[clang] 9a14c36 - [X86] Replace avx512f integer add reduction builtins with generic builtin

[Simon Pilgrim via cfe-commits]

Author: Simon Pilgrim
Date: 2022-05-02T14:39:17+01:00
New Revision: 9a14c369c422b244db78f1a9f947a891a75d912f

URL: https://github.com/llvm/llvm-project/commit/9a14c369c422b244db78f1a9f947a891a75d912f
DIFF: https://github.com/llvm/llvm-project/commit/9a14c369c422b244db78f1a9f947a891a75d912f.diff

LOG: [X86] Replace avx512f integer add reduction builtins with generic builtin

D124741 added the generic "__builtin_reduce_add" which we can use to replace the x86 specific integer add reduction builtins - internally these were mapping to the same intrinsic already so there are no test changes required.

Differential Revision: https://reviews.llvm.org/D124757

Added: 
    

Modified: 
    clang/include/clang/Basic/BuiltinsX86.def
    clang/lib/CodeGen/CGBuiltin.cpp
    clang/lib/Headers/avx512fintrin.h

Removed: 
    


################################################################################
diff  --git a/clang/include/clang/Basic/BuiltinsX86.def b/clang/include/clang/Basic/BuiltinsX86.def
index 449acfbec5630..6de7e6f89b47d 100644
--- a/clang/include/clang/Basic/BuiltinsX86.def
+++ b/clang/include/clang/Basic/BuiltinsX86.def
@@ -1990,8 +1990,6 @@ TARGET_BUILTIN(__builtin_ia32_selectss_128, "V4fUcV4fV4f", "ncV:128:", "avx512f"
 TARGET_BUILTIN(__builtin_ia32_selectsd_128, "V2dUcV2dV2d", "ncV:128:", "avx512f")
 
 // generic reduction intrinsics
-TARGET_BUILTIN(__builtin_ia32_reduce_add_d512, "iV16i", "ncV:512:", "avx512f")
-TARGET_BUILTIN(__builtin_ia32_reduce_add_q512, "OiV8Oi", "ncV:512:", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_reduce_fadd_pd512, "ddV8d", "ncV:512:", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_reduce_fadd_ps512, "ffV16f", "ncV:512:", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_reduce_fadd_ph512, "xxV32x", "ncV:512:", "avx512fp16")

diff  --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
index 4c203904a0e0e..caea5d16b2659 100644
--- a/clang/lib/CodeGen/CGBuiltin.cpp
+++ b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -14496,12 +14496,6 @@ Value *CodeGenFunction::EmitX86BuiltinExpr(unsigned BuiltinID,
     return EmitX86FunnelShift(*this, Ops[1], Ops[0], Ops[2], true);
 
   // Reductions
-  case X86::BI__builtin_ia32_reduce_add_d512:
-  case X86::BI__builtin_ia32_reduce_add_q512: {
-    Function *F =
-        CGM.getIntrinsic(Intrinsic::vector_reduce_add, Ops[0]->getType());
-    return Builder.CreateCall(F, {Ops[0]});
-  }
   case X86::BI__builtin_ia32_reduce_fadd_pd512:
   case X86::BI__builtin_ia32_reduce_fadd_ps512:
   case X86::BI__builtin_ia32_reduce_fadd_ph512:

diff  --git a/clang/lib/Headers/avx512fintrin.h b/clang/lib/Headers/avx512fintrin.h
index d10493fdd0e66..c7f3c96106076 100644
--- a/clang/lib/Headers/avx512fintrin.h
+++ b/clang/lib/Headers/avx512fintrin.h
@@ -9319,7 +9319,7 @@ _mm512_mask_abs_pd(__m512d __W, __mmask8 __K, __m512d __A)
  */
 
 static __inline__ long long __DEFAULT_FN_ATTRS512 _mm512_reduce_add_epi64(__m512i __W) {
-  return __builtin_ia32_reduce_add_q512(__W);
+  return __builtin_reduce_add((__v8di)__W);
 }
 
 static __inline__ long long __DEFAULT_FN_ATTRS512 _mm512_reduce_mul_epi64(__m512i __W) {
@@ -9337,7 +9337,7 @@ static __inline__ long long __DEFAULT_FN_ATTRS512 _mm512_reduce_or_epi64(__m512i
 static __inline__ long long __DEFAULT_FN_ATTRS512
 _mm512_mask_reduce_add_epi64(__mmask8 __M, __m512i __W) {
   __W = _mm512_maskz_mov_epi64(__M, __W);
-  return __builtin_ia32_reduce_add_q512(__W);
+  return __builtin_reduce_add((__v8di)__W);
 }
 
 static __inline__ long long __DEFAULT_FN_ATTRS512
@@ -9383,7 +9383,7 @@ _mm512_mask_reduce_mul_pd(__mmask8 __M, __m512d __W) {
 
 static __inline__ int __DEFAULT_FN_ATTRS512
 _mm512_reduce_add_epi32(__m512i __W) {
-  return __builtin_ia32_reduce_add_d512((__v16si)__W);
+  return __builtin_reduce_add((__v16si)__W);
 }
 
 static __inline__ int __DEFAULT_FN_ATTRS512
@@ -9404,7 +9404,7 @@ _mm512_reduce_or_epi32(__m512i __W) {
 static __inline__ int __DEFAULT_FN_ATTRS512
 _mm512_mask_reduce_add_epi32( __mmask16 __M, __m512i __W) {
   __W = _mm512_maskz_mov_epi32(__M, __W);
-  return __builtin_ia32_reduce_add_d512((__v16si)__W);
+  return __builtin_reduce_add((__v16si)__W);
 }
 
 static __inline__ int __DEFAULT_FN_ATTRS512