AVX-512. 66/n. Extend vpalignr insn patterns.

gcc/
	* config/i386/sse.md
	(define_mode_iterator SSESCALARMODE): Add V4TI mode.
	(define_insn "<ssse3_avx2>_palignr<mode>_mask"): New.
	(define_insn "<ssse3_avx2>_palignr<mode>"): Add EVEX version.

Co-Authored-By: Andrey Turetskiy <andrey.turetskiy@intel.com>
Co-Authored-By: Anna Tikhonova <anna.tikhonova@intel.com>
Co-Authored-By: Ilya Tocar <ilya.tocar@intel.com>
Co-Authored-By: Ilya Verbin <ilya.verbin@intel.com>
Co-Authored-By: Kirill Yukhin <kirill.yukhin@intel.com>
Co-Authored-By: Maxim Kuznetsov <maxim.kuznetsov@intel.com>
Co-Authored-By: Michael Zolotukhin <michael.v.zolotukhin@intel.com>
From-SVN: r216183

diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index 646ec01..6553b21 100644 (file)
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -8,6 +8,20 @@
            Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
 
        * config/i386/sse.md
+       (define_mode_iterator SSESCALARMODE): Add V4TI mode.
+       (define_insn "<ssse3_avx2>_palignr<mode>_mask"): New.
+       (define_insn "<ssse3_avx2>_palignr<mode>"): Add EVEX version.
+
+2014-10-14  Alexander Ivchenko  <alexander.ivchenko@intel.com>
+           Maxim Kuznetsov  <maxim.kuznetsov@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>
+
+       * config/i386/sse.md
        (define_expand "mul<mode>3<mask_name>"): Add masking.
 
 2014-10-14  Alexander Ivchenko  <alexander.ivchenko@intel.com>

diff --git a/gcc/config/i386/sse.md b/gcc/config/i386/sse.md
index 18614ca..135cb04 100644 (file)
--- a/gcc/config/i386/sse.md
+++ b/gcc/config/i386/sse.md
@@ -350,7 +350,7 @@
 
 ;; ??? This should probably be dropped in favor of VIMAX_AVX2.
 (define_mode_iterator SSESCALARMODE
-  [(V2TI "TARGET_AVX2") TI])
+  [(V4TI "TARGET_AVX512BW") (V2TI "TARGET_AVX2") TI])
 
 (define_mode_iterator VI12_AVX2
   [(V64QI "TARGET_AVX512BW") (V32QI "TARGET_AVX2") V16QI
@@ -13455,11 +13455,33 @@
    (set (attr "prefix_rex") (symbol_ref "x86_extended_reg_mentioned_p (insn)"))
    (set_attr "mode" "DI")])
 
+(define_insn "<ssse3_avx2>_palignr<mode>_mask"
+  [(set (match_operand:VI1_AVX2 0 "register_operand" "=v")
+        (vec_merge:VI1_AVX2
+         (unspec:VI1_AVX2
+           [(match_operand:VI1_AVX2 1 "register_operand" "v")
+            (match_operand:VI1_AVX2 2 "nonimmediate_operand" "vm")
+            (match_operand:SI 3 "const_0_to_255_mul_8_operand" "n")]
+           UNSPEC_PALIGNR)
+       (match_operand:VI1_AVX2 4 "vector_move_operand" "0C")
+       (match_operand:<avx512fmaskmode> 5 "register_operand" "Yk")))]
+  "TARGET_AVX512BW && (<MODE_SIZE> == 64 || TARGET_AVX512VL)"
+{
+  operands[3] = GEN_INT (INTVAL (operands[3]) / 8);
+  return "vpalignr\t{%3, %2, %1, %0%{%5%}%N4|%0%{%5%}%N4, %1, %2, %3}";
+}
+  [(set_attr "type" "sseishft")
+   (set_attr "atom_unit" "sishuf")
+   (set_attr "prefix_extra" "1")
+   (set_attr "length_immediate" "1")
+   (set_attr "prefix" "evex")
+   (set_attr "mode" "<sseinsnmode>")])
+
 (define_insn "<ssse3_avx2>_palignr<mode>"
-  [(set (match_operand:SSESCALARMODE 0 "register_operand" "=x,x")
+  [(set (match_operand:SSESCALARMODE 0 "register_operand" "=x,v")
        (unspec:SSESCALARMODE
-         [(match_operand:SSESCALARMODE 1 "register_operand" "0,x")
-          (match_operand:SSESCALARMODE 2 "nonimmediate_operand" "xm,xm")
+         [(match_operand:SSESCALARMODE 1 "register_operand" "0,v")
+          (match_operand:SSESCALARMODE 2 "nonimmediate_operand" "xm,vm")
           (match_operand:SI 3 "const_0_to_255_mul_8_operand" "n,n")]
          UNSPEC_PALIGNR))]
   "TARGET_SSSE3"