundepi32_ps (m512, mmask16, m512i, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_cvt_roundepi32_ps (mmask16, m512i, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_cvt_roundepu32_ps (m512i, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_cvt_roundepu32_ps (m512, mmask16, m512i, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_cvt_roundepu32_ps (mmask16, m512i, 7); /* { dg-error "incorrect rounding operand" } */
+
+ ui = _mm_cvt_roundss_u32 (m128, 7); /* { dg-error "incorrect rounding operand" } */
+ i = _mm_cvt_roundss_i32 (m128, 7); /* { dg-error "incorrect rounding operand" } */
+
+ ui = _mm_cvt_roundsd_u32 (m128d, 7); /* { dg-error "incorrect rounding operand" } */
+ i = _mm_cvt_roundsd_i32 (m128d, 7); /* { dg-error "incorrect rounding operand" } */
+
+ m256 = _mm512_cvt_roundpd_ps (m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m256 = _mm512_mask_cvt_roundpd_ps (m256, mmask8, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m256 = _mm512_maskz_cvt_roundpd_ps (mmask8, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+
+ m512d = _mm512_max_round_pd (m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_max_round_pd (m512d, mmask8, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_max_round_pd (mmask8, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_max_round_ps (m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_max_round_ps (m512, mmask16, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_max_round_ps (mmask16, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_min_round_pd (m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_min_round_pd (m512d, mmask8, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_min_round_pd (mmask8, m512d, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_min_round_ps (m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_min_round_ps (m512, mmask16, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_min_round_ps (mmask16, m512, m512, 7); /* { dg-error "incorrect rounding operand" } */
+
+ m256i = _mm512_cvtt_roundpd_epi32 (m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m256i = _mm512_mask_cvtt_roundpd_epi32 (m256i, mmask8, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m256i = _mm512_maskz_cvtt_roundpd_epi32 (mmask8, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m256i = _mm512_cvtt_roundpd_epu32 (m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m256i = _mm512_mask_cvtt_roundpd_epu32 (m256i, mmask8, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m256i = _mm512_maskz_cvtt_roundpd_epu32 (mmask8, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+
+ m512i = _mm512_cvtt_roundps_epi32 (m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512i = _mm512_mask_cvtt_roundps_epi32 (m512i, mmask16, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512i = _mm512_maskz_cvtt_roundps_epi32 (mmask16, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512i = _mm512_cvtt_roundps_epu32 (m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512i = _mm512_mask_cvtt_roundps_epu32 (m512i, mmask16, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512i = _mm512_maskz_cvtt_roundps_epu32 (mmask16, m512, 7); /* { dg-error "incorrect rounding operand" } */
+
+ m512d = _mm512_fixupimm_round_pd (m512d, m512d, m512i, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_fixupimm_round_pd (m512d, mmask8, m512d, m512i, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_fixupimm_round_pd (mmask8, m512d, m512d, m512i, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_fixupimm_round_ps (m512, m512, m512i, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_fixupimm_round_ps (m512, mmask16, m512, m512i, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_fixupimm_round_ps (mmask16, m512, m512, m512i, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ m128d = _mm_fixupimm_round_sd (m128d, m128d, m128i, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ m128d = _mm_mask_fixupimm_round_sd (m128d, mmask8, m128d, m128i, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ m128d = _mm_maskz_fixupimm_round_sd (mmask8, m128d, m128d, m128i, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ m128 = _mm_fixupimm_round_ss (m128, m128, m128i, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ m128 = _mm_mask_fixupimm_round_ss (m128, mmask8, m128, m128i, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ m128 = _mm_maskz_fixupimm_round_ss (mmask8, m128, m128, m128i, 4, 7); /* { dg-error "incorrect rounding operand" } */
+
+ ui = _mm_cvtt_roundss_u32 (m128, 7); /* { dg-error "incorrect rounding operand" } */
+ i = _mm_cvtt_roundss_i32 (m128, 7); /* { dg-error "incorrect rounding operand" } */
+
+ ui = _mm_cvtt_roundsd_u32 (m128d, 7); /* { dg-error "incorrect rounding operand" } */
+ i = _mm_cvtt_roundsd_i32 (m128d, 7); /* { dg-error "incorrect rounding operand" } */
+
+ m512d = _mm512_cvt_roundps_pd (m256, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_cvt_roundps_pd (m512d, mmask8, m256, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_cvt_roundps_pd (mmask8, m256, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_cvt_roundph_ps (m256i, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_cvt_roundph_ps (m512, mmask16, m256i, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_cvt_roundph_ps (mmask16, m256i, 7); /* { dg-error "incorrect rounding operand" } */
+
+ m512 = _mm512_getexp_round_ps (m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_getexp_round_ps (m512, mmask16, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_getexp_round_ps (mmask16, m512, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_getexp_round_pd (m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_getexp_round_pd (m512d, mmask8, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_getexp_round_pd (mmask8, m512d, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_getmant_round_pd (m512d, 0, 0, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_getmant_round_pd (m512d, mmask8, m512d, 0, 0, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_getmant_round_pd (mmask8, m512d, 0, 0, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_getmant_round_ps (m512, 0, 0, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_getmant_round_ps (m512, mmask16, m512, 0, 0, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_getmant_round_ps (mmask16, m512, 0, 0, 7); /* { dg-error "incorrect rounding operand" } */
+
+ m512 = _mm512_roundscale_round_ps (m512, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_roundscale_round_ps (m512, mmask16, m512, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_roundscale_round_ps (mmask16, m512, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_roundscale_round_pd (m512d, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_roundscale_round_pd (m512d, mmask8, m512d, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_roundscale_round_pd (mmask8, m512d, 4, 7); /* { dg-error "incorrect rounding operand" } */
+
+ mmask8 = _mm512_cmp_round_pd_mask (m512d, m512d, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ mmask16 = _mm512_cmp_round_ps_mask (m512, m512, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ mmask8 = _mm512_mask_cmp_round_pd_mask (mmask8, m512d, m512d, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ mmask16 = _mm512_mask_cmp_round_ps_mask (mmask16, m512, m512, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ mmask8 = _mm_cmp_round_sd_mask (m128d, m128d, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ mmask8 = _mm_mask_cmp_round_sd_mask (mmask8, m128d, m128d, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ mmask16 = _mm_cmp_round_ss_mask (m128, m128, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ mmask16 = _mm_mask_cmp_round_ss_mask (mmask8, m128, m128, 4, 7); /* { dg-error "incorrect rounding operand" } */
+
+ i = _mm_comi_round_ss (m128, m128, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ i = _mm_comi_round_sd (m128d, m128d, 4, 7); /* { dg-error "incorrect rounding operand" } */
+}
+
+void
+test_round_sae (void)
+{
+ m512d = _mm512_sqrt_round_pd (m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_sqrt_round_pd (m512d, mmask8, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_sqrt_round_pd (mmask8, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_sqrt_round_ps (m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_sqrt_round_ps (m512, mmask16, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_sqrt_round_ps (mmask16, m512, 5); /* { dg-error "incorrect rounding operand" } */
+
+ m512d = _mm512_add_round_pd (m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_add_round_pd (m512d, mmask8, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_add_round_pd (mmask8, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_add_round_ps (m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_add_round_ps (m512, mmask16, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_add_round_ps (mmask16, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_sub_round_pd (m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_sub_round_pd (m512d, mmask8, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_sub_round_pd (mmask8, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_sub_round_ps (m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_sub_round_ps (m512, mmask16, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_sub_round_ps (mmask16, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+
