;
; AVX512-LABEL: testv32i8:
; AVX512: # %bb.0:
-; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT: vpmovzxbd {{.*#+}} zmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero,xmm1[8],zero,zero,zero,xmm1[9],zero,zero,zero,xmm1[10],zero,zero,zero,xmm1[11],zero,zero,zero,xmm1[12],zero,zero,zero,xmm1[13],zero,zero,zero,xmm1[14],zero,zero,zero,xmm1[15],zero,zero,zero
+; AVX512-NEXT: vpmovzxbd {{.*#+}} zmm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
; AVX512-NEXT: vplzcntd %zmm1, %zmm1
; AVX512-NEXT: vpmovdb %zmm1, %xmm1
-; AVX512-NEXT: vmovdqa {{.*#+}} xmm2 = [24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24]
-; AVX512-NEXT: vpsubb %xmm2, %xmm1, %xmm1
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm0
; AVX512-NEXT: vpmovzxbd {{.*#+}} zmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
; AVX512-NEXT: vplzcntd %zmm0, %zmm0
; AVX512-NEXT: vpmovdb %zmm0, %xmm0
-; AVX512-NEXT: vpsubb %xmm2, %xmm0, %xmm0
-; AVX512-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
+; AVX512-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm0
+; AVX512-NEXT: vpsubb {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
; AVX512-NEXT: retq
;
; X32-AVX-LABEL: testv32i8:
;
; AVX512-LABEL: testv32i8u:
; AVX512: # %bb.0:
-; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT: vpmovzxbd {{.*#+}} zmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero,xmm1[8],zero,zero,zero,xmm1[9],zero,zero,zero,xmm1[10],zero,zero,zero,xmm1[11],zero,zero,zero,xmm1[12],zero,zero,zero,xmm1[13],zero,zero,zero,xmm1[14],zero,zero,zero,xmm1[15],zero,zero,zero
+; AVX512-NEXT: vpmovzxbd {{.*#+}} zmm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
; AVX512-NEXT: vplzcntd %zmm1, %zmm1
; AVX512-NEXT: vpmovdb %zmm1, %xmm1
-; AVX512-NEXT: vmovdqa {{.*#+}} xmm2 = [24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24]
-; AVX512-NEXT: vpsubb %xmm2, %xmm1, %xmm1
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm0
; AVX512-NEXT: vpmovzxbd {{.*#+}} zmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
; AVX512-NEXT: vplzcntd %zmm0, %zmm0
; AVX512-NEXT: vpmovdb %zmm0, %xmm0
-; AVX512-NEXT: vpsubb %xmm2, %xmm0, %xmm0
-; AVX512-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
+; AVX512-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm0
+; AVX512-NEXT: vpsubb {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
; AVX512-NEXT: retq
;
; X32-AVX-LABEL: testv32i8u:
;
; AVX512CDBW-LABEL: testv32i16:
; AVX512CDBW: # %bb.0:
-; AVX512CDBW-NEXT: vextracti64x4 $1, %zmm0, %ymm1
-; AVX512CDBW-NEXT: vpmovzxwd {{.*#+}} zmm1 = ymm1[0],zero,ymm1[1],zero,ymm1[2],zero,ymm1[3],zero,ymm1[4],zero,ymm1[5],zero,ymm1[6],zero,ymm1[7],zero,ymm1[8],zero,ymm1[9],zero,ymm1[10],zero,ymm1[11],zero,ymm1[12],zero,ymm1[13],zero,ymm1[14],zero,ymm1[15],zero
