public inbox for gcc-cvs@sourceware.org
help / color / mirror / Atom feed
* [gcc r12-110] Add folding and remove expanders for x86 *pcmp{et, gt}* builtins [PR target/98911]
@ 2021-04-25  1:18 hongtao Liu
  0 siblings, 0 replies; only message in thread
From: hongtao Liu @ 2021-04-25  1:18 UTC (permalink / raw)
  To: gcc-cvs

https://gcc.gnu.org/g:c54a9f7259fce1a221270f21d8076c960c7d0163

commit r12-110-gc54a9f7259fce1a221270f21d8076c960c7d0163
Author: liuhongt <hongtao.liu@intel.com>
Date:   Tue Feb 23 11:17:40 2021 +0800

    Add folding and remove expanders for x86 *pcmp{et,gt}* builtins [PR target/98911]
    
    gcc/ChangeLog:
    
            PR target/98911
            * config/i386/i386-builtin.def (BDESC): Change the icode of
            the following builtins to CODE_FOR_nothing.
            * config/i386/i386.c (ix86_gimple_fold_builtin): Fold
            IX86_BUILTIN_PCMPEQB128, IX86_BUILTIN_PCMPEQW128,
            IX86_BUILTIN_PCMPEQD128, IX86_BUILTIN_PCMPEQQ,
            IX86_BUILTIN_PCMPEQB256, IX86_BUILTIN_PCMPEQW256,
            IX86_BUILTIN_PCMPEQD256, IX86_BUILTIN_PCMPEQQ256,
            IX86_BUILTIN_PCMPGTB128, IX86_BUILTIN_PCMPGTW128,
            IX86_BUILTIN_PCMPGTD128, IX86_BUILTIN_PCMPGTQ,
            IX86_BUILTIN_PCMPGTB256, IX86_BUILTIN_PCMPGTW256,
            IX86_BUILTIN_PCMPGTD256, IX86_BUILTIN_PCMPGTQ256.
            * config/i386/sse.md (avx2_eq<mode>3): Deleted.
            (sse2_eq<mode>3): Ditto.
            (sse4_1_eqv2di3): Ditto.
            (sse2_gt<mode>3): Rename to ..
            (*sse2_gt<mode>3): .. this.
    
    gcc/testsuite/ChangeLog:
    
            PR target/98911
            * gcc.target/i386/pr98911.c: New test.
            * gcc.target/i386/funcspec-8.c: Replace __builtin_ia32_pcmpgtq
            with __builtin_ia32_pcmpistrm128 since it has been folded.

Diff:
---
 gcc/config/i386/i386-builtin.def           |  32 ++++----
 gcc/config/i386/i386.c                     |  44 +++++++++++
 gcc/config/i386/sse.md                     |  26 +------
 gcc/testsuite/gcc.target/i386/funcspec-8.c |  17 +++--
 gcc/testsuite/gcc.target/i386/pr98911.c    | 116 +++++++++++++++++++++++++++++
 5 files changed, 186 insertions(+), 49 deletions(-)

diff --git a/gcc/config/i386/i386-builtin.def b/gcc/config/i386/i386-builtin.def
index e3ed4e1578f..4dbd4f23647 100644
--- a/gcc/config/i386/i386-builtin.def
+++ b/gcc/config/i386/i386-builtin.def
@@ -773,12 +773,12 @@ BDESC (OPTION_MASK_ISA_SSE2, 0, CODE_FOR_xorv2di3, "__builtin_ia32_pxor128", IX8
 BDESC (OPTION_MASK_ISA_SSE2, 0, CODE_FOR_sse2_uavgv16qi3, "__builtin_ia32_pavgb128", IX86_BUILTIN_PAVGB128, UNKNOWN, (int) V16QI_FTYPE_V16QI_V16QI)
 BDESC (OPTION_MASK_ISA_SSE2, 0, CODE_FOR_sse2_uavgv8hi3, "__builtin_ia32_pavgw128", IX86_BUILTIN_PAVGW128, UNKNOWN, (int) V8HI_FTYPE_V8HI_V8HI)
 