+ m512d = _mm512_mul_round_pd (m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_mul_round_pd (m512d, mmask8, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_mul_round_pd (mmask8, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mul_round_ps (m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_mul_round_ps (m512, mmask16, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_mul_round_ps (mmask16, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_div_round_pd (m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_div_round_pd (m512d, mmask8, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_div_round_pd (mmask8, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_div_round_ps (m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_div_round_ps (m512, mmask16, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_div_round_ps (mmask16, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+
+ m512d = _mm512_scalef_round_pd(m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_scalef_round_pd(m512d, mmask8, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_scalef_round_pd(mmask8, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_scalef_round_ps(m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_scalef_round_ps(m512, mmask16, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_scalef_round_ps(mmask16, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+
+ m512d = _mm512_fmadd_round_pd (m512d, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_fmadd_round_pd (m512d, mmask8, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask3_fmadd_round_pd (m512d, m512d, m512d, mmask8, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_fmadd_round_pd (mmask8, m512d, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_fmadd_round_ps (m512, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_fmadd_round_ps (m512, mmask16, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask3_fmadd_round_ps (m512, m512, m512, mmask16, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_fmadd_round_ps (mmask16, m512, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_fmsub_round_pd (m512d, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_fmsub_round_pd (m512d, mmask8, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask3_fmsub_round_pd (m512d, m512d, m512d, mmask8, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_fmsub_round_pd (mmask8, m512d, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_fmsub_round_ps (m512, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_fmsub_round_ps (m512, mmask16, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask3_fmsub_round_ps (m512, m512, m512, mmask16, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_fmsub_round_ps (mmask16, m512, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_fmaddsub_round_pd (m512d, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_fmaddsub_round_pd (m512d, mmask8, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask3_fmaddsub_round_pd (m512d, m512d, m512d, mmask8, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_fmaddsub_round_pd (mmask8, m512d, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_fmaddsub_round_ps (m512, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_fmaddsub_round_ps (m512, mmask16, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask3_fmaddsub_round_ps (m512, m512, m512, mmask16, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_fmaddsub_round_ps (mmask16, m512, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_fmsubadd_round_pd (m512d, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_fmsubadd_round_pd (m512d, mmask8, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask3_fmsubadd_round_pd (m512d, m512d, m512d, mmask8, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_fmsubadd_round_pd (mmask8, m512d, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_fmsubadd_round_ps (m512, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_fmsubadd_round_ps (m512, mmask16, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask3_fmsubadd_round_ps (m512, m512, m512, mmask16, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_fmsubadd_round_ps (mmask16, m512, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_fnmadd_round_pd (m512d, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_fnmadd_round_pd (m512d, mmask8, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask3_fnmadd_round_pd (m512d, m512d, m512d, mmask8, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_fnmadd_round_pd (mmask8, m512d, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_fnmadd_round_ps (m512, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_fnmadd_round_ps (m512, mmask16, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask3_fnmadd_round_ps (m512, m512, m512, mmask16, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_fnmadd_round_ps (mmask16, m512, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_fnmsub_round_pd (m512d, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_fnmsub_round_pd (m512d, mmask8, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask3_fnmsub_round_pd (m512d, m512d, m512d, mmask8, 5); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_fnmsub_round_pd (mmask8, m512d, m512d, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_fnmsub_round_ps (m512, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_fnmsub_round_ps (m512, mmask16, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask3_fnmsub_round_ps (m512, m512, m512, mmask16, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_fnmsub_round_ps (mmask16, m512, m512, m512, 5); /* { dg-error "incorrect rounding operand" } */
+
+ m256i = _mm512_cvt_roundpd_epi32 (m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m256i = _mm512_mask_cvt_roundpd_epi32 (m256i, mmask8, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m256i = _mm512_maskz_cvt_roundpd_epi32 (mmask8, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m256i = _mm512_cvt_roundpd_epu32 (m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m256i = _mm512_mask_cvt_roundpd_epu32 (m256i, mmask8, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m256i = _mm512_maskz_cvt_roundpd_epu32 (mmask8, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+
+ m512i = _mm512_cvt_roundps_epi32 (m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512i = _mm512_mask_cvt_roundps_epi32 (m512i, mmask16, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512i = _mm512_maskz_cvt_roundps_epi32 (mmask16, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512i = _mm512_cvt_roundps_epu32 (m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512i = _mm512_mask_cvt_roundps_epu32 (m512i, mmask16, m512, 5); /* { dg-error "incorrect rounding operand" } */
+ m512i = _mm512_maskz_cvt_roundps_epu32 (mmask16, m512, 5); /* { dg-error "incorrect rounding operand" } */
+
+ m128 = _mm_cvt_roundu32_ss (m128, 4, 5); /* { dg-error "incorrect rounding operand" } */
+ m128 = _mm_cvt_roundi32_ss (m128, 4, 5); /* { dg-error "incorrect rounding operand" } */
+
+ m512 = _mm512_cvt_roundepi32_ps (m512i, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_cvt_roundepi32_ps (m512, mmask16, m512i, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_cvt_roundepi32_ps (mmask16, m512i, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_cvt_roundepu32_ps (m512i, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_cvt_roundepu32_ps (m512, mmask16, m512i, 5); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_cvt_roundepu32_ps (mmask16, m512i, 5); /* { dg-error "incorrect rounding operand" } */
+
+ ui = _mm_cvt_roundss_u32 (m128, 5); /* { dg-error "incorrect rounding operand" } */
+ i = _mm_cvt_roundss_i32 (m128, 5); /* { dg-error "incorrect rounding operand" } */
+
+ ui = _mm_cvt_roundsd_u32 (m128d, 5); /* { dg-error "incorrect rounding operand" } */
+ i = _mm_cvt_roundsd_i32 (m128d, 5); /* { dg-error "incorrect rounding operand" } */
+
+ m256 = _mm512_cvt_roundpd_ps (m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m256 = _mm512_mask_cvt_roundpd_ps (m256, mmask8, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+ m256 = _mm512_maskz_cvt_roundpd_ps (mmask8, m512d, 5); /* { dg-error "incorrect rounding operand" } */
+}
+
+void
+test_sae_only (void)
+{
+ m512d = _mm512_max_round_pd (m512d, m512d, 3); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_max_round_pd (m512d, mmask8, m512d, m512d, 3); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_max_round_pd (mmask8, m512d, m512d, 3); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_max_round_ps (m512, m512, 3); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_max_round_ps (m512, mmask16, m512, m512, 3); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_max_round_ps (mmask16, m512, m512, 3); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_min_round_pd (m512d, m512d, 3); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_min_round_pd (m512d, mmask8, m512d, m512d, 3); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_min_round_pd (mmask8, m512d, m512d, 3); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_min_round_ps (m512, m512, 3); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_min_round_ps (m512, mmask16, m512, m512, 3); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_min_round_ps (mmask16, m512, m512, 3); /* { dg-error "incorrect rounding operand" } */