+; AVX512CDBW-NEXT: vpmovzxwd {{.*#+}} zmm1 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero
; AVX512CDBW-NEXT: vplzcntd %zmm1, %zmm1
; AVX512CDBW-NEXT: vpmovdw %zmm1, %ymm1
-; AVX512CDBW-NEXT: vmovdqa {{.*#+}} ymm2 = [16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
-; AVX512CDBW-NEXT: vpsubw %ymm2, %ymm1, %ymm1
+; AVX512CDBW-NEXT: vextracti64x4 $1, %zmm0, %ymm0
; AVX512CDBW-NEXT: vpmovzxwd {{.*#+}} zmm0 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero
; AVX512CDBW-NEXT: vplzcntd %zmm0, %zmm0
; AVX512CDBW-NEXT: vpmovdw %zmm0, %ymm0
-; AVX512CDBW-NEXT: vpsubw %ymm2, %ymm0, %ymm0
-; AVX512CDBW-NEXT: vinserti64x4 $1, %ymm1, %zmm0, %zmm0
+; AVX512CDBW-NEXT: vinserti64x4 $1, %ymm0, %zmm1, %zmm0
+; AVX512CDBW-NEXT: vpsubw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %zmm0
; AVX512CDBW-NEXT: retq
;
; AVX512BW-LABEL: testv32i16:
;
; AVX512CDBW-LABEL: testv32i16u:
; AVX512CDBW: # %bb.0:
-; AVX512CDBW-NEXT: vextracti64x4 $1, %zmm0, %ymm1
-; AVX512CDBW-NEXT: vpmovzxwd {{.*#+}} zmm1 = ymm1[0],zero,ymm1[1],zero,ymm1[2],zero,ymm1[3],zero,ymm1[4],zero,ymm1[5],zero,ymm1[6],zero,ymm1[7],zero,ymm1[8],zero,ymm1[9],zero,ymm1[10],zero,ymm1[11],zero,ymm1[12],zero,ymm1[13],zero,ymm1[14],zero,ymm1[15],zero
+; AVX512CDBW-NEXT: vpmovzxwd {{.*#+}} zmm1 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero
; AVX512CDBW-NEXT: vplzcntd %zmm1, %zmm1
; AVX512CDBW-NEXT: vpmovdw %zmm1, %ymm1
-; AVX512CDBW-NEXT: vmovdqa {{.*#+}} ymm2 = [16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
-; AVX512CDBW-NEXT: vpsubw %ymm2, %ymm1, %ymm1
+; AVX512CDBW-NEXT: vextracti64x4 $1, %zmm0, %ymm0
; AVX512CDBW-NEXT: vpmovzxwd {{.*#+}} zmm0 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero
; AVX512CDBW-NEXT: vplzcntd %zmm0, %zmm0
; AVX512CDBW-NEXT: vpmovdw %zmm0, %ymm0
-; AVX512CDBW-NEXT: vpsubw %ymm2, %ymm0, %ymm0
-; AVX512CDBW-NEXT: vinserti64x4 $1, %ymm1, %zmm0, %zmm0
+; AVX512CDBW-NEXT: vinserti64x4 $1, %ymm0, %zmm1, %zmm0
+; AVX512CDBW-NEXT: vpsubw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %zmm0
; AVX512CDBW-NEXT: retq
;
; AVX512BW-LABEL: testv32i16u:
; AVX512CD-LABEL: testv64i8:
; AVX512CD: # %bb.0:
; AVX512CD-NEXT: vextracti64x4 $1, %zmm0, %ymm1
-; AVX512CD-NEXT: vextracti128 $1, %ymm1, %xmm2
-; AVX512CD-NEXT: vpmovzxbd {{.*#+}} zmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero,xmm2[4],zero,zero,zero,xmm2[5],zero,zero,zero,xmm2[6],zero,zero,zero,xmm2[7],zero,zero,zero,xmm2[8],zero,zero,zero,xmm2[9],zero,zero,zero,xmm2[10],zero,zero,zero,xmm2[11],zero,zero,zero,xmm2[12],zero,zero,zero,xmm2[13],zero,zero,zero,xmm2[14],zero,zero,zero,xmm2[15],zero,zero,zero