-BDESC (OPTION_MASK_ISA_SSE2, 0, CODE_FOR_sse2_eqv16qi3, "__builtin_ia32_pcmpeqb128", IX86_BUILTIN_PCMPEQB128, UNKNOWN, (int) V16QI_FTYPE_V16QI_V16QI)
-BDESC (OPTION_MASK_ISA_SSE2, 0, CODE_FOR_sse2_eqv8hi3, "__builtin_ia32_pcmpeqw128", IX86_BUILTIN_PCMPEQW128, UNKNOWN, (int) V8HI_FTYPE_V8HI_V8HI)
-BDESC (OPTION_MASK_ISA_SSE2, 0, CODE_FOR_sse2_eqv4si3, "__builtin_ia32_pcmpeqd128", IX86_BUILTIN_PCMPEQD128, UNKNOWN, (int) V4SI_FTYPE_V4SI_V4SI )
-BDESC (OPTION_MASK_ISA_SSE2, 0, CODE_FOR_sse2_gtv16qi3, "__builtin_ia32_pcmpgtb128", IX86_BUILTIN_PCMPGTB128, UNKNOWN, (int) V16QI_FTYPE_V16QI_V16QI)
-BDESC (OPTION_MASK_ISA_SSE2, 0, CODE_FOR_sse2_gtv8hi3, "__builtin_ia32_pcmpgtw128", IX86_BUILTIN_PCMPGTW128, UNKNOWN, (int) V8HI_FTYPE_V8HI_V8HI)
-BDESC (OPTION_MASK_ISA_SSE2, 0, CODE_FOR_sse2_gtv4si3, "__builtin_ia32_pcmpgtd128", IX86_BUILTIN_PCMPGTD128, UNKNOWN, (int) V4SI_FTYPE_V4SI_V4SI )
+BDESC (OPTION_MASK_ISA_SSE2, 0, CODE_FOR_nothing, "__builtin_ia32_pcmpeqb128", IX86_BUILTIN_PCMPEQB128, UNKNOWN, (int) V16QI_FTYPE_V16QI_V16QI)
+BDESC (OPTION_MASK_ISA_SSE2, 0, CODE_FOR_nothing, "__builtin_ia32_pcmpeqw128", IX86_BUILTIN_PCMPEQW128, UNKNOWN, (int) V8HI_FTYPE_V8HI_V8HI)
+BDESC (OPTION_MASK_ISA_SSE2, 0, CODE_FOR_nothing, "__builtin_ia32_pcmpeqd128", IX86_BUILTIN_PCMPEQD128, UNKNOWN, (int) V4SI_FTYPE_V4SI_V4SI )
+BDESC (OPTION_MASK_ISA_SSE2, 0, CODE_FOR_nothing, "__builtin_ia32_pcmpgtb128", IX86_BUILTIN_PCMPGTB128, UNKNOWN, (int) V16QI_FTYPE_V16QI_V16QI)
+BDESC (OPTION_MASK_ISA_SSE2, 0, CODE_FOR_nothing, "__builtin_ia32_pcmpgtw128", IX86_BUILTIN_PCMPGTW128, UNKNOWN, (int) V8HI_FTYPE_V8HI_V8HI)
+BDESC (OPTION_MASK_ISA_SSE2, 0, CODE_FOR_nothing, "__builtin_ia32_pcmpgtd128", IX86_BUILTIN_PCMPGTD128, UNKNOWN, (int) V4SI_FTYPE_V4SI_V4SI )
 
 BDESC (OPTION_MASK_ISA_SSE2, 0, CODE_FOR_umaxv16qi3, "__builtin_ia32_pmaxub128", IX86_BUILTIN_PMAXUB128, UNKNOWN, (int) V16QI_FTYPE_V16QI_V16QI)
 BDESC (OPTION_MASK_ISA_SSE2, 0, CODE_FOR_smaxv8hi3, "__builtin_ia32_pmaxsw128", IX86_BUILTIN_PMAXSW128, UNKNOWN, (int) V8HI_FTYPE_V8HI_V8HI)
@@ -919,7 +919,7 @@ BDESC (OPTION_MASK_ISA_SSE4_1, 0, CODE_FOR_sse4_1_zero_extendv2siv2di2, "__built
 BDESC (OPTION_MASK_ISA_SSE4_1, 0, CODE_FOR_sse4_1_phminposuw, "__builtin_ia32_phminposuw128", IX86_BUILTIN_PHMINPOSUW128, UNKNOWN, (int) V8HI_FTYPE_V8HI)
 