+
+ m256i = _mm512_cvtt_roundpd_epi32 (m512d, 3); /* { dg-error "incorrect rounding operand" } */
+ m256i = _mm512_mask_cvtt_roundpd_epi32 (m256i, mmask8, m512d, 3); /* { dg-error "incorrect rounding operand" } */
+ m256i = _mm512_maskz_cvtt_roundpd_epi32 (mmask8, m512d, 3); /* { dg-error "incorrect rounding operand" } */
+ m256i = _mm512_cvtt_roundpd_epu32 (m512d, 3); /* { dg-error "incorrect rounding operand" } */
+ m256i = _mm512_mask_cvtt_roundpd_epu32 (m256i, mmask8, m512d, 3); /* { dg-error "incorrect rounding operand" } */
+ m256i = _mm512_maskz_cvtt_roundpd_epu32 (mmask8, m512d, 3); /* { dg-error "incorrect rounding operand" } */
+
+ m512i = _mm512_cvtt_roundps_epi32 (m512, 3); /* { dg-error "incorrect rounding operand" } */
+ m512i = _mm512_mask_cvtt_roundps_epi32 (m512i, mmask16, m512, 3); /* { dg-error "incorrect rounding operand" } */
+ m512i = _mm512_maskz_cvtt_roundps_epi32 (mmask16, m512, 3); /* { dg-error "incorrect rounding operand" } */
+ m512i = _mm512_cvtt_roundps_epu32 (m512, 3); /* { dg-error "incorrect rounding operand" } */
+ m512i = _mm512_mask_cvtt_roundps_epu32 (m512i, mmask16, m512, 3); /* { dg-error "incorrect rounding operand" } */
+ m512i = _mm512_maskz_cvtt_roundps_epu32 (mmask16, m512, 3); /* { dg-error "incorrect rounding operand" } */
+
+ m512d = _mm512_fixupimm_round_pd (m512d, m512d, m512i, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_fixupimm_round_pd (m512d, mmask8, m512d, m512i, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_fixupimm_round_pd (mmask8, m512d, m512d, m512i, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_fixupimm_round_ps (m512, m512, m512i, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_fixupimm_round_ps (m512, mmask16, m512, m512i, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_fixupimm_round_ps (mmask16, m512, m512, m512i, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ m128d = _mm_fixupimm_round_sd (m128d, m128d, m128i, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ m128d = _mm_mask_fixupimm_round_sd (m128d, mmask8, m128d, m128i, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ m128d = _mm_maskz_fixupimm_round_sd (mmask8, m128d, m128d, m128i, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ m128 = _mm_fixupimm_round_ss (m128, m128, m128i, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ m128 = _mm_mask_fixupimm_round_ss (m128, mmask8, m128, m128i, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ m128 = _mm_maskz_fixupimm_round_ss (mmask8, m128, m128, m128i, 4, 3); /* { dg-error "incorrect rounding operand" } */
+
+ ui = _mm_cvtt_roundss_u32 (m128, 3); /* { dg-error "incorrect rounding operand" } */
+ i = _mm_cvtt_roundss_i32 (m128, 3); /* { dg-error "incorrect rounding operand" } */
+
+ ui = _mm_cvtt_roundsd_u32 (m128d, 3); /* { dg-error "incorrect rounding operand" } */
+ i = _mm_cvtt_roundsd_i32 (m128d, 3); /* { dg-error "incorrect rounding operand" } */
+
+ m512d = _mm512_cvt_roundps_pd (m256, 3); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_cvt_roundps_pd (m512d, mmask8, m256, 3); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_cvt_roundps_pd (mmask8, m256, 3); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_cvt_roundph_ps (m256i, 3); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_cvt_roundph_ps (m512, mmask16, m256i, 3); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_cvt_roundph_ps (mmask16, m256i, 3); /* { dg-error "incorrect rounding operand" } */
+
+ m512 = _mm512_getexp_round_ps (m512, 3); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_getexp_round_ps (m512, mmask16, m512, 3); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_getexp_round_ps (mmask16, m512, 3); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_getexp_round_pd (m512d, 3); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_getexp_round_pd (m512d, mmask8, m512d, 3); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_getexp_round_pd (mmask8, m512d, 3); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_getmant_round_pd (m512d, 0, 0, 3); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_getmant_round_pd (m512d, mmask8, m512d, 0, 0, 3); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_getmant_round_pd (mmask8, m512d, 0, 0, 3); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_getmant_round_ps (m512, 0, 0, 3); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_getmant_round_ps (m512, mmask16, m512, 0, 0, 3); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_getmant_round_ps (mmask16, m512, 0, 0, 3); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_roundscale_round_ps (m512, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_mask_roundscale_round_ps (m512, mmask16, m512, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ m512 = _mm512_maskz_roundscale_round_ps (mmask16, m512, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_roundscale_round_pd (m512d, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_mask_roundscale_round_pd (m512d, mmask8, m512d, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ m512d = _mm512_maskz_roundscale_round_pd (mmask8, m512d, 4, 3); /* { dg-error "incorrect rounding operand" } */
+
+ mmask8 = _mm512_cmp_round_pd_mask (m512d, m512d, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ mmask16 = _mm512_cmp_round_ps_mask (m512, m512, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ mmask8 = _mm512_mask_cmp_round_pd_mask (mmask8, m512d, m512d, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ mmask16 = _mm512_mask_cmp_round_ps_mask (mmask16, m512, m512, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ mmask8 = _mm_cmp_round_sd_mask (m128d, m128d, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ mmask8 = _mm_mask_cmp_round_sd_mask (mmask8, m128d, m128d, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ mmask16 = _mm_cmp_round_ss_mask (m128, m128, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ mmask16 = _mm_mask_cmp_round_ss_mask (mmask8, m128, m128, 4, 3); /* { dg-error "incorrect rounding operand" } */
+
+ i = _mm_comi_round_ss (m128, m128, 4, 3); /* { dg-error "incorrect rounding operand" } */
+ i = _mm_comi_round_sd (m128d, m128d, 4, 3); /* { dg-error "incorrect rounding operand" } */
+}
--- /dev/null
+/* { dg-do compile { target { ! { ia32 } } } } */
+/* { dg-options "-O0 -mavx512f" } */
+
+#include <x86intrin.h>
+
+long long l;
+unsigned long long ul;
+__m128d m128d;
+__m128 m128;
+
+void
+test_round_64 (void)
+{
+ m128d = _mm_cvt_roundu64_sd (m128d, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ m128d = _mm_cvt_roundi64_sd (m128d, 4, 7); /* { dg-error "incorrect rounding operand" } */
+
+ m128 = _mm_cvt_roundu64_ss (m128, 4, 7); /* { dg-error "incorrect rounding operand" } */
+ m128 = _mm_cvt_roundi64_ss (m128, 4, 7); /* { dg-error "incorrect rounding operand" } */
+
+ ul = _mm_cvt_roundss_u64 (m128, 7); /* { dg-error "incorrect rounding operand" } */
+ l = _mm_cvt_roundss_i64 (m128, 7); /* { dg-error "incorrect rounding operand" } */
+
+ ul = _mm_cvt_roundsd_u64 (m128d, 7); /* { dg-error "incorrect rounding operand" } */
+ l = _mm_cvt_roundsd_i64 (m128d, 7); /* { dg-error "incorrect rounding operand" } */
+
+ ul = _mm_cvtt_roundss_u64 (m128, 7); /* { dg-error "incorrect rounding operand" } */
+ l = _mm_cvtt_roundss_i64 (m128, 7); /* { dg-error "incorrect rounding operand" } */
+
+ ul = _mm_cvtt_roundsd_u64 (m128d, 7); /* { dg-error "incorrect rounding operand" } */
+ l = _mm_cvtt_roundsd_i64 (m128d, 7); /* { dg-error "incorrect rounding operand" } */
+}
+
+void
+test_round_sae_64 (void)
+{
+ m128d = _mm_cvt_roundu64_sd (m128d, 4, 5); /* { dg-error "incorrect rounding operand" } */
+ m128d = _mm_cvt_roundi64_sd (m128d, 4, 5); /* { dg-error "incorrect rounding operand" } */
+
+ m128 = _mm_cvt_roundu64_ss (m128, 4, 5); /* { dg-error "incorrect rounding operand" } */
+ m128 = _mm_cvt_roundi64_ss (m128, 4, 5); /* { dg-error "incorrect rounding operand" } */
+
+ ul = _mm_cvt_roundss_u64 (m128, 5); /* { dg-error "incorrect rounding operand" } */
+ l = _mm_cvt_roundss_i64 (m128, 5); /* { dg-error "incorrect rounding operand" } */
+
+ ul = _mm_cvt_roundsd_u64 (m128d, 5); /* { dg-error "incorrect rounding operand" } */
+ l = _mm_cvt_roundsd_i64 (m128d, 5); /* { dg-error "incorrect rounding operand" } */
+}
+
+void
+test_sae_only_64 (void)
+{
+ ul = _mm_cvtt_roundss_u64 (m128, 3); /* { dg-error "incorrect rounding operand" } */
+ l = _mm_cvtt_roundss_i64 (m128, 3); /* { dg-error "incorrect rounding operand" } */
+
+ ul = _mm_cvtt_roundsd_u64 (m128d, 3); /* { dg-error "incorrect rounding operand" } */
+ l = _mm_cvtt_roundsd_i64 (m128d, 3); /* { dg-error "incorrect rounding operand" } */
+}
--- /dev/null
+# Copyright (C) 2009-2013 Free Software Foundation, Inc.
+
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with GCC; see the file COPYING3. If not see
+# <http://www.gnu.org/licenses/>.
+
+# The x86-64 AVX512F ABI testsuite needs one additional assembler file for most
+# testcases. For simplicity we will just link it into each test.
+
+load_lib c-torture.exp
+load_lib target-supports.exp
+load_lib torture-options.exp
+
+if { (![istarget x86_64-*-*] && ![istarget i?86-*-*])
+ || ![is-effective-target lp64]
+ || ![is-effective-target avx512f] } then {
+ return
+}
+
+
+# If the linker used understands -M <mapfile>, pass it to clear hardware
+# capabilities set by the Sun assembler.