+; AVX512CD-NEXT: vpmovzxbd {{.*#+}} zmm2 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero,xmm1[8],zero,zero,zero,xmm1[9],zero,zero,zero,xmm1[10],zero,zero,zero,xmm1[11],zero,zero,zero,xmm1[12],zero,zero,zero,xmm1[13],zero,zero,zero,xmm1[14],zero,zero,zero,xmm1[15],zero,zero,zero
; AVX512CD-NEXT: vplzcntd %zmm2, %zmm2
; AVX512CD-NEXT: vpmovdb %zmm2, %xmm2
-; AVX512CD-NEXT: vmovdqa {{.*#+}} xmm3 = [24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24]
-; AVX512CD-NEXT: vpsubb %xmm3, %xmm2, %xmm2
+; AVX512CD-NEXT: vextracti128 $1, %ymm1, %xmm1
; AVX512CD-NEXT: vpmovzxbd {{.*#+}} zmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero,xmm1[8],zero,zero,zero,xmm1[9],zero,zero,zero,xmm1[10],zero,zero,zero,xmm1[11],zero,zero,zero,xmm1[12],zero,zero,zero,xmm1[13],zero,zero,zero,xmm1[14],zero,zero,zero,xmm1[15],zero,zero,zero
; AVX512CD-NEXT: vplzcntd %zmm1, %zmm1
; AVX512CD-NEXT: vpmovdb %zmm1, %xmm1
-; AVX512CD-NEXT: vpsubb %xmm3, %xmm1, %xmm1
-; AVX512CD-NEXT: vinserti128 $1, %xmm2, %ymm1, %ymm1
-; AVX512CD-NEXT: vextracti128 $1, %ymm0, %xmm2
-; AVX512CD-NEXT: vpmovzxbd {{.*#+}} zmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero,xmm2[4],zero,zero,zero,xmm2[5],zero,zero,zero,xmm2[6],zero,zero,zero,xmm2[7],zero,zero,zero,xmm2[8],zero,zero,zero,xmm2[9],zero,zero,zero,xmm2[10],zero,zero,zero,xmm2[11],zero,zero,zero,xmm2[12],zero,zero,zero,xmm2[13],zero,zero,zero,xmm2[14],zero,zero,zero,xmm2[15],zero,zero,zero
-; AVX512CD-NEXT: vplzcntd %zmm2, %zmm2
-; AVX512CD-NEXT: vpmovdb %zmm2, %xmm2
-; AVX512CD-NEXT: vpsubb %xmm3, %xmm2, %xmm2
+; AVX512CD-NEXT: vinserti128 $1, %xmm1, %ymm2, %ymm1
+; AVX512CD-NEXT: vmovdqa {{.*#+}} ymm2 = [24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24]
+; AVX512CD-NEXT: vpsubb %ymm2, %ymm1, %ymm1
+; AVX512CD-NEXT: vpmovzxbd {{.*#+}} zmm3 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
+; AVX512CD-NEXT: vplzcntd %zmm3, %zmm3
+; AVX512CD-NEXT: vpmovdb %zmm3, %xmm3
+; AVX512CD-NEXT: vextracti128 $1, %ymm0, %xmm0
; AVX512CD-NEXT: vpmovzxbd {{.*#+}} zmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
; AVX512CD-NEXT: vplzcntd %zmm0, %zmm0
; AVX512CD-NEXT: vpmovdb %zmm0, %xmm0
-; AVX512CD-NEXT: vpsubb %xmm3, %xmm0, %xmm0
-; AVX512CD-NEXT: vinserti128 $1, %xmm2, %ymm0, %ymm0
+; AVX512CD-NEXT: vinserti128 $1, %xmm0, %ymm3, %ymm0
+; AVX512CD-NEXT: vpsubb %ymm2, %ymm0, %ymm0
; AVX512CD-NEXT: vinserti64x4 $1, %ymm1, %zmm0, %zmm0
; AVX512CD-NEXT: retq
;
; AVX512CDBW-LABEL: testv64i8:
; AVX512CDBW: # %bb.0:
; AVX512CDBW-NEXT: vextracti64x4 $1, %zmm0, %ymm1
-; AVX512CDBW-NEXT: vextracti128 $1, %ymm1, %xmm2
-; AVX512CDBW-NEXT: vpmovzxbd {{.*#+}} zmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero,xmm2[4],zero,zero,zero,xmm2[5],zero,zero,zero,xmm2[6],zero,zero,zero,xmm2[7],zero,zero,zero,xmm2[8],zero,zero,zero,xmm2[9],zero,zero,zero,xmm2[10],zero,zero,zero,xmm2[11],zero,zero,zero,xmm2[12],zero,zero,zero,xmm2[13],zero,zero,zero,xmm2[14],zero,zero,zero,xmm2[15],zero,zero,zero