 BDESC (OPTION_MASK_ISA_SSE4_1, 0, CODE_FOR_sse4_1_packusdw, "__builtin_ia32_packusdw128", IX86_BUILTIN_PACKUSDW128, UNKNOWN, (int) V8HI_FTYPE_V4SI_V4SI)
-BDESC (OPTION_MASK_ISA_SSE4_1, 0, CODE_FOR_sse4_1_eqv2di3, "__builtin_ia32_pcmpeqq", IX86_BUILTIN_PCMPEQQ, UNKNOWN, (int) V2DI_FTYPE_V2DI_V2DI)
+BDESC (OPTION_MASK_ISA_SSE4_1, 0, CODE_FOR_nothing, "__builtin_ia32_pcmpeqq", IX86_BUILTIN_PCMPEQQ, UNKNOWN, (int) V2DI_FTYPE_V2DI_V2DI)
 BDESC (OPTION_MASK_ISA_SSE4_1, 0, CODE_FOR_smaxv16qi3, "__builtin_ia32_pmaxsb128", IX86_BUILTIN_PMAXSB128, UNKNOWN, (int) V16QI_FTYPE_V16QI_V16QI)
 BDESC (OPTION_MASK_ISA_SSE4_1, 0, CODE_FOR_smaxv4si3, "__builtin_ia32_pmaxsd128", IX86_BUILTIN_PMAXSD128, UNKNOWN, (int) V4SI_FTYPE_V4SI_V4SI)
 BDESC (OPTION_MASK_ISA_SSE4_1, 0, CODE_FOR_umaxv4si3, "__builtin_ia32_pmaxud128", IX86_BUILTIN_PMAXUD128, UNKNOWN, (int) V4SI_FTYPE_V4SI_V4SI)
@@ -962,7 +962,7 @@ BDESC (OPTION_MASK_ISA_SSE4_1, 0, CODE_FOR_sse4_1_ptestv2di, "__builtin_ia32_pte
 BDESC (OPTION_MASK_ISA_SSE4_1, 0, CODE_FOR_sse4_1_ptestv2di, "__builtin_ia32_ptestnzc128", IX86_BUILTIN_PTESTNZC, GTU, (int) INT_FTYPE_V2DI_V2DI_PTEST)
 