+set flags ""
+set clearcap_ldflags "-Wl,-M,$srcdir/gcc.target/i386/clearcap.map"
+
+if [check_no_compiler_messages mapfile executable {
+ int main (void) { return 0; }
+ } $clearcap_ldflags ] {
+ set flags $clearcap_ldflags
+}
+
+torture-init
+set-torture-options $C_TORTURE_OPTIONS
+set additional_flags "-W -Wall -mavx512f $flags"
+
+foreach src [lsort [glob -nocomplain $srcdir/$subdir/test_*.c]] {
+ if {[runtest_file_p $runtests $src]} {
+ if { ([istarget *-*-darwin*]) } then {
+ # FIXME: Darwin isn't tested.
+ c-torture-execute [list $src \
+ $srcdir/$subdir/asm-support-darwin.s] \
+ $additional_flags
+ } else {
+ c-torture-execute [list $src \
+ $srcdir/$subdir/asm-support.S] \
+ $additional_flags
+ }
+ }
+}
+
+torture-finish
--- /dev/null
+#ifndef INCLUDED_ARGS_H
+#define INCLUDED_ARGS_H
+
+#include <immintrin.h>
+#include <string.h>
+
+/* Assertion macro. */
+#define assert(test) if (!(test)) abort()
+
+#ifdef __GNUC__
+#define ATTRIBUTE_UNUSED __attribute__((__unused__))
+#else
+#define ATTRIBUTE_UNUSED
+#endif
+
+/* This defines the calling sequences for integers and floats. */
+#define I0 rdi
+#define I1 rsi
+#define I2 rdx
+#define I3 rcx
+#define I4 r8
+#define I5 r9
+#define F0 zmm0
+#define F1 zmm1
+#define F2 zmm2
+#define F3 zmm3
+#define F4 zmm4
+#define F5 zmm5
+#define F6 zmm6
+#define F7 zmm7
+
+typedef union {
+ float _float[16];
+ double _double[8];
+ long _long[8];
+ int _int[16];
+ unsigned long _ulong[8];
+ __m64 _m64[8];
+ __m128 _m128[4];
+ __m256 _m256[2];
+ __m512 _m512[1];
+} ZMM_T;
+
+typedef union {
+ float _float;
+ double _double;
+ long double _ldouble;
+ unsigned long _ulong[2];
+} X87_T;
+extern void (*callthis)(void);
+extern unsigned long rax,rbx,rcx,rdx,rsi,rdi,rsp,rbp,r8,r9,r10,r11,r12,r13,r14,r15;
+ZMM_T zmm_regs[32];
+X87_T x87_regs[8];
+extern volatile unsigned long volatile_var;
+extern void snapshot (void);
+extern void snapshot_ret (void);
+#define WRAP_CALL(N) \
+ (callthis = (void (*)()) (N), (typeof (&N)) snapshot)
+#define WRAP_RET(N) \
+ (callthis = (void (*)()) (N), (typeof (&N)) snapshot_ret)
+
+/* Clear all integer registers. */
+#define clear_int_hardware_registers \
+ asm __volatile__ ("xor %%rax, %%rax\n\t" \
+ "xor %%rbx, %%rbx\n\t" \
+ "xor %%rcx, %%rcx\n\t" \
+ "xor %%rdx, %%rdx\n\t" \
+ "xor %%rsi, %%rsi\n\t" \
+ "xor %%rdi, %%rdi\n\t" \
+ "xor %%r8, %%r8\n\t" \
+ "xor %%r9, %%r9\n\t" \
+ "xor %%r10, %%r10\n\t" \
+ "xor %%r11, %%r11\n\t" \
+ "xor %%r12, %%r12\n\t" \
+ "xor %%r13, %%r13\n\t" \
+ "xor %%r14, %%r14\n\t" \
+ "xor %%r15, %%r15\n\t" \
+ ::: "rax", "rbx", "rcx", "rdx", "rsi", "rdi", "r8", \
+ "r9", "r10", "r11", "r12", "r13", "r14", "r15");
+
+/* This is the list of registers available for passing arguments. Not all of
+ these are used or even really available. */
+struct IntegerRegisters
+{
+ unsigned long rax, rbx, rcx, rdx, rsi, rdi, r8, r9, r10, r11, r12, r13, r14, r15;
+};
+struct FloatRegisters
+{
+ double mm0, mm1, mm2, mm3, mm4, mm5, mm6, mm7;
+ long double st0, st1, st2, st3, st4, st5, st6, st7;
+ ZMM_T zmm0, zmm1, zmm2, zmm3, zmm4, zmm5, zmm6, zmm7, zmm8, zmm9,
+ zmm10, zmm11, zmm12, zmm13, zmm14, zmm15, zmm16, zmm17, zmm18,
+ zmm19, zmm20, zmm21, zmm22, zmm23, zmm24, zmm25, zmm26, zmm27,
+ zmm28, zmm29, zmm30, zmm31;
+};
+
+/* Implemented in scalarargs.c */
+extern struct IntegerRegisters iregs;
+extern struct FloatRegisters fregs;
+extern unsigned int num_iregs, num_fregs;
+
+#define check_int_arguments do { \
+ assert (num_iregs <= 0 || iregs.I0 == I0); \
+ assert (num_iregs <= 1 || iregs.I1 == I1); \
+ assert (num_iregs <= 2 || iregs.I2 == I2); \
+ assert (num_iregs <= 3 || iregs.I3 == I3); \
+ assert (num_iregs <= 4 || iregs.I4 == I4); \
+ assert (num_iregs <= 5 || iregs.I5 == I5); \
+ } while (0)
+
+#define check_char_arguments check_int_arguments
+#define check_short_arguments check_int_arguments
+#define check_long_arguments check_int_arguments
+
+/* Clear register struct. */
+#define clear_struct_registers \
+ rax = rbx = rcx = rdx = rdi = rsi = rbp = rsp \
+ = r8 = r9 = r10 = r11 = r12 = r13 = r14 = r15 = 0; \
+ memset (&iregs, 0, sizeof (iregs)); \
+ memset (&fregs, 0, sizeof (fregs)); \
+ memset (zmm_regs, 0, sizeof (zmm_regs)); \
+ memset (x87_regs, 0, sizeof (x87_regs));
+
+/* Clear both hardware and register structs for integers. */
+#define clear_int_registers \
+ clear_struct_registers \
+ clear_int_hardware_registers
+
+/* TODO: Do the checking. */
+#define check_f_arguments(T) do { \
+ assert (num_fregs <= 0 || fregs.zmm0._ ## T [0] == zmm_regs[0]._ ## T [0]); \
+ assert (num_fregs <= 1 || fregs.zmm1._ ## T [0] == zmm_regs[1]._ ## T [0]); \
+ assert (num_fregs <= 2 || fregs.zmm2._ ## T [0] == zmm_regs[2]._ ## T [0]); \
+ assert (num_fregs <= 3 || fregs.zmm3._ ## T [0] == zmm_regs[3]._ ## T [0]); \
+ assert (num_fregs <= 4 || fregs.zmm4._ ## T [0] == zmm_regs[4]._ ## T [0]); \
+ assert (num_fregs <= 5 || fregs.zmm5._ ## T [0] == zmm_regs[5]._ ## T [0]); \
+ assert (num_fregs <= 6 || fregs.zmm6._ ## T [0] == zmm_regs[6]._ ## T [0]); \
+ assert (num_fregs <= 7 || fregs.zmm7._ ## T [0] == zmm_regs[7]._ ## T [0]); \
+ } while (0)
+
+#define check_float_arguments check_f_arguments(float)
+#define check_double_arguments check_f_arguments(double)
+
+#define check_vector_arguments(T,O) do { \
+ assert (num_fregs <= 0 \