+; AVX512CDBW-NEXT: vpmovzxbd {{.*#+}} zmm2 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero,xmm1[8],zero,zero,zero,xmm1[9],zero,zero,zero,xmm1[10],zero,zero,zero,xmm1[11],zero,zero,zero,xmm1[12],zero,zero,zero,xmm1[13],zero,zero,zero,xmm1[14],zero,zero,zero,xmm1[15],zero,zero,zero
; AVX512CDBW-NEXT: vplzcntd %zmm2, %zmm2
; AVX512CDBW-NEXT: vpmovdb %zmm2, %xmm2
-; AVX512CDBW-NEXT: vmovdqa {{.*#+}} xmm3 = [24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24]
-; AVX512CDBW-NEXT: vpsubb %xmm3, %xmm2, %xmm2
+; AVX512CDBW-NEXT: vextracti128 $1, %ymm1, %xmm1
; AVX512CDBW-NEXT: vpmovzxbd {{.*#+}} zmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero,xmm1[8],zero,zero,zero,xmm1[9],zero,zero,zero,xmm1[10],zero,zero,zero,xmm1[11],zero,zero,zero,xmm1[12],zero,zero,zero,xmm1[13],zero,zero,zero,xmm1[14],zero,zero,zero,xmm1[15],zero,zero,zero
; AVX512CDBW-NEXT: vplzcntd %zmm1, %zmm1
; AVX512CDBW-NEXT: vpmovdb %zmm1, %xmm1
-; AVX512CDBW-NEXT: vpsubb %xmm3, %xmm1, %xmm1
-; AVX512CDBW-NEXT: vinserti128 $1, %xmm2, %ymm1, %ymm1
-; AVX512CDBW-NEXT: vextracti128 $1, %ymm0, %xmm2
-; AVX512CDBW-NEXT: vpmovzxbd {{.*#+}} zmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero,xmm2[4],zero,zero,zero,xmm2[5],zero,zero,zero,xmm2[6],zero,zero,zero,xmm2[7],zero,zero,zero,xmm2[8],zero,zero,zero,xmm2[9],zero,zero,zero,xmm2[10],zero,zero,zero,xmm2[11],zero,zero,zero,xmm2[12],zero,zero,zero,xmm2[13],zero,zero,zero,xmm2[14],zero,zero,zero,xmm2[15],zero,zero,zero
+; AVX512CDBW-NEXT: vinserti128 $1, %xmm1, %ymm2, %ymm1
+; AVX512CDBW-NEXT: vpmovzxbd {{.*#+}} zmm2 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
; AVX512CDBW-NEXT: vplzcntd %zmm2, %zmm2
; AVX512CDBW-NEXT: vpmovdb %zmm2, %xmm2
-; AVX512CDBW-NEXT: vpsubb %xmm3, %xmm2, %xmm2
+; AVX512CDBW-NEXT: vextracti128 $1, %ymm0, %xmm0
; AVX512CDBW-NEXT: vpmovzxbd {{.*#+}} zmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
; AVX512CDBW-NEXT: vplzcntd %zmm0, %zmm0
; AVX512CDBW-NEXT: vpmovdb %zmm0, %xmm0
-; AVX512CDBW-NEXT: vpsubb %xmm3, %xmm0, %xmm0
-; AVX512CDBW-NEXT: vinserti128 $1, %xmm2, %ymm0, %ymm0
+; AVX512CDBW-NEXT: vinserti128 $1, %xmm0, %ymm2, %ymm0
; AVX512CDBW-NEXT: vinserti64x4 $1, %ymm1, %zmm0, %zmm0
+; AVX512CDBW-NEXT: vpsubb {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %zmm0
; AVX512CDBW-NEXT: retq
;
; AVX512BW-LABEL: testv64i8:
; AVX512CD-LABEL: testv64i8u:
; AVX512CD: # %bb.0:
; AVX512CD-NEXT: vextracti64x4 $1, %zmm0, %ymm1
-; AVX512CD-NEXT: vextracti128 $1, %ymm1, %xmm2