 /* SSE4.2 */
-BDESC (OPTION_MASK_ISA_SSE4_2, 0, CODE_FOR_sse4_2_gtv2di3, "__builtin_ia32_pcmpgtq", IX86_BUILTIN_PCMPGTQ, UNKNOWN, (int) V2DI_FTYPE_V2DI_V2DI)
+BDESC (OPTION_MASK_ISA_SSE4_2, 0, CODE_FOR_nothing, "__builtin_ia32_pcmpgtq", IX86_BUILTIN_PCMPGTQ, UNKNOWN, (int) V2DI_FTYPE_V2DI_V2DI)
 BDESC (OPTION_MASK_ISA_SSE4_2 | OPTION_MASK_ISA_CRC32, 0, CODE_FOR_sse4_2_crc32qi, "__builtin_ia32_crc32qi", IX86_BUILTIN_CRC32QI, UNKNOWN, (int) UINT_FTYPE_UINT_UCHAR)
 BDESC (OPTION_MASK_ISA_SSE4_2 | OPTION_MASK_ISA_CRC32, 0, CODE_FOR_sse4_2_crc32hi, "__builtin_ia32_crc32hi", IX86_BUILTIN_CRC32HI, UNKNOWN, (int) UINT_FTYPE_UINT_USHORT)
 BDESC (OPTION_MASK_ISA_SSE4_2 | OPTION_MASK_ISA_CRC32, 0, CODE_FOR_sse4_2_crc32si, "__builtin_ia32_crc32si", IX86_BUILTIN_CRC32SI, UNKNOWN, (int) UINT_FTYPE_UINT_UINT)
@@ -1149,14 +1149,14 @@ BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_avx2_uavgv32qi3, "__builtin_ia32_pavgb2
 BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_avx2_uavgv16hi3, "__builtin_ia32_pavgw256",  IX86_BUILTIN_PAVGW256, UNKNOWN, (int) V16HI_FTYPE_V16HI_V16HI)
 BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_avx2_pblendvb, "__builtin_ia32_pblendvb256", IX86_BUILTIN_PBLENDVB256, UNKNOWN, (int) V32QI_FTYPE_V32QI_V32QI_V32QI)
 BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_avx2_pblendw, "__builtin_ia32_pblendw256", IX86_BUILTIN_PBLENDVW256, UNKNOWN, (int) V16HI_FTYPE_V16HI_V16HI_INT)
-BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_avx2_eqv32qi3, "__builtin_ia32_pcmpeqb256", IX86_BUILTIN_PCMPEQB256, UNKNOWN, (int) V32QI_FTYPE_V32QI_V32QI)
-BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_avx2_eqv16hi3, "__builtin_ia32_pcmpeqw256", IX86_BUILTIN_PCMPEQW256, UNKNOWN, (int) V16HI_FTYPE_V16HI_V16HI)
-BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_avx2_eqv8si3, "__builtin_ia32_pcmpeqd256", IX86_BUILTIN_PCMPEQD256, UNKNOWN, (int) V8SI_FTYPE_V8SI_V8SI )
-BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_avx2_eqv4di3, "__builtin_ia32_pcmpeqq256", IX86_BUILTIN_PCMPEQQ256, UNKNOWN, (int) V4DI_FTYPE_V4DI_V4DI )
-BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_avx2_gtv32qi3, "__builtin_ia32_pcmpgtb256", IX86_BUILTIN_PCMPGTB256, UNKNOWN, (int) V32QI_FTYPE_V32QI_V32QI)
-BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_avx2_gtv16hi3, "__builtin_ia32_pcmpgtw256", IX86_BUILTIN_PCMPGTW256, UNKNOWN, (int) V16HI_FTYPE_V16HI_V16HI)
-BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_avx2_gtv8si3, "__builtin_ia32_pcmpgtd256", IX86_BUILTIN_PCMPGTD256, UNKNOWN, (int) V8SI_FTYPE_V8SI_V8SI )
-BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_avx2_gtv4di3, "__builtin_ia32_pcmpgtq256", IX86_BUILTIN_PCMPGTQ256, UNKNOWN, (int) V4DI_FTYPE_V4DI_V4DI )
+BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_nothing, "__builtin_ia32_pcmpeqb256", IX86_BUILTIN_PCMPEQB256, UNKNOWN, (int) V32QI_FTYPE_V32QI_V32QI)
+BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_nothing, "__builtin_ia32_pcmpeqw256", IX86_BUILTIN_PCMPEQW256, UNKNOWN, (int) V16HI_FTYPE_V16HI_V16HI)
+BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_nothing, "__builtin_ia32_pcmpeqd256", IX86_BUILTIN_PCMPEQD256, UNKNOWN, (int) V8SI_FTYPE_V8SI_V8SI )
+BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_nothing, "__builtin_ia32_pcmpeqq256", IX86_BUILTIN_PCMPEQQ256, UNKNOWN, (int) V4DI_FTYPE_V4DI_V4DI )
+BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_nothing, "__builtin_ia32_pcmpgtb256", IX86_BUILTIN_PCMPGTB256, UNKNOWN, (int) V32QI_FTYPE_V32QI_V32QI)
+BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_nothing, "__builtin_ia32_pcmpgtw256", IX86_BUILTIN_PCMPGTW256, UNKNOWN, (int) V16HI_FTYPE_V16HI_V16HI)
+BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_nothing, "__builtin_ia32_pcmpgtd256", IX86_BUILTIN_PCMPGTD256, UNKNOWN, (int) V8SI_FTYPE_V8SI_V8SI )
+BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_nothing, "__builtin_ia32_pcmpgtq256", IX86_BUILTIN_PCMPGTQ256, UNKNOWN, (int) V4DI_FTYPE_V4DI_V4DI )
 BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_avx2_phaddwv16hi3, "__builtin_ia32_phaddw256", IX86_BUILTIN_PHADDW256, UNKNOWN, (int) V16HI_FTYPE_V16HI_V16HI)
 BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_avx2_phadddv8si3, "__builtin_ia32_phaddd256", IX86_BUILTIN_PHADDD256, UNKNOWN, (int) V8SI_FTYPE_V8SI_V8SI)
 BDESC (OPTION_MASK_ISA_AVX2, 0, CODE_FOR_avx2_phaddswv16hi3, "__builtin_ia32_phaddsw256", IX86_BUILTIN_PHADDSW256, UNKNOWN, (int) V16HI_FTYPE_V16HI_V16HI)
diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
index d3c09bfd7b9..adcef1e98bf 100644
--- a/gcc/config/i386/i386.c
+++ b/gcc/config/i386/i386.c
@@ -67,6 +67,7 @@ along with GCC; see the file COPYING3.  If not see
 #include "pass_manager.h"
 #include "target-globals.h"
 #include "gimple-iterator.h"
+#include "gimple-fold.h"
 #include "tree-vectorizer.h"
 #include "shrink-wrap.h"
 #include "builtins.h"
@@ -17865,6 +17866,7 @@ ix86_gimple_fold_builtin (gimple_stmt_iterator *gsi)
   tree decl = NULL_TREE;
   tree arg0, arg1, arg2;
   enum rtx_code rcode;
+  enum tree_code tcode;
   unsigned HOST_WIDE_INT count;
   bool is_vshift;
 