+ || memcmp (((char *) &fregs.zmm0) + (O), \
+ &zmm_regs[0], \
+ sizeof (__ ## T) - (O)) == 0); \
+ assert (num_fregs <= 1 \
+ || memcmp (((char *) &fregs.zmm1) + (O), \
+ &zmm_regs[1], \
+ sizeof (__ ## T) - (O)) == 0); \
+ assert (num_fregs <= 2 \
+ || memcmp (((char *) &fregs.zmm2) + (O), \
+ &zmm_regs[2], \
+ sizeof (__ ## T) - (O)) == 0); \
+ assert (num_fregs <= 3 \
+ || memcmp (((char *) &fregs.zmm3) + (O), \
+ &zmm_regs[3], \
+ sizeof (__ ## T) - (O)) == 0); \
+ assert (num_fregs <= 4 \
+ || memcmp (((char *) &fregs.zmm4) + (O), \
+ &zmm_regs[4], \
+ sizeof (__ ## T) - (O)) == 0); \
+ assert (num_fregs <= 5 \
+ || memcmp (((char *) &fregs.zmm5) + (O), \
+ &zmm_regs[5], \
+ sizeof (__ ## T) - (O)) == 0); \
+ assert (num_fregs <= 6 \
+ || memcmp (((char *) &fregs.zmm6) + (O), \
+ &zmm_regs[6], \
+ sizeof (__ ## T) - (O)) == 0); \
+ assert (num_fregs <= 7 \
+ || memcmp (((char *) &fregs.zmm7) + (O), \
+ &zmm_regs[7], \
+ sizeof (__ ## T) - (O)) == 0); \
+ } while (0)
+
+#define check_m64_arguments check_vector_arguments(m64, 0)
+#define check_m128_arguments check_vector_arguments(m128, 0)
+#define check_m256_arguments check_vector_arguments(m256, 0)
+#define check_m512_arguments check_vector_arguments(m512, 0)
+
+#endif /* INCLUDED_ARGS_H */
--- /dev/null
+ .file "snapshot.S"
+ .text
+ .p2align 4,,15
+.globl snapshot
+ .type snapshot, @function
+snapshot:
+.LFB3:
+ movq %rax, rax(%rip)
+ movq %rbx, rbx(%rip)
+ movq %rcx, rcx(%rip)
+ movq %rdx, rdx(%rip)
+ movq %rdi, rdi(%rip)
+ movq %rsi, rsi(%rip)
+ movq %rbp, rbp(%rip)
+ movq %rsp, rsp(%rip)
+ movq %r8, r8(%rip)
+ movq %r9, r9(%rip)
+ movq %r10, r10(%rip)
+ movq %r11, r11(%rip)
+ movq %r12, r12(%rip)
+ movq %r13, r13(%rip)
+ movq %r14, r14(%rip)
+ movq %r15, r15(%rip)
+ vmovdqu32 %zmm0, zmm_regs+0(%rip)
+ vmovdqu32 %zmm1, zmm_regs+64(%rip)
+ vmovdqu32 %zmm2, zmm_regs+128(%rip)
+ vmovdqu32 %zmm3, zmm_regs+192(%rip)
+ vmovdqu32 %zmm4, zmm_regs+256(%rip)
+ vmovdqu32 %zmm5, zmm_regs+320(%rip)
+ vmovdqu32 %zmm6, zmm_regs+384(%rip)
+ vmovdqu32 %zmm7, zmm_regs+448(%rip)
+ vmovdqu32 %zmm8, zmm_regs+512(%rip)
+ vmovdqu32 %zmm9, zmm_regs+576(%rip)
+ vmovdqu32 %zmm10, zmm_regs+640(%rip)
+ vmovdqu32 %zmm11, zmm_regs+704(%rip)
+ vmovdqu32 %zmm12, zmm_regs+768(%rip)
+ vmovdqu32 %zmm13, zmm_regs+832(%rip)
+ vmovdqu32 %zmm14, zmm_regs+896(%rip)
+ vmovdqu32 %zmm15, zmm_regs+960(%rip)
+ vmovdqu32 %zmm16, zmm_regs+1024(%rip)
+ vmovdqu32 %zmm17, zmm_regs+1088(%rip)
+ vmovdqu32 %zmm18, zmm_regs+1152(%rip)
+ vmovdqu32 %zmm19, zmm_regs+1216(%rip)
+ vmovdqu32 %zmm20, zmm_regs+1280(%rip)
+ vmovdqu32 %zmm21, zmm_regs+1344(%rip)
+ vmovdqu32 %zmm22, zmm_regs+1408(%rip)
+ vmovdqu32 %zmm23, zmm_regs+1472(%rip)
+ vmovdqu32 %zmm24, zmm_regs+1536(%rip)
+ vmovdqu32 %zmm25, zmm_regs+1600(%rip)
+ vmovdqu32 %zmm26, zmm_regs+1664(%rip)
+ vmovdqu32 %zmm27, zmm_regs+1728(%rip)
+ vmovdqu32 %zmm28, zmm_regs+1792(%rip)
+ vmovdqu32 %zmm29, zmm_regs+1856(%rip)
+ vmovdqu32 %zmm30, zmm_regs+1920(%rip)
+ vmovdqu32 %zmm31, zmm_regs+1984(%rip)
+ jmp *callthis(%rip)
+.LFE3:
+ .size snapshot, .-snapshot
+
+ .p2align 4,,15
+.globl snapshot_ret
+ .type snapshot_ret, @function
+snapshot_ret:
+ movq %rdi, rdi(%rip)
+ subq $8, %rsp
+ call *callthis(%rip)
+ addq $8, %rsp
+ movq %rax, rax(%rip)
+ movq %rdx, rdx(%rip)
+ vmovdqu32 %zmm0, zmm_regs+0(%rip)
+ vmovdqu32 %zmm1, zmm_regs+64(%rip)
+ fstpt x87_regs(%rip)
+ fstpt x87_regs+16(%rip)
+ fldt x87_regs+16(%rip)
+ fldt x87_regs(%rip)
+ ret
+ .size snapshot_ret, .-snapshot_ret
+
+ .comm callthis,8,8
+ .comm rax,8,8
+ .comm rbx,8,8
+ .comm rcx,8,8
+ .comm rdx,8,8
+ .comm rsi,8,8
+ .comm rdi,8,8
+ .comm rsp,8,8
+ .comm rbp,8,8
+ .comm r8,8,8
+ .comm r9,8,8
+ .comm r10,8,8
+ .comm r11,8,8
+ .comm r12,8,8
+ .comm r13,8,8
+ .comm r14,8,8
+ .comm r15,8,8
+ .comm zmm_regs,2048,64
+ .comm x87_regs,128,32
+ .comm volatile_var,8,8
--- /dev/null
+#include <stdlib.h>
+#include "cpuid.h"
+
+static void avx512f_test (void);
+
+int
+main ()
+{
+ unsigned int eax, ebx, ecx, edx;
+
+ if (!__get_cpuid (1, &eax, &ebx, &ecx, &edx))
+ return 0;
+
+#define DEBUG
+ /* Run AVX test only if host has AVX support. */
+ if ((ecx & bit_OSXSAVE) == bit_OSXSAVE)
+ {
+ if (__get_cpuid_max (0, NULL) < 7)
+ return 0;
+
+ __cpuid_count (7, 0, eax, ebx, ecx, edx);
+
+ if ((ebx & bit_AVX512F) == bit_AVX512F)
+ {
+ avx512f_test ();
+#ifdef DEBUG
+ printf ("PASSED\n");
+#endif
+ }
+#ifdef DEBUG
+ else
+ printf ("SKIPPED\n");
+#endif
+ }
+#ifdef DEBUG
+ else
+ printf ("SKIPPED\n");
+#endif
+
+ return 0;
+}
--- /dev/null
+#include <stdio.h>
+#include "avx512f-check.h"
+#include "args.h"
+
+struct IntegerRegisters iregs;
+struct FloatRegisters fregs;
+unsigned int num_iregs, num_fregs;
+
+__m512
+fun_test_returning___m512 (void)
+{
+ volatile_var++;
+ return (__m512){73,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0};
+}
+
+__m512 test_512;
+
+static void
+avx512f_test (void)
+{
+ unsigned failed = 0;
+ ZMM_T zmmt1, zmmt2;
+
+ clear_struct_registers;
+ test_512 = (__m512){73,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0};
+ zmmt1._m512[0] = test_512;
+ zmmt2._m512[0] = WRAP_RET (fun_test_returning___m512)();
+ if (memcmp (&zmmt1, &zmmt2, sizeof (zmmt2)) != 0)
+ printf ("fail m512\n"), failed++;
+ if (failed)
+ abort ();
+}
--- /dev/null
+#include <stdio.h>
+#include "avx512f-check.h"
+#include "args.h"