-; AVX512CD-NEXT: vpmovzxbd {{.*#+}} zmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero,xmm2[4],zero,zero,zero,xmm2[5],zero,zero,zero,xmm2[6],zero,zero,zero,xmm2[7],zero,zero,zero,xmm2[8],zero,zero,zero,xmm2[9],zero,zero,zero,xmm2[10],zero,zero,zero,xmm2[11],zero,zero,zero,xmm2[12],zero,zero,zero,xmm2[13],zero,zero,zero,xmm2[14],zero,zero,zero,xmm2[15],zero,zero,zero
+; AVX512CD-NEXT: vpmovzxbd {{.*#+}} zmm2 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero,xmm1[8],zero,zero,zero,xmm1[9],zero,zero,zero,xmm1[10],zero,zero,zero,xmm1[11],zero,zero,zero,xmm1[12],zero,zero,zero,xmm1[13],zero,zero,zero,xmm1[14],zero,zero,zero,xmm1[15],zero,zero,zero
; AVX512CD-NEXT: vplzcntd %zmm2, %zmm2
; AVX512CD-NEXT: vpmovdb %zmm2, %xmm2
-; AVX512CD-NEXT: vmovdqa {{.*#+}} xmm3 = [24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24]
-; AVX512CD-NEXT: vpsubb %xmm3, %xmm2, %xmm2
+; AVX512CD-NEXT: vextracti128 $1, %ymm1, %xmm1
; AVX512CD-NEXT: vpmovzxbd {{.*#+}} zmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero,xmm1[8],zero,zero,zero,xmm1[9],zero,zero,zero,xmm1[10],zero,zero,zero,xmm1[11],zero,zero,zero,xmm1[12],zero,zero,zero,xmm1[13],zero,zero,zero,xmm1[14],zero,zero,zero,xmm1[15],zero,zero,zero
; AVX512CD-NEXT: vplzcntd %zmm1, %zmm1
; AVX512CD-NEXT: vpmovdb %zmm1, %xmm1
-; AVX512CD-NEXT: vpsubb %xmm3, %xmm1, %xmm1
-; AVX512CD-NEXT: vinserti128 $1, %xmm2, %ymm1, %ymm1
-; AVX512CD-NEXT: vextracti128 $1, %ymm0, %xmm2
-; AVX512CD-NEXT: vpmovzxbd {{.*#+}} zmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero,xmm2[4],zero,zero,zero,xmm2[5],zero,zero,zero,xmm2[6],zero,zero,zero,xmm2[7],zero,zero,zero,xmm2[8],zero,zero,zero,xmm2[9],zero,zero,zero,xmm2[10],zero,zero,zero,xmm2[11],zero,zero,zero,xmm2[12],zero,zero,zero,xmm2[13],zero,zero,zero,xmm2[14],zero,zero,zero,xmm2[15],zero,zero,zero
-; AVX512CD-NEXT: vplzcntd %zmm2, %zmm2
-; AVX512CD-NEXT: vpmovdb %zmm2, %xmm2
-; AVX512CD-NEXT: vpsubb %xmm3, %xmm2, %xmm2
+; AVX512CD-NEXT: vinserti128 $1, %xmm1, %ymm2, %ymm1
+; AVX512CD-NEXT: vmovdqa {{.*#+}} ymm2 = [24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24]
+; AVX512CD-NEXT: vpsubb %ymm2, %ymm1, %ymm1
+; AVX512CD-NEXT: vpmovzxbd {{.*#+}} zmm3 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
+; AVX512CD-NEXT: vplzcntd %zmm3, %zmm3
+; AVX512CD-NEXT: vpmovdb %zmm3, %xmm3
+; AVX512CD-NEXT: vextracti128 $1, %ymm0, %xmm0
; AVX512CD-NEXT: vpmovzxbd {{.*#+}} zmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
; AVX512CD-NEXT: vplzcntd %zmm0, %zmm0
; AVX512CD-NEXT: vpmovdb %zmm0, %xmm0
-; AVX512CD-NEXT: vpsubb %xmm3, %xmm0, %xmm0
-; AVX512CD-NEXT: vinserti128 $1, %xmm2, %ymm0, %ymm0
+; AVX512CD-NEXT: vinserti128 $1, %xmm0, %ymm3, %ymm0
+; AVX512CD-NEXT: vpsubb %ymm2, %ymm0, %ymm0
; AVX512CD-NEXT: vinserti64x4 $1, %ymm1, %zmm0, %zmm0
; AVX512CD-NEXT: retq
;
; AVX512CDBW-LABEL: testv64i8u:
; AVX512CDBW: # %bb.0:
; AVX512CDBW-NEXT: vextracti64x4 $1, %zmm0, %ymm1