@@ -17946,6 +17948,48 @@ ix86_gimple_fold_builtin (gimple_stmt_iterator *gsi)
 	}
       break;
 
+    case IX86_BUILTIN_PCMPEQB128:
+    case IX86_BUILTIN_PCMPEQW128:
+    case IX86_BUILTIN_PCMPEQD128:
+    case IX86_BUILTIN_PCMPEQQ:
+    case IX86_BUILTIN_PCMPEQB256:
+    case IX86_BUILTIN_PCMPEQW256:
+    case IX86_BUILTIN_PCMPEQD256:
+    case IX86_BUILTIN_PCMPEQQ256:
+      tcode = EQ_EXPR;
+      goto do_cmp;
+
+    case IX86_BUILTIN_PCMPGTB128:
+    case IX86_BUILTIN_PCMPGTW128:
+    case IX86_BUILTIN_PCMPGTD128:
+    case IX86_BUILTIN_PCMPGTQ:
+    case IX86_BUILTIN_PCMPGTB256:
+    case IX86_BUILTIN_PCMPGTW256:
+    case IX86_BUILTIN_PCMPGTD256:
+    case IX86_BUILTIN_PCMPGTQ256:
+      tcode = GT_EXPR;
+
+    do_cmp:
+      gcc_assert (n_args == 2);
+      arg0 = gimple_call_arg (stmt, 0);
+      arg1 = gimple_call_arg (stmt, 1);
+      {
+	location_t loc = gimple_location (stmt);
+	tree type = TREE_TYPE (arg0);
+	tree zero_vec = build_zero_cst (type);
+	tree minus_one_vec = build_minus_one_cst (type);
+	tree cmp_type = truth_type_for (type);
+	gimple_seq stmts = NULL;
+	tree cmp = gimple_build (&stmts, tcode, cmp_type, arg0, arg1);
+	gsi_insert_before (gsi, stmts, GSI_SAME_STMT);
+	gimple *g = gimple_build_assign (gimple_call_lhs (stmt),
+					 VEC_COND_EXPR, cmp,
+					 minus_one_vec, zero_vec);
+	gimple_set_location (g, loc);
+	gsi_replace (gsi, g, false);
+      }
+      return true;
+
     case IX86_BUILTIN_PSLLD:
     case IX86_BUILTIN_PSLLD128:
     case IX86_BUILTIN_PSLLD128_MASK:
diff --git a/gcc/config/i386/sse.md b/gcc/config/i386/sse.md
index 319099d7b7f..897cf3eaea9 100644
--- a/gcc/config/i386/sse.md
+++ b/gcc/config/i386/sse.md
@@ -12875,14 +12875,6 @@
 ;;
 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 
-(define_expand "avx2_eq<mode>3"
-  [(set (match_operand:VI_256 0 "register_operand")
-	(eq:VI_256
-	  (match_operand:VI_256 1 "nonimmediate_operand")
-	  (match_operand:VI_256 2 "nonimmediate_operand")))]
-  "TARGET_AVX2"
-  "ix86_fixup_binary_operands_no_copy (EQ, <MODE>mode, operands);")
-
 (define_insn "*avx2_eq<mode>3"
   [(set (match_operand:VI_256 0 "register_operand" "=x")
 	(eq:VI_256
@@ -13058,22 +13050,6 @@
    (set_attr "prefix" "orig,vex")
    (set_attr "mode" "TI")])
 
-(define_expand "sse2_eq<mode>3"
-  [(set (match_operand:VI124_128 0 "register_operand")
-	(eq:VI124_128
-	  (match_operand:VI124_128 1 "vector_operand")
-	  (match_operand:VI124_128 2 "vector_operand")))]
-  "TARGET_SSE2 && !TARGET_XOP "
-  "ix86_fixup_binary_operands_no_copy (EQ, <MODE>mode, operands);")
-
-(define_expand "sse4_1_eqv2di3"
-  [(set (match_operand:V2DI 0 "register_operand")
-	(eq:V2DI
-	  (match_operand:V2DI 1 "vector_operand")
-	  (match_operand:V2DI 2 "vector_operand")))]
-  "TARGET_SSE4_1"
-  "ix86_fixup_binary_operands_no_copy (EQ, V2DImode, operands);")
-
 (define_insn "sse4_2_gtv2di3"
   [(set (match_operand:V2DI 0 "register_operand" "=Yr,*x,x")
 	(gt:V2DI
@@ -13144,7 +13120,7 @@
    (set_attr "prefix" "evex")
    (set_attr "mode" "<sseinsnmode>")])
 
-(define_insn "sse2_gt<mode>3"
+(define_insn "*sse2_gt<mode>3"
   [(set (match_operand:VI124_128 0 "register_operand" "=x,x")
 	(gt:VI124_128
 	  (match_operand:VI124_128 1 "register_operand" "0,x")
diff --git a/gcc/testsuite/gcc.target/i386/funcspec-8.c b/gcc/testsuite/gcc.target/i386/funcspec-8.c
index 8bd95fc2d39..0a6c709003a 100644
--- a/gcc/testsuite/gcc.target/i386/funcspec-8.c
+++ b/gcc/testsuite/gcc.target/i386/funcspec-8.c
@@ -8,6 +8,7 @@ typedef float     __m128  __attribute__ ((__vector_size__ (16), __may_alias__));
 typedef double    __m128d __attribute__ ((__vector_size__ (16), __may_alias__));
 typedef int	  __m128w __attribute__ ((__vector_size__ (16), __may_alias__));
 typedef long long __m128i __attribute__ ((__vector_size__ (16), __may_alias__));
+typedef char __m128qi __attribute__ ((__vector_size__ (16), __may_alias__));
 
 #ifdef __SSE3__
 #error "-msse3 should not be set for this test"
@@ -70,19 +71,19 @@ generic_blendvpd  (__m128d a, __m128d b, __m128d c)
 #error "-msse4.2 should not be set for this test"
 #endif
 
-__m128i sse4_2_pcmpgtq (__m128i a, __m128i b) __attribute__((__target__("sse4.2")));
-__m128i generic_pcmpgtq (__m128i ab, __m128i b);
+__m128qi sse4_2_cmpistrm (__m128qi a, __m128qi b) __attribute__((__target__("sse4.2")));
+__m128qi generic_cmpistrm (__m128qi a, __m128qi b);
 
-__m128i
-sse4_2_pcmpgtq (__m128i a, __m128i b)
+__m128qi
+sse4_2_cmpistrm (__m128qi a, __m128qi b)
 {
-  return __builtin_ia32_pcmpgtq (a, b);
+  return  __builtin_ia32_pcmpistrm128 (a, b, 0);
 }
 
-__m128i
-generic_pcmpgtq (__m128i a, __m128i b)
+__m128qi
+generic_comistrm (__m128qi a, __m128qi b)
 {
-  return __builtin_ia32_pcmpgtq (a, b);			/* { dg-error "needs isa option" } */
+  return  __builtin_ia32_pcmpistrm128 (a, b, 0);	/* { dg-error "needs isa option" } */
 }
 
 #ifdef __SSE4A__
diff --git a/gcc/testsuite/gcc.target/i386/pr98911.c b/gcc/testsuite/gcc.target/i386/pr98911.c
new file mode 100644
index 00000000000..f08cea04858
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/pr98911.c
@@ -0,0 +1,116 @@
+/* PR target/98911  */
+/* { dg-do compile } */
+/* { dg-options "-O -mavx2" } */
+/* { dg-final { scan-assembler-times "pcmpeqb" 2 } } */
+/* { dg-final { scan-assembler-times "pcmpeqw" 2 } } */
+/* { dg-final { scan-assembler-times "pcmpeqd" 2 } } */
+/* { dg-final { scan-assembler-times "pcmpeqq" 2 } } */
+/* { dg-final { scan-assembler-times "pcmpgtb" 2 } } */
+/* { dg-final { scan-assembler-times "pcmpgtw" 2 } } */
+/* { dg-final { scan-assembler-times "pcmpgtd" 2 } } */
+/* { dg-final { scan-assembler-times "pcmpgtq" 2 } } */
+
+typedef char v16qi __attribute__ ((vector_size (16)));
+typedef char v32qi __attribute__ ((vector_size (32)));
+typedef short v8hi __attribute__ ((vector_size (16)));
+typedef short v16hi __attribute__ ((vector_size (32)));
+typedef int v4si __attribute__ ((vector_size (16)));
+typedef int v8si __attribute__ ((vector_size (32)));
+typedef long long v2di __attribute__ ((vector_size (16)));
+typedef long long v4di __attribute__ ((vector_size (32)));
+
+v16qi
+f1 (v16qi a, v16qi b)
+{
+  return __builtin_ia32_pcmpeqb128 (a, b);
+}
+
+v8hi
+f2 (v8hi a, v8hi b)
+{
+  return __builtin_ia32_pcmpeqw128 (a, b);
+}
+
+v4si
+f3 (v4si a, v4si b)
+{
+  return __builtin_ia32_pcmpeqd128 (a, b);
+}
+
+v2di
+f4 (v2di a, v2di b)
+{
+  return __builtin_ia32_pcmpeqq (a, b);
+}
+
+v16qi
+f5 (v16qi a, v16qi b)
+{
+  return __builtin_ia32_pcmpgtb128 (a, b);
+}
+
+v8hi
+f6 (v8hi a, v8hi b)
+{
+  return __builtin_ia32_pcmpgtw128 (a, b);
+}
+
+v4si
+f7 (v4si a, v4si b)
+{
+  return __builtin_ia32_pcmpgtd128 (a, b);
+}
+
+v2di
+f8 (v2di a, v2di b)
+{
+  return __builtin_ia32_pcmpgtq (a, b);
+}
+
+v32qi
+f9 (v32qi a, v32qi b)
+{
+  return __builtin_ia32_pcmpeqb256 (a, b);
+}
+
+v16hi
+f10 (v16hi a, v16hi b)
+{
+  return __builtin_ia32_pcmpeqw256 (a, b);
+}
+
+v8si
+f11 (v8si a, v8si b)
+{
+  return __builtin_ia32_pcmpeqd256 (a, b);
+}
+
+v4di
+f12 (v4di a, v4di b)
+{
+  return __builtin_ia32_pcmpeqq256 (a, b);
+}
+
+v32qi
+f13 (v32qi a, v32qi b)
+{
+  return __builtin_ia32_pcmpgtb256 (a, b);
+}
+
+v16hi
+f14 (v16hi a, v16hi b)
+{
+  return __builtin_ia32_pcmpgtw256 (a, b);
+}
+
+v8si
+f15 (v8si a, v8si b)
+{
+  return __builtin_ia32_pcmpgtd256 (a, b);
+}
+
+v4di
+f16 (v4di a, v4di b)
+{
+  return __builtin_ia32_pcmpgtq256 (a, b);
+}


^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2021-04-25  1:18 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-04-25  1:18 [gcc r12-110] Add folding and remove expanders for x86 *pcmp{et, gt}* builtins [PR target/98911] hongtao Liu

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).