+
+struct IntegerRegisters iregs;
+struct FloatRegisters fregs;
+unsigned int num_iregs, num_fregs;
+
+/* This struct holds values for argument checking. */
+struct
+{
+ ZMM_T i0, i1, i2, i3, i4, i5, i6, i7, i8, i9, i10, i11, i12, i13, i14, i15, i16, i17, i18, i19, i20, i21, i22, i23;
+} values;
+
+char *pass;
+int failed = 0;
+
+#undef assert
+#define assert(c) do { \
+ if (!(c)) {failed++; printf ("failed %s\n", pass); } \
+} while (0)
+
+#define compare(X1,X2,T) do { \
+ assert (memcmp (&X1, &X2, sizeof (T)) == 0); \
+} while (0)
+
+fun_check_passing_m512_8_values (__m512 i0 ATTRIBUTE_UNUSED, __m512 i1 ATTRIBUTE_UNUSED, __m512 i2 ATTRIBUTE_UNUSED, __m512 i3 ATTRIBUTE_UNUSED, __m512 i4 ATTRIBUTE_UNUSED, __m512 i5 ATTRIBUTE_UNUSED, __m512 i6 ATTRIBUTE_UNUSED, __m512 i7 ATTRIBUTE_UNUSED)
+{
+ /* Check argument values. */
+ compare (values.i0, i0, __m512);
+ compare (values.i1, i1, __m512);
+ compare (values.i2, i2, __m512);
+ compare (values.i3, i3, __m512);
+ compare (values.i4, i4, __m512);
+ compare (values.i5, i5, __m512);
+ compare (values.i6, i6, __m512);
+ compare (values.i7, i7, __m512);
+}
+
+void
+fun_check_passing_m512_8_regs (__m512 i0 ATTRIBUTE_UNUSED, __m512 i1 ATTRIBUTE_UNUSED, __m512 i2 ATTRIBUTE_UNUSED, __m512 i3 ATTRIBUTE_UNUSED, __m512 i4 ATTRIBUTE_UNUSED, __m512 i5 ATTRIBUTE_UNUSED, __m512 i6 ATTRIBUTE_UNUSED, __m512 i7 ATTRIBUTE_UNUSED)
+{
+ /* Check register contents. */
+ check_m512_arguments;
+}
+
+void
+fun_check_passing_m512_20_values (__m512 i0 ATTRIBUTE_UNUSED, __m512 i1 ATTRIBUTE_UNUSED, __m512 i2 ATTRIBUTE_UNUSED, __m512 i3 ATTRIBUTE_UNUSED, __m512 i4 ATTRIBUTE_UNUSED, __m512 i5 ATTRIBUTE_UNUSED, __m512 i6 ATTRIBUTE_UNUSED, __m512 i7 ATTRIBUTE_UNUSED, __m512 i8 ATTRIBUTE_UNUSED, __m512 i9 ATTRIBUTE_UNUSED, __m512 i10 ATTRIBUTE_UNUSED, __m512 i11 ATTRIBUTE_UNUSED, __m512 i12 ATTRIBUTE_UNUSED, __m512 i13 ATTRIBUTE_UNUSED, __m512 i14 ATTRIBUTE_UNUSED, __m512 i15 ATTRIBUTE_UNUSED, __m512 i16 ATTRIBUTE_UNUSED, __m512 i17 ATTRIBUTE_UNUSED, __m512 i18 ATTRIBUTE_UNUSED, __m512 i19 ATTRIBUTE_UNUSED)
+{
+ /* Check argument values. */
+ compare (values.i0, i0, __m512);
+ compare (values.i1, i1, __m512);
+ compare (values.i2, i2, __m512);
+ compare (values.i3, i3, __m512);
+ compare (values.i4, i4, __m512);
+ compare (values.i5, i5, __m512);
+ compare (values.i6, i6, __m512);
+ compare (values.i7, i7, __m512);
+ compare (values.i8, i8, __m512);
+ compare (values.i9, i9, __m512);
+ compare (values.i10, i10, __m512);
+ compare (values.i11, i11, __m512);
+ compare (values.i12, i12, __m512);
+ compare (values.i13, i13, __m512);
+ compare (values.i14, i14, __m512);
+ compare (values.i15, i15, __m512);
+ compare (values.i16, i16, __m512);
+ compare (values.i17, i17, __m512);
+ compare (values.i18, i18, __m512);
+ compare (values.i19, i19, __m512);
+}
+
+void
+fun_check_passing_m512_20_regs (__m512 i0 ATTRIBUTE_UNUSED, __m512 i1 ATTRIBUTE_UNUSED, __m512 i2 ATTRIBUTE_UNUSED, __m512 i3 ATTRIBUTE_UNUSED, __m512 i4 ATTRIBUTE_UNUSED, __m512 i5 ATTRIBUTE_UNUSED, __m512 i6 ATTRIBUTE_UNUSED, __m512 i7 ATTRIBUTE_UNUSED, __m512 i8 ATTRIBUTE_UNUSED, __m512 i9 ATTRIBUTE_UNUSED, __m512 i10 ATTRIBUTE_UNUSED, __m512 i11 ATTRIBUTE_UNUSED, __m512 i12 ATTRIBUTE_UNUSED, __m512 i13 ATTRIBUTE_UNUSED, __m512 i14 ATTRIBUTE_UNUSED, __m512 i15 ATTRIBUTE_UNUSED, __m512 i16 ATTRIBUTE_UNUSED, __m512 i17 ATTRIBUTE_UNUSED, __m512 i18 ATTRIBUTE_UNUSED, __m512 i19 ATTRIBUTE_UNUSED)
+{
+ /* Check register contents. */
+ check_m512_arguments;
+}
+
+
+#define def_check_passing8(_i0, _i1, _i2, _i3, _i4, _i5, _i6, _i7, _func1, _func2, TYPE) \
+ values.i0.TYPE[0] = _i0; \
+ values.i1.TYPE[0] = _i1; \
+ values.i2.TYPE[0] = _i2; \
+ values.i3.TYPE[0] = _i3; \
+ values.i4.TYPE[0] = _i4; \
+ values.i5.TYPE[0] = _i5; \
+ values.i6.TYPE[0] = _i6; \
+ values.i7.TYPE[0] = _i7; \
+ WRAP_CALL(_func1) (_i0, _i1, _i2, _i3, _i4, _i5, _i6, _i7); \
+ \
+ clear_struct_registers; \
+ fregs.F0.TYPE[0] = _i0; \
+ fregs.F1.TYPE[0] = _i1; \
+ fregs.F2.TYPE[0] = _i2; \
+ fregs.F3.TYPE[0] = _i3; \
+ fregs.F4.TYPE[0] = _i4; \
+ fregs.F5.TYPE[0] = _i5; \
+ fregs.F6.TYPE[0] = _i6; \
+ fregs.F7.TYPE[0] = _i7; \
+ num_fregs = 8; \
+ WRAP_CALL(_func2) (_i0, _i1, _i2, _i3, _i4, _i5, _i6, _i7);
+
+#define def_check_passing20(_i0, _i1, _i2, _i3, _i4, _i5, _i6, _i7, _i8, _i9, _i10, _i11, _i12, _i13, _i14, _i15, _i16, _i17, _i18, _i19, _func1, _func2, TYPE) \
+ values.i0.TYPE[0] = _i0; \
+ values.i1.TYPE[0] = _i1; \
+ values.i2.TYPE[0] = _i2; \
+ values.i3.TYPE[0] = _i3; \
+ values.i4.TYPE[0] = _i4; \
+ values.i5.TYPE[0] = _i5; \
+ values.i6.TYPE[0] = _i6; \
+ values.i7.TYPE[0] = _i7; \
+ values.i8.TYPE[0] = _i8; \
+ values.i9.TYPE[0] = _i9; \
+ values.i10.TYPE[0] = _i10; \
+ values.i11.TYPE[0] = _i11; \
+ values.i12.TYPE[0] = _i12; \
+ values.i13.TYPE[0] = _i13; \
+ values.i14.TYPE[0] = _i14; \
+ values.i15.TYPE[0] = _i15; \
+ values.i16.TYPE[0] = _i16; \
+ values.i17.TYPE[0] = _i17; \
+ values.i18.TYPE[0] = _i18; \
+ values.i19.TYPE[0] = _i19; \
+ WRAP_CALL(_func1) (_i0, _i1, _i2, _i3, _i4, _i5, _i6, _i7, _i8, _i9, _i10, _i11, _i12, _i13, _i14, _i15, _i16, _i17, _i18, _i19); \
+ \
+ clear_struct_registers; \
+ fregs.F0.TYPE[0] = _i0; \
+ fregs.F1.TYPE[0] = _i1; \
+ fregs.F2.TYPE[0] = _i2; \
+ fregs.F3.TYPE[0] = _i3; \
+ fregs.F4.TYPE[0] = _i4; \
+ fregs.F5.TYPE[0] = _i5; \
+ fregs.F6.TYPE[0] = _i6; \
+ fregs.F7.TYPE[0] = _i7; \
+ num_fregs = 8; \
+ WRAP_CALL(_func2) (_i0, _i1, _i2, _i3, _i4, _i5, _i6, _i7, _i8, _i9, _i10, _i11, _i12, _i13, _i14, _i15, _i16, _i17, _i18, _i19);
+
+void
+test_m512_on_stack ()
+{
+ __m512 x[8];
+ int i;
+ for (i = 0; i < 8; i++)
+ x[i] = (__m512){32+i, 0, 0, 0, 0, 0, 0, 0};
+ pass = "m512-8";
+ def_check_passing8(x[0], x[1], x[2], x[3], x[4], x[5], x[6], x[7], fun_check_passing_m512_8_values, fun_check_passing_m512_8_regs, _m512);
+}
+
+void
+test_too_many_m512 ()
+{
+ __m512 x[20];
+ int i;
+ for (i = 0; i < 20; i++)
+ x[i] = (__m512){32+i, 0, 0, 0, 0, 0, 0, 0};
+ pass = "m512-20";
+ def_check_passing20(x[0], x[1], x[2], x[3], x[4], x[5], x[6], x[7], x[8], x[9], x[10], x[11], x[12], x[13], x[14], x[15], x[16], x[17], x[18], x[19], fun_check_passing_m512_20_values, fun_check_passing_m512_20_regs, _m512);
+}
+
+static void
+avx512f_test (void)
+{
+ test_m512_on_stack ();
+ test_too_many_m512 ();
+ if (failed)
+ abort ();
+}
--- /dev/null
+#include "avx512f-check.h"
+#include "args.h"
+
+struct IntegerRegisters iregs;
+struct FloatRegisters fregs;
+unsigned int num_iregs, num_fregs;
+
+struct m512_struct
+{
+ __m512 x;
+};
+
+struct m512_2_struct
+{
+ __m512 x1, x2;
+};
+
+/* Check that the struct is passed as the individual members in fregs. */
+void
+check_struct_passing1 (struct m512_struct ms1 ATTRIBUTE_UNUSED,
+ struct m512_struct ms2 ATTRIBUTE_UNUSED,
+ struct m512_struct ms3 ATTRIBUTE_UNUSED,
+ struct m512_struct ms4 ATTRIBUTE_UNUSED,
+ struct m512_struct ms5 ATTRIBUTE_UNUSED,
+ struct m512_struct ms6 ATTRIBUTE_UNUSED,
+ struct m512_struct ms7 ATTRIBUTE_UNUSED,
+ struct m512_struct ms8 ATTRIBUTE_UNUSED)
+{
+ /* Check the passing on the stack by comparing the address of the
+ stack elements to the expected place on the stack. */
+ assert ((unsigned long)&ms1.x == rsp+8);
+ assert ((unsigned long)&ms2.x == rsp+72);
+ assert ((unsigned long)&ms3.x == rsp+136);
+ assert ((unsigned long)&ms4.x == rsp+200);
+ assert ((unsigned long)&ms5.x == rsp+264);
+ assert ((unsigned long)&ms6.x == rsp+328);
+ assert ((unsigned long)&ms7.x == rsp+392);
+ assert ((unsigned long)&ms8.x == rsp+456);
+}
+
+void
+check_struct_passing2 (struct m512_2_struct ms ATTRIBUTE_UNUSED)
+{
+ /* Check the passing on the stack by comparing the address of the
+ stack elements to the expected place on the stack. */
+ assert ((unsigned long)&ms.x1 == rsp+8);
+ assert ((unsigned long)&ms.x2 == rsp+72);
+}
+
+static void
+avx512f_test (void)
+{
+ struct m512_struct m512s [8];
+ struct m512_2_struct m512_2s = {
+ { 48.394, 39.3, -397.9, 3484.9, -8.394, -93.3, 7.9, 84.94,
+ 48.3941, 39.31, -397.91, 3484.91, -8.3941, -93.31, 7.91, 84.941 },
+ { -8.394, -3.3, -39.9, 34.9, 7.9, 84.94, -48.394, 39.3,
+ -8.3942, -3.32, -39.92, 34.92, 7.92, 84.942, -48.3942, 39.32 }
+ };
+ int i;
+
+ for (i = 0; i < 8; i++)
+ m512s[i].x = (__m512){32+i, 0, i, 0, -i, 0, i - 12, i + 8,
+ 32+i, 0, i, 0, -i, 0, i - 12, i + 8};
+
+ clear_struct_registers;
+ for (i = 0; i < 8; i++)
+ (&fregs.zmm0)[i]._m512[0] = m512s[i].x;
+ num_fregs = 8;
+ WRAP_CALL (check_struct_passing1)(m512s[0], m512s[1], m512s[2], m512s[3],
+ m512s[4], m512s[5], m512s[6], m512s[7]);
+ WRAP_CALL (check_struct_passing2)(m512_2s);
+}
--- /dev/null
+#include "avx512f-check.h"
+#include "args.h"
+
+struct IntegerRegisters iregs;
+struct FloatRegisters fregs;
+unsigned int num_iregs, num_fregs;
+
+union un1
+{
+ __m512 x;
+ float f;
+};
+
+union un2
+{
+ __m512 x;
+ double d;
+};
+
+union un3
+{
+ __m512 x;
+ __m128 v;
+};
+
+union un4
+{
+ __m512 x;
+ long double ld;
+};
+
+union un5
+{
+ __m512 x;
+ int i;
+};
+
+union un6
+{
+ __m512 x;
+ __m256 v;
+};
+
+
+void
+check_union_passing1(union un1 u1 ATTRIBUTE_UNUSED,
+ union un1 u2 ATTRIBUTE_UNUSED,
+ union un1 u3 ATTRIBUTE_UNUSED,
+ union un1 u4 ATTRIBUTE_UNUSED,
+ union un1 u5 ATTRIBUTE_UNUSED,
+ union un1 u6 ATTRIBUTE_UNUSED,
+ union un1 u7 ATTRIBUTE_UNUSED,
+ union un1 u8 ATTRIBUTE_UNUSED)
+{
+ /* Check the passing on the stack by comparing the address of the
+ stack elements to the expected place on the stack. */
+ assert ((unsigned long)&u1.x == rsp+8);
+ assert ((unsigned long)&u1.f == rsp+8);
+ assert ((unsigned long)&u2.x == rsp+72);
+ assert ((unsigned long)&u2.f == rsp+72);
+ assert ((unsigned long)&u3.x == rsp+136);
+ assert ((unsigned long)&u3.f == rsp+136);
+ assert ((unsigned long)&u4.x == rsp+200);
+ assert ((unsigned long)&u4.f == rsp+200);
+ assert ((unsigned long)&u5.x == rsp+264);
+ assert ((unsigned long)&u5.f == rsp+264);
+ assert ((unsigned long)&u6.x == rsp+328);
+ assert ((unsigned long)&u6.f == rsp+328);
+ assert ((unsigned long)&u7.x == rsp+392);
+ assert ((unsigned long)&u7.f == rsp+392);
+ assert ((unsigned long)&u8.x == rsp+456);
+ assert ((unsigned long)&u8.f == rsp+456);
+}
+
+void
+check_union_passing2(union un2 u1 ATTRIBUTE_UNUSED,
+ union un2 u2 ATTRIBUTE_UNUSED,
+ union un2 u3 ATTRIBUTE_UNUSED,
+ union un2 u4 ATTRIBUTE_UNUSED,
+ union un2 u5 ATTRIBUTE_UNUSED,
+ union un2 u6 ATTRIBUTE_UNUSED,
+ union un2 u7 ATTRIBUTE_UNUSED,
+ union un2 u8 ATTRIBUTE_UNUSED)
+{
+ /* Check the passing on the stack by comparing the address of the
+ stack elements to the expected place on the stack. */
+ assert ((unsigned long)&u1.x == rsp+8);
+ assert ((unsigned long)&u1.d == rsp+8);
+ assert ((unsigned long)&u2.x == rsp+72);
+ assert ((unsigned long)&u2.d == rsp+72);
+ assert ((unsigned long)&u3.x == rsp+136);
+ assert ((unsigned long)&u3.d == rsp+136);
+ assert ((unsigned long)&u4.x == rsp+200);
+ assert ((unsigned long)&u4.d == rsp+200);
+ assert ((unsigned long)&u5.x == rsp+264);
+ assert ((unsigned long)&u5.d == rsp+264);
+ assert ((unsigned long)&u6.x == rsp+328);
+ assert ((unsigned long)&u6.d == rsp+328);
+ assert ((unsigned long)&u7.x == rsp+392);
+ assert ((unsigned long)&u7.d == rsp+392);
+ assert ((unsigned long)&u8.x == rsp+456);
+ assert ((unsigned long)&u8.d == rsp+456);
+}
+
+void
+check_union_passing3(union un3 u1 ATTRIBUTE_UNUSED,
+ union un3 u2 ATTRIBUTE_UNUSED,
+ union un3 u3 ATTRIBUTE_UNUSED,
+ union un3 u4 ATTRIBUTE_UNUSED,
+ union un3 u5 ATTRIBUTE_UNUSED,
+ union un3 u6 ATTRIBUTE_UNUSED,
+ union un3 u7 ATTRIBUTE_UNUSED,
+ union un3 u8 ATTRIBUTE_UNUSED)
+{
+ /* Check the passing on the stack by comparing the address of the
+ stack elements to the expected place on the stack. */
+ assert ((unsigned long)&u1.x == rsp+8);
+ assert ((unsigned long)&u1.v == rsp+8);
+ assert ((unsigned long)&u2.x == rsp+72);
+ assert ((unsigned long)&u2.v == rsp+72);
+ assert ((unsigned long)&u3.x == rsp+136);
+ assert ((unsigned long)&u3.v == rsp+136);
+ assert ((unsigned long)&u4.x == rsp+200);
+ assert ((unsigned long)&u4.v == rsp+200);
+ assert ((unsigned long)&u5.x == rsp+264);
+ assert ((unsigned long)&u5.v == rsp+264);
+ assert ((unsigned long)&u6.x == rsp+328);
+ assert ((unsigned long)&u6.v == rsp+328);
+ assert ((unsigned long)&u7.x == rsp+392);
+ assert ((unsigned long)&u7.v == rsp+392);
+ assert ((unsigned long)&u8.x == rsp+456);
+ assert ((unsigned long)&u8.v == rsp+456);
+}
+
+void
+check_union_passing4(union un4 u ATTRIBUTE_UNUSED)
+{
+ /* Check the passing on the stack by comparing the address of the
+ stack elements to the expected place on the stack. */
+ assert ((unsigned long)&u.x == rsp+8);
+ assert ((unsigned long)&u.ld == rsp+8);
+}
+
+void
+check_union_passing5(union un5 u ATTRIBUTE_UNUSED)
+{
+ /* Check the passing on the stack by comparing the address of the
+ stack elements to the expected place on the stack. */
+ assert ((unsigned long)&u.x == rsp+8);
+ assert ((unsigned long)&u.i == rsp+8);
+}
+
+void
+check_union_passing6(union un6 u1 ATTRIBUTE_UNUSED,
+ union un6 u2 ATTRIBUTE_UNUSED,
+ union un6 u3 ATTRIBUTE_UNUSED,
+ union un6 u4 ATTRIBUTE_UNUSED,
+ union un6 u5 ATTRIBUTE_UNUSED,
+ union un6 u6 ATTRIBUTE_UNUSED,
+ union un6 u7 ATTRIBUTE_UNUSED,
+ union un6 u8 ATTRIBUTE_UNUSED)
+{
+ assert ((unsigned long)&u1.x == rsp+8);
+ assert ((unsigned long)&u1.v == rsp+8);
+ assert ((unsigned long)&u2.x == rsp+72);
+ assert ((unsigned long)&u2.v == rsp+72);
+ assert ((unsigned long)&u3.x == rsp+136);
+ assert ((unsigned long)&u3.v == rsp+136);
+ assert ((unsigned long)&u4.x == rsp+200);
+ assert ((unsigned long)&u4.v == rsp+200);
+ assert ((unsigned long)&u5.x == rsp+264);
+ assert ((unsigned long)&u5.v == rsp+264);
+ assert ((unsigned long)&u6.x == rsp+328);
+ assert ((unsigned long)&u6.v == rsp+328);
+ assert ((unsigned long)&u7.x == rsp+392);
+ assert ((unsigned long)&u7.v == rsp+392);
+ assert ((unsigned long)&u8.x == rsp+456);
+ assert ((unsigned long)&u8.v == rsp+456);
+}
+
+#define check_union_passing1 WRAP_CALL(check_union_passing1)
+#define check_union_passing2 WRAP_CALL(check_union_passing2)
+#define check_union_passing3 WRAP_CALL(check_union_passing3)
+#define check_union_passing4 WRAP_CALL(check_union_passing4)
+#define check_union_passing5 WRAP_CALL(check_union_passing5)
+#define check_union_passing6 WRAP_CALL(check_union_passing6)
+
+static void
+avx512f_test (void)
+{
+ union un1 u1[8];
+ union un2 u2[8];
+ union un3 u3[8];
+ union un4 u4;
+ union un5 u5;
+ union un6 u6[8];
+ int i;
+
+ for (i = 0; i < 8; i++)
+ u1[i].x = (__m512){32+i, 0, i, 0, -i, 0, i - 12, i + 8,
+ 32+i, 0, i, 0, -i, 0, i - 12, i + 8};
+
+ clear_struct_registers;
+ for (i = 0; i < 8; i++)
+ (&fregs.zmm0)[i]._m512[0] = u1[i].x;
+ num_fregs = 8;
+ check_union_passing1(u1[0], u1[1], u1[2], u1[3],
+ u1[4], u1[5], u1[6], u1[7]);
+
+ clear_struct_registers;
+ for (i = 0; i < 8; i++)
+ {
+ u2[i].x = u1[i].x;
+ (&fregs.zmm0)[i]._m512[0] = u2[i].x;
+ }
+ num_fregs = 8;
+ check_union_passing2(u2[0], u2[1], u2[2], u2[3],
+ u2[4], u2[5], u2[6], u2[7]);
+
+ clear_struct_registers;
+ for (i = 0; i < 8; i++)
+ {
+ u3[i].x = u1[i].x;
+ (&fregs.zmm0)[i]._m512[0] = u3[i].x;
+ }
+ num_fregs = 8;
+ check_union_passing3(u3[0], u3[1], u3[2], u3[3],
+ u3[4], u3[5], u3[6], u3[7]);
+
+ check_union_passing4(u4);
+ check_union_passing5(u5);
+
+ clear_struct_registers;
+ for (i = 0; i < 8; i++)
+ {
+ u6[i].x = u1[i].x;
+ (&fregs.zmm0)[i]._m512[0] = u6[i].x;
+ }
+ num_fregs = 8;
+ check_union_passing6(u6[0], u6[1], u6[2], u6[3],
+ u6[4], u6[5], u6[6], u6[7]);
+}
}
+# Return 1 if avx512f instructions can be compiled.
+
+proc check_effective_target_avx512f { } {
+ return [check_no_compiler_messages avx512f object {
+ typedef double __m512d __attribute__ ((__vector_size__ (64)));
+
+ void _mm512_add (__m512d a)
+ {
+ __builtin_ia32_addpd512_mask (a, a, a, 1, 4);
+ }
+ } "-O2 -mavx512f" ]
+}
+
# Return 1 if avx instructions can be compiled.
proc check_effective_target_avx { } {
projects / platform / upstream / gcc.git / commitdiff