-; AVX512CDBW-NEXT: vextracti128 $1, %ymm1, %xmm2
-; AVX512CDBW-NEXT: vpmovzxbd {{.*#+}} zmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero,xmm2[4],zero,zero,zero,xmm2[5],zero,zero,zero,xmm2[6],zero,zero,zero,xmm2[7],zero,zero,zero,xmm2[8],zero,zero,zero,xmm2[9],zero,zero,zero,xmm2[10],zero,zero,zero,xmm2[11],zero,zero,zero,xmm2[12],zero,zero,zero,xmm2[13],zero,zero,zero,xmm2[14],zero,zero,zero,xmm2[15],zero,zero,zero
+; AVX512CDBW-NEXT: vpmovzxbd {{.*#+}} zmm2 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero,xmm1[8],zero,zero,zero,xmm1[9],zero,zero,zero,xmm1[10],zero,zero,zero,xmm1[11],zero,zero,zero,xmm1[12],zero,zero,zero,xmm1[13],zero,zero,zero,xmm1[14],zero,zero,zero,xmm1[15],zero,zero,zero
; AVX512CDBW-NEXT: vplzcntd %zmm2, %zmm2
; AVX512CDBW-NEXT: vpmovdb %zmm2, %xmm2
-; AVX512CDBW-NEXT: vmovdqa {{.*#+}} xmm3 = [24,24,24,24,24,24,24,24,24,24,24,24,24,24,24,24]
-; AVX512CDBW-NEXT: vpsubb %xmm3, %xmm2, %xmm2
+; AVX512CDBW-NEXT: vextracti128 $1, %ymm1, %xmm1
; AVX512CDBW-NEXT: vpmovzxbd {{.*#+}} zmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero,xmm1[8],zero,zero,zero,xmm1[9],zero,zero,zero,xmm1[10],zero,zero,zero,xmm1[11],zero,zero,zero,xmm1[12],zero,zero,zero,xmm1[13],zero,zero,zero,xmm1[14],zero,zero,zero,xmm1[15],zero,zero,zero
; AVX512CDBW-NEXT: vplzcntd %zmm1, %zmm1
; AVX512CDBW-NEXT: vpmovdb %zmm1, %xmm1
-; AVX512CDBW-NEXT: vpsubb %xmm3, %xmm1, %xmm1
-; AVX512CDBW-NEXT: vinserti128 $1, %xmm2, %ymm1, %ymm1
-; AVX512CDBW-NEXT: vextracti128 $1, %ymm0, %xmm2
-; AVX512CDBW-NEXT: vpmovzxbd {{.*#+}} zmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero,xmm2[4],zero,zero,zero,xmm2[5],zero,zero,zero,xmm2[6],zero,zero,zero,xmm2[7],zero,zero,zero,xmm2[8],zero,zero,zero,xmm2[9],zero,zero,zero,xmm2[10],zero,zero,zero,xmm2[11],zero,zero,zero,xmm2[12],zero,zero,zero,xmm2[13],zero,zero,zero,xmm2[14],zero,zero,zero,xmm2[15],zero,zero,zero
+; AVX512CDBW-NEXT: vinserti128 $1, %xmm1, %ymm2, %ymm1
+; AVX512CDBW-NEXT: vpmovzxbd {{.*#+}} zmm2 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
; AVX512CDBW-NEXT: vplzcntd %zmm2, %zmm2
; AVX512CDBW-NEXT: vpmovdb %zmm2, %xmm2
-; AVX512CDBW-NEXT: vpsubb %xmm3, %xmm2, %xmm2
+; AVX512CDBW-NEXT: vextracti128 $1, %ymm0, %xmm0
; AVX512CDBW-NEXT: vpmovzxbd {{.*#+}} zmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
; AVX512CDBW-NEXT: vplzcntd %zmm0, %zmm0
; AVX512CDBW-NEXT: vpmovdb %zmm0, %xmm0
-; AVX512CDBW-NEXT: vpsubb %xmm3, %xmm0, %xmm0
-; AVX512CDBW-NEXT: vinserti128 $1, %xmm2, %ymm0, %ymm0
+; AVX512CDBW-NEXT: vinserti128 $1, %xmm0, %ymm2, %ymm0
; AVX512CDBW-NEXT: vinserti64x4 $1, %ymm1, %zmm0, %zmm0
+; AVX512CDBW-NEXT: vpsubb {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %zmm0
; AVX512CDBW-NEXT: retq
;
; AVX512BW-LABEL: testv64i8u: