[X86] Reduce the number of FMA builtins needed by the frontend by adding negates...

The backend should be able to combine the negates to create fmsub, fnmadd, and fnmsub. faddsub converting to fsubadd still needs work I think, but should be very doable.

This matches what we already do for the masked builtins.

This only covers the packed builtins. Scalar builtins will be done after FMA4 is fixed.

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@317873 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/include/clang/Basic/BuiltinsX86.def b/include/clang/Basic/BuiltinsX86.def
index 92e8b978925e73570a06d25052c187c048e09bc7..ee655035639485b835539e88abd57e617542c780 100644 (file)
--- a/include/clang/Basic/BuiltinsX86.def
+++ b/include/clang/Basic/BuiltinsX86.def
@@ -684,34 +684,20 @@ TARGET_BUILTIN(__builtin_ia32_vfmaddps, "V4fV4fV4fV4f", "", "fma|fma4")
 TARGET_BUILTIN(__builtin_ia32_vfmaddpd, "V2dV2dV2dV2d", "", "fma|fma4")
 TARGET_BUILTIN(__builtin_ia32_vfmaddss3, "V4fV4fV4fV4f", "", "fma|fma4")
 TARGET_BUILTIN(__builtin_ia32_vfmaddsd3, "V2dV2dV2dV2d", "", "fma|fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmsubps, "V4fV4fV4fV4f", "", "fma|fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmsubpd, "V2dV2dV2dV2d", "", "fma|fma4")
 TARGET_BUILTIN(__builtin_ia32_vfmsubss3, "V4fV4fV4fV4f", "", "fma|fma4")
 TARGET_BUILTIN(__builtin_ia32_vfmsubsd3, "V2dV2dV2dV2d", "", "fma|fma4")
-TARGET_BUILTIN(__builtin_ia32_vfnmaddps, "V4fV4fV4fV4f", "", "fma|fma4")
-TARGET_BUILTIN(__builtin_ia32_vfnmaddpd, "V2dV2dV2dV2d", "", "fma|fma4")
 TARGET_BUILTIN(__builtin_ia32_vfnmaddss3, "V4fV4fV4fV4f", "", "fma|fma4")
 TARGET_BUILTIN(__builtin_ia32_vfnmaddsd3, "V2dV2dV2dV2d", "", "fma|fma4")
-TARGET_BUILTIN(__builtin_ia32_vfnmsubps, "V4fV4fV4fV4f", "", "fma|fma4")
-TARGET_BUILTIN(__builtin_ia32_vfnmsubpd, "V2dV2dV2dV2d", "", "fma|fma4")
 TARGET_BUILTIN(__builtin_ia32_vfnmsubss3, "V4fV4fV4fV4f", "", "fma|fma4")
 TARGET_BUILTIN(__builtin_ia32_vfnmsubsd3, "V2dV2dV2dV2d", "", "fma|fma4")
 TARGET_BUILTIN(__builtin_ia32_vfmaddsubps, "V4fV4fV4fV4f", "", "fma|fma4")
 TARGET_BUILTIN(__builtin_ia32_vfmaddsubpd, "V2dV2dV2dV2d", "", "fma|fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmsubaddps, "V4fV4fV4fV4f", "", "fma|fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmsubaddpd, "V2dV2dV2dV2d", "", "fma|fma4")
 TARGET_BUILTIN(__builtin_ia32_vfmaddps256, "V8fV8fV8fV8f", "", "fma|fma4")
 TARGET_BUILTIN(__builtin_ia32_vfmaddpd256, "V4dV4dV4dV4d", "", "fma|fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmsubps256, "V8fV8fV8fV8f", "", "fma|fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmsubpd256, "V4dV4dV4dV4d", "", "fma|fma4")
 TARGET_BUILTIN(__builtin_ia32_vfnmaddps256, "V8fV8fV8fV8f", "", "fma|fma4")
 TARGET_BUILTIN(__builtin_ia32_vfnmaddpd256, "V4dV4dV4dV4d", "", "fma|fma4")
-TARGET_BUILTIN(__builtin_ia32_vfnmsubps256, "V8fV8fV8fV8f", "", "fma|fma4")
-TARGET_BUILTIN(__builtin_ia32_vfnmsubpd256, "V4dV4dV4dV4d", "", "fma|fma4")
 TARGET_BUILTIN(__builtin_ia32_vfmaddsubps256, "V8fV8fV8fV8f", "", "fma|fma4")
 TARGET_BUILTIN(__builtin_ia32_vfmaddsubpd256, "V4dV4dV4dV4d", "", "fma|fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmsubaddps256, "V8fV8fV8fV8f", "", "fma|fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmsubaddpd256, "V4dV4dV4dV4d", "", "fma|fma4")
 
 TARGET_BUILTIN(__builtin_ia32_vfmaddpd128_mask, "V2dV2dV2dV2dUc", "", "avx512vl")
 TARGET_BUILTIN(__builtin_ia32_vfmaddpd128_mask3, "V2dV2dV2dV2dUc", "", "avx512vl")

diff --git a/lib/Headers/fma4intrin.h b/lib/Headers/fma4intrin.h
index 8431821bba553e68b2b41da4fbd9d8e244e57466..cda20e7d810914a249f4d8767b1188de3ba11352 100644 (file)
--- a/lib/Headers/fma4intrin.h
+++ b/lib/Headers/fma4intrin.h
@@ -60,13 +60,13 @@ _mm_macc_sd(__m128d __A, __m128d __B, __m128d __C)
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_msub_ps(__m128 __A, __m128 __B, __m128 __C)
 {
-  return (__m128)__builtin_ia32_vfmsubps((__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
+  return (__m128)__builtin_ia32_vfmaddps((__v4sf)__A, (__v4sf)__B, -(__v4sf)__C);
 }
 
 static __inline__ __m128d __DEFAULT_FN_ATTRS
 _mm_msub_pd(__m128d __A, __m128d __B, __m128d __C)
 {
-  return (__m128d)__builtin_ia32_vfmsubpd((__v2df)__A, (__v2df)__B, (__v2df)__C);
+  return (__m128d)__builtin_ia32_vfmaddpd((__v2df)__A, (__v2df)__B, -(__v2df)__C);
 }
 
 static __inline__ __m128 __DEFAULT_FN_ATTRS
@@ -84,13 +84,13 @@ _mm_msub_sd(__m128d __A, __m128d __B, __m128d __C)
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_nmacc_ps(__m128 __A, __m128 __B, __m128 __C)
 {
-  return (__m128)__builtin_ia32_vfnmaddps((__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
+  return (__m128)__builtin_ia32_vfmaddps(-(__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
 }
 
 static __inline__ __m128d __DEFAULT_FN_ATTRS
 _mm_nmacc_pd(__m128d __A, __m128d __B, __m128d __C)
 {
-  return (__m128d)__builtin_ia32_vfnmaddpd((__v2df)__A, (__v2df)__B, (__v2df)__C);
+  return (__m128d)__builtin_ia32_vfmaddpd(-(__v2df)__A, (__v2df)__B, (__v2df)__C);
 }
 
 static __inline__ __m128 __DEFAULT_FN_ATTRS
@@ -108,13 +108,13 @@ _mm_nmacc_sd(__m128d __A, __m128d __B, __m128d __C)
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_nmsub_ps(__m128 __A, __m128 __B, __m128 __C)
 {
-  return (__m128)__builtin_ia32_vfnmsubps((__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
+  return (__m128)__builtin_ia32_vfmaddps(-(__v4sf)__A, (__v4sf)__B, -(__v4sf)__C);
 }
 
 static __inline__ __m128d __DEFAULT_FN_ATTRS
 _mm_nmsub_pd(__m128d __A, __m128d __B, __m128d __C)
 {
-  return (__m128d)__builtin_ia32_vfnmsubpd((__v2df)__A, (__v2df)__B, (__v2df)__C);
+  return (__m128d)__builtin_ia32_vfmaddpd(-(__v2df)__A, (__v2df)__B, -(__v2df)__C);
 }
 
 static __inline__ __m128 __DEFAULT_FN_ATTRS
@@ -144,13 +144,13 @@ _mm_maddsub_pd(__m128d __A, __m128d __B, __m128d __C)
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_msubadd_ps(__m128 __A, __m128 __B, __m128 __C)
 {
-  return (__m128)__builtin_ia32_vfmsubaddps((__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
+  return (__m128)__builtin_ia32_vfmaddsubps((__v4sf)__A, (__v4sf)__B, -(__v4sf)__C);
 }
 
 static __inline__ __m128d __DEFAULT_FN_ATTRS
 _mm_msubadd_pd(__m128d __A, __m128d __B, __m128d __C)
 {
-  return (__m128d)__builtin_ia32_vfmsubaddpd((__v2df)__A, (__v2df)__B, (__v2df)__C);
+  return (__m128d)__builtin_ia32_vfmaddsubpd((__v2df)__A, (__v2df)__B, -(__v2df)__C);
 }
 
 static __inline__ __m256 __DEFAULT_FN_ATTRS
@@ -168,37 +168,37 @@ _mm256_macc_pd(__m256d __A, __m256d __B, __m256d __C)
 static __inline__ __m256 __DEFAULT_FN_ATTRS
 _mm256_msub_ps(__m256 __A, __m256 __B, __m256 __C)
 {
-  return (__m256)__builtin_ia32_vfmsubps256((__v8sf)__A, (__v8sf)__B, (__v8sf)__C);
+  return (__m256)__builtin_ia32_vfmaddps256((__v8sf)__A, (__v8sf)__B, -(__v8sf)__C);
 }
 
 static __inline__ __m256d __DEFAULT_FN_ATTRS
 _mm256_msub_pd(__m256d __A, __m256d __B, __m256d __C)
 {
-  return (__m256d)__builtin_ia32_vfmsubpd256((__v4df)__A, (__v4df)__B, (__v4df)__C);
+  return (__m256d)__builtin_ia32_vfmaddpd256((__v4df)__A, (__v4df)__B, -(__v4df)__C);
 }
 
 static __inline__ __m256 __DEFAULT_FN_ATTRS
 _mm256_nmacc_ps(__m256 __A, __m256 __B, __m256 __C)
 {
-  return (__m256)__builtin_ia32_vfnmaddps256((__v8sf)__A, (__v8sf)__B, (__v8sf)__C);
+  return (__m256)__builtin_ia32_vfmaddps256(-(__v8sf)__A, (__v8sf)__B, (__v8sf)__C);
 }
 
 static __inline__ __m256d __DEFAULT_FN_ATTRS
 _mm256_nmacc_pd(__m256d __A, __m256d __B, __m256d __C)
 {
-  return (__m256d)__builtin_ia32_vfnmaddpd256((__v4df)__A, (__v4df)__B, (__v4df)__C);
+  return (__m256d)__builtin_ia32_vfmaddpd256(-(__v4df)__A, (__v4df)__B, (__v4df)__C);
 }
 
 static __inline__ __m256 __DEFAULT_FN_ATTRS
 _mm256_nmsub_ps(__m256 __A, __m256 __B, __m256 __C)
 {
-  return (__m256)__builtin_ia32_vfnmsubps256((__v8sf)__A, (__v8sf)__B, (__v8sf)__C);
+  return (__m256)__builtin_ia32_vfmaddps256(-(__v8sf)__A, (__v8sf)__B, -(__v8sf)__C);
 }
 
 static __inline__ __m256d __DEFAULT_FN_ATTRS
 _mm256_nmsub_pd(__m256d __A, __m256d __B, __m256d __C)
 {
-  return (__m256d)__builtin_ia32_vfnmsubpd256((__v4df)__A, (__v4df)__B, (__v4df)__C);
+  return (__m256d)__builtin_ia32_vfmaddpd256(-(__v4df)__A, (__v4df)__B, -(__v4df)__C);
 }
 
 static __inline__ __m256 __DEFAULT_FN_ATTRS
@@ -216,13 +216,13 @@ _mm256_maddsub_pd(__m256d __A, __m256d __B, __m256d __C)
 static __inline__ __m256 __DEFAULT_FN_ATTRS
 _mm256_msubadd_ps(__m256 __A, __m256 __B, __m256 __C)
 {
-  return (__m256)__builtin_ia32_vfmsubaddps256((__v8sf)__A, (__v8sf)__B, (__v8sf)__C);
+  return (__m256)__builtin_ia32_vfmaddsubps256((__v8sf)__A, (__v8sf)__B, -(__v8sf)__C);
 }
 
 static __inline__ __m256d __DEFAULT_FN_ATTRS
 _mm256_msubadd_pd(__m256d __A, __m256d __B, __m256d __C)
 {
-  return (__m256d)__builtin_ia32_vfmsubaddpd256((__v4df)__A, (__v4df)__B, (__v4df)__C);
+  return (__m256d)__builtin_ia32_vfmaddsubpd256((__v4df)__A, (__v4df)__B, -(__v4df)__C);
 }
 
 #undef __DEFAULT_FN_ATTRS

diff --git a/lib/Headers/fmaintrin.h b/lib/Headers/fmaintrin.h
index 235931ccb6bc4ce19c62a609189b91d0bdc2ceb6..86a1198879b92a59b21605590e156d351689dbab 100644 (file)
--- a/lib/Headers/fmaintrin.h
+++ b/lib/Headers/fmaintrin.h
@@ -58,13 +58,13 @@ _mm_fmadd_sd(__m128d __A, __m128d __B, __m128d __C)
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_fmsub_ps(__m128 __A, __m128 __B, __m128 __C)
 {
-  return (__m128)__builtin_ia32_vfmsubps((__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
+  return (__m128)__builtin_ia32_vfmaddps((__v4sf)__A, (__v4sf)__B, -(__v4sf)__C);
 }
 
 static __inline__ __m128d __DEFAULT_FN_ATTRS
 _mm_fmsub_pd(__m128d __A, __m128d __B, __m128d __C)
 {
-  return (__m128d)__builtin_ia32_vfmsubpd((__v2df)__A, (__v2df)__B, (__v2df)__C);
+  return (__m128d)__builtin_ia32_vfmaddpd((__v2df)__A, (__v2df)__B, -(__v2df)__C);
 }
 
 static __inline__ __m128 __DEFAULT_FN_ATTRS
@@ -82,13 +82,13 @@ _mm_fmsub_sd(__m128d __A, __m128d __B, __m128d __C)
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_fnmadd_ps(__m128 __A, __m128 __B, __m128 __C)
 {
-  return (__m128)__builtin_ia32_vfnmaddps((__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
+  return (__m128)__builtin_ia32_vfmaddps(-(__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
 }
 
 static __inline__ __m128d __DEFAULT_FN_ATTRS
 _mm_fnmadd_pd(__m128d __A, __m128d __B, __m128d __C)
 {
-  return (__m128d)__builtin_ia32_vfnmaddpd((__v2df)__A, (__v2df)__B, (__v2df)__C);
+  return (__m128d)__builtin_ia32_vfmaddpd(-(__v2df)__A, (__v2df)__B, (__v2df)__C);
 }
 
 static __inline__ __m128 __DEFAULT_FN_ATTRS
@@ -106,13 +106,13 @@ _mm_fnmadd_sd(__m128d __A, __m128d __B, __m128d __C)
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_fnmsub_ps(__m128 __A, __m128 __B, __m128 __C)
 {
-  return (__m128)__builtin_ia32_vfnmsubps((__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
+  return (__m128)__builtin_ia32_vfmaddps(-(__v4sf)__A, (__v4sf)__B, -(__v4sf)__C);
 }
 
 static __inline__ __m128d __DEFAULT_FN_ATTRS
 _mm_fnmsub_pd(__m128d __A, __m128d __B, __m128d __C)
 {
-  return (__m128d)__builtin_ia32_vfnmsubpd((__v2df)__A, (__v2df)__B, (__v2df)__C);
+  return (__m128d)__builtin_ia32_vfmaddpd(-(__v2df)__A, (__v2df)__B, -(__v2df)__C);
 }
 
 static __inline__ __m128 __DEFAULT_FN_ATTRS
@@ -142,13 +142,13 @@ _mm_fmaddsub_pd(__m128d __A, __m128d __B, __m128d __C)
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_fmsubadd_ps(__m128 __A, __m128 __B, __m128 __C)
 {
-  return (__m128)__builtin_ia32_vfmsubaddps((__v4sf)__A, (__v4sf)__B, (__v4sf)__C);
+  return (__m128)__builtin_ia32_vfmaddsubps((__v4sf)__A, (__v4sf)__B, -(__v4sf)__C);
 }
 
 static __inline__ __m128d __DEFAULT_FN_ATTRS
 _mm_fmsubadd_pd(__m128d __A, __m128d __B, __m128d __C)
 {
-  return (__m128d)__builtin_ia32_vfmsubaddpd((__v2df)__A, (__v2df)__B, (__v2df)__C);
+  return (__m128d)__builtin_ia32_vfmaddsubpd((__v2df)__A, (__v2df)__B, -(__v2df)__C);
 }
 
 static __inline__ __m256 __DEFAULT_FN_ATTRS
@@ -166,37 +166,37 @@ _mm256_fmadd_pd(__m256d __A, __m256d __B, __m256d __C)
 static __inline__ __m256 __DEFAULT_FN_ATTRS
 _mm256_fmsub_ps(__m256 __A, __m256 __B, __m256 __C)
 {
-  return (__m256)__builtin_ia32_vfmsubps256((__v8sf)__A, (__v8sf)__B, (__v8sf)__C);
+  return (__m256)__builtin_ia32_vfmaddps256((__v8sf)__A, (__v8sf)__B, -(__v8sf)__C);
 }
 
 static __inline__ __m256d __DEFAULT_FN_ATTRS
 _mm256_fmsub_pd(__m256d __A, __m256d __B, __m256d __C)
 {
-  return (__m256d)__builtin_ia32_vfmsubpd256((__v4df)__A, (__v4df)__B, (__v4df)__C);
+  return (__m256d)__builtin_ia32_vfmaddpd256((__v4df)__A, (__v4df)__B, -(__v4df)__C);
 }
 
 static __inline__ __m256 __DEFAULT_FN_ATTRS
 _mm256_fnmadd_ps(__m256 __A, __m256 __B, __m256 __C)
 {
-  return (__m256)__builtin_ia32_vfnmaddps256((__v8sf)__A, (__v8sf)__B, (__v8sf)__C);
+  return (__m256)__builtin_ia32_vfmaddps256(-(__v8sf)__A, (__v8sf)__B, (__v8sf)__C);
 }
 
 static __inline__ __m256d __DEFAULT_FN_ATTRS
 _mm256_fnmadd_pd(__m256d __A, __m256d __B, __m256d __C)
 {
-  return (__m256d)__builtin_ia32_vfnmaddpd256((__v4df)__A, (__v4df)__B, (__v4df)__C);
+  return (__m256d)__builtin_ia32_vfmaddpd256(-(__v4df)__A, (__v4df)__B, (__v4df)__C);
 }
 
 static __inline__ __m256 __DEFAULT_FN_ATTRS
 _mm256_fnmsub_ps(__m256 __A, __m256 __B, __m256 __C)
 {
-  return (__m256)__builtin_ia32_vfnmsubps256((__v8sf)__A, (__v8sf)__B, (__v8sf)__C);
+  return (__m256)__builtin_ia32_vfmaddps256(-(__v8sf)__A, (__v8sf)__B, -(__v8sf)__C);
 }
 
 static __inline__ __m256d __DEFAULT_FN_ATTRS
 _mm256_fnmsub_pd(__m256d __A, __m256d __B, __m256d __C)
 {
-  return (__m256d)__builtin_ia32_vfnmsubpd256((__v4df)__A, (__v4df)__B, (__v4df)__C);
+  return (__m256d)__builtin_ia32_vfmaddpd256(-(__v4df)__A, (__v4df)__B, -(__v4df)__C);
 }
 
 static __inline__ __m256 __DEFAULT_FN_ATTRS
@@ -214,13 +214,13 @@ _mm256_fmaddsub_pd(__m256d __A, __m256d __B, __m256d __C)
 static __inline__ __m256 __DEFAULT_FN_ATTRS
 _mm256_fmsubadd_ps(__m256 __A, __m256 __B, __m256 __C)
 {
-  return (__m256)__builtin_ia32_vfmsubaddps256((__v8sf)__A, (__v8sf)__B, (__v8sf)__C);
+  return (__m256)__builtin_ia32_vfmaddsubps256((__v8sf)__A, (__v8sf)__B, -(__v8sf)__C);
 }
 
 static __inline__ __m256d __DEFAULT_FN_ATTRS
 _mm256_fmsubadd_pd(__m256d __A, __m256d __B, __m256d __C)
 {
-  return (__m256d)__builtin_ia32_vfmsubaddpd256((__v4df)__A, (__v4df)__B, (__v4df)__C);
+  return (__m256d)__builtin_ia32_vfmaddsubpd256((__v4df)__A, (__v4df)__B, -(__v4df)__C);
 }
 
 #undef __DEFAULT_FN_ATTRS

diff --git a/test/CodeGen/fma-builtins.c b/test/CodeGen/fma-builtins.c
index 91a2efe7637ea08483a4a7a70c882c591c30b96e..d5905d33a97957df22ab17aef5ea42eda169c6dd 100644 (file)
--- a/test/CodeGen/fma-builtins.c
+++ b/test/CodeGen/fma-builtins.c
@@ -4,161 +4,213 @@
 #include <immintrin.h>
 
 __m128 test_mm_fmadd_ps(__m128 a, __m128 b, __m128 c) {
+  // CHECK-LABEL: test_mm_fmadd_ps
   // CHECK: @llvm.x86.fma.vfmadd.ps
   return _mm_fmadd_ps(a, b, c);
 }
 
 __m128d test_mm_fmadd_pd(__m128d a, __m128d b, __m128d c) {
+  // CHECK-LABEL: test_mm_fmadd_pd
   // CHECK: @llvm.x86.fma.vfmadd.pd
   return _mm_fmadd_pd(a, b, c);
 }
 
 __m128 test_mm_fmadd_ss(__m128 a, __m128 b, __m128 c) {
+  // CHECK-LABEL: test_mm_fmadd_ss
   // CHECK: @llvm.x86.fma.vfmadd.ss
   return _mm_fmadd_ss(a, b, c);
 }
 
 __m128d test_mm_fmadd_sd(__m128d a, __m128d b, __m128d c) {
+  // CHECK-LABEL: test_mm_fmadd_sd
   // CHECK: @llvm.x86.fma.vfmadd.sd
   return _mm_fmadd_sd(a, b, c);
 }
 
 __m128 test_mm_fmsub_ps(__m128 a, __m128 b, __m128 c) {
-  // CHECK: @llvm.x86.fma.vfmsub.ps
+  // CHECK-LABEL: test_mm_fmsub_ps
+  // CHECK: [[NEG:%.+]] = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmadd.ps(<4 x float> %{{.+}}, <4 x float> %{{.+}}, <4 x float> [[NEG]])
   return _mm_fmsub_ps(a, b, c);
 }
 
 __m128d test_mm_fmsub_pd(__m128d a, __m128d b, __m128d c) {
-  // CHECK: @llvm.x86.fma.vfmsub.pd
+  // CHECK-LABEL: test_mm_fmsub_pd
+  // CHECK: [[NEG:%.+]] = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmadd.pd(<2 x double> %{{.+}}, <2 x double> %{{.+}}, <2 x double> [[NEG]])
   return _mm_fmsub_pd(a, b, c);
 }
 
 __m128 test_mm_fmsub_ss(__m128 a, __m128 b, __m128 c) {
+  // CHECK-LABEL: test_mm_fmsub_ss
   // CHECK: @llvm.x86.fma.vfmsub.ss
   return _mm_fmsub_ss(a, b, c);
 }
 
 __m128d test_mm_fmsub_sd(__m128d a, __m128d b, __m128d c) {
+  // CHECK-LABEL: test_mm_fmsub_sd
   // CHECK: @llvm.x86.fma.vfmsub.sd
   return _mm_fmsub_sd(a, b, c);
 }
 
 __m128 test_mm_fnmadd_ps(__m128 a, __m128 b, __m128 c) {
-  // CHECK: @llvm.x86.fma.vfnmadd.ps
+  // CHECK-LABEL: test_mm_fnmadd_ps
+  // CHECK: [[NEG:%.+]] = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmadd.ps(<4 x float> [[NEG]], <4 x float> %{{.+}}, <4 x float> %{{.+}})
   return _mm_fnmadd_ps(a, b, c);
 }
 
 __m128d test_mm_fnmadd_pd(__m128d a, __m128d b, __m128d c) {
-  // CHECK: @llvm.x86.fma.vfnmadd.pd
+  // CHECK-LABEL: test_mm_fnmadd_pd
+  // CHECK: [[NEG:%.+]] = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmadd.pd(<2 x double> [[NEG]], <2 x double> %{{.+}}, <2 x double> %{{.+}})
   return _mm_fnmadd_pd(a, b, c);
 }
 
 __m128 test_mm_fnmadd_ss(__m128 a, __m128 b, __m128 c) {
+  // CHECK-LABEL: test_mm_fnmadd_ss
   // CHECK: @llvm.x86.fma.vfnmadd.ss
   return _mm_fnmadd_ss(a, b, c);
 }
 
 __m128d test_mm_fnmadd_sd(__m128d a, __m128d b, __m128d c) {
+  // CHECK-LABEL: test_mm_fnmadd_sd
   // CHECK: @llvm.x86.fma.vfnmadd.sd
   return _mm_fnmadd_sd(a, b, c);
 }
 
 __m128 test_mm_fnmsub_ps(__m128 a, __m128 b, __m128 c) {
-  // CHECK: @llvm.x86.fma.vfnmsub.ps
+  // CHECK-LABEL: test_mm_fnmsub_ps
+  // CHECK: [[NEG:%.+]] = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %{{.+}}
+  // CHECK: [[NEG2:%.+]] = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmadd.ps(<4 x float> [[NEG]], <4 x float> %{{.+}}, <4 x float> [[NEG2]])
   return _mm_fnmsub_ps(a, b, c);
 }
 
 __m128d test_mm_fnmsub_pd(__m128d a, __m128d b, __m128d c) {
-  // CHECK: @llvm.x86.fma.vfnmsub.pd
+  // CHECK-LABEL: test_mm_fnmsub_pd
+  // CHECK: [[NEG:%.+]] = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, %{{.+}}
+  // CHECK: [[NEG2:%.+]] = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmadd.pd(<2 x double> [[NEG]], <2 x double> %{{.+}}, <2 x double> [[NEG2]])
   return _mm_fnmsub_pd(a, b, c);
 }
 
 __m128 test_mm_fnmsub_ss(__m128 a, __m128 b, __m128 c) {
+  // CHECK-LABEL: test_mm_fnmsub_ss
   // CHECK: @llvm.x86.fma.vfnmsub.ss
   return _mm_fnmsub_ss(a, b, c);
 }
 
 __m128d test_mm_fnmsub_sd(__m128d a, __m128d b, __m128d c) {
+  // CHECK-LABEL: test_mm_fnmsub_sd
   // CHECK: @llvm.x86.fma.vfnmsub.sd
   return _mm_fnmsub_sd(a, b, c);
 }
 
 __m128 test_mm_fmaddsub_ps(__m128 a, __m128 b, __m128 c) {
+  // CHECK-LABEL: test_mm_fmaddsub_ps
   // CHECK: @llvm.x86.fma.vfmaddsub.ps
   return _mm_fmaddsub_ps(a, b, c);
 }
 
 __m128d test_mm_fmaddsub_pd(__m128d a, __m128d b, __m128d c) {
+  // CHECK-LABEL: test_mm_fmaddsub_pd
   // CHECK: @llvm.x86.fma.vfmaddsub.pd
   return _mm_fmaddsub_pd(a, b, c);
 }
 
 __m128 test_mm_fmsubadd_ps(__m128 a, __m128 b, __m128 c) {
-  // CHECK: @llvm.x86.fma.vfmsubadd.ps
+  // CHECK-LABEL: test_mm_fmsubadd_ps
+  // CHECK: [[NEG:%.+]] = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmaddsub.ps(<4 x float> %{{.+}}, <4 x float> %{{.+}}, <4 x float> [[NEG]])
   return _mm_fmsubadd_ps(a, b, c);
 }
 
 __m128d test_mm_fmsubadd_pd(__m128d a, __m128d b, __m128d c) {
-  // CHECK: @llvm.x86.fma.vfmsubadd.pd
+  // CHECK-LABEL: test_mm_fmsubadd_pd
+  // CHECK: [[NEG:%.+]] = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmaddsub.pd(<2 x double> %{{.+}}, <2 x double> %{{.+}}, <2 x double> [[NEG]])
   return _mm_fmsubadd_pd(a, b, c);
 }
 
 __m256 test_mm256_fmadd_ps(__m256 a, __m256 b, __m256 c) {
+  // CHECK-LABEL: test_mm256_fmadd_ps
   // CHECK: @llvm.x86.fma.vfmadd.ps.256
   return _mm256_fmadd_ps(a, b, c);
 }
 
 __m256d test_mm256_fmadd_pd(__m256d a, __m256d b, __m256d c) {
+  // CHECK-LABEL: test_mm256_fmadd_pd
   // CHECK: @llvm.x86.fma.vfmadd.pd.256
   return _mm256_fmadd_pd(a, b, c);
 }
 
 __m256 test_mm256_fmsub_ps(__m256 a, __m256 b, __m256 c) {
-  // CHECK: @llvm.x86.fma.vfmsub.ps.256
+  // CHECK-LABEL: test_mm256_fmsub_ps
+  // CHECK: [[NEG:%.+]] = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %{{.*}}
+  // CHECK: @llvm.x86.fma.vfmadd.ps.256(<8 x float> %{{.+}}, <8 x float> %{{.+}}, <8 x float> [[NEG]])
   return _mm256_fmsub_ps(a, b, c);
 }
 
 __m256d test_mm256_fmsub_pd(__m256d a, __m256d b, __m256d c) {
-  // CHECK: @llvm.x86.fma.vfmsub.pd.256
+  // CHECK-LABEL: test_mm256_fmsub_pd
+  // CHECK: [[NEG:%.+]] = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmadd.pd.256(<4 x double> %{{.+}}, <4 x double> %{{.+}}, <4 x double> [[NEG]])
   return _mm256_fmsub_pd(a, b, c);
 }
 
 __m256 test_mm256_fnmadd_ps(__m256 a, __m256 b, __m256 c) {
-  // CHECK: @llvm.x86.fma.vfnmadd.ps.256
+  // CHECK-LABEL: test_mm256_fnmadd_ps
+  // CHECK: [[NEG:%.+]] = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %{{.*}}
+  // CHECK: @llvm.x86.fma.vfmadd.ps.256(<8 x float> [[NEG]], <8 x float> %{{.+}}, <8 x float> %{{.+}})
   return _mm256_fnmadd_ps(a, b, c);
 }
 
 __m256d test_mm256_fnmadd_pd(__m256d a, __m256d b, __m256d c) {
-  // CHECK: @llvm.x86.fma.vfnmadd.pd.256
+  // CHECK-LABEL: test_mm256_fnmadd_pd
+  // CHECK: [[NEG:%.+]] = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmadd.pd.256(<4 x double> [[NEG]], <4 x double> %{{.+}}, <4 x double> %{{.+}})
   return _mm256_fnmadd_pd(a, b, c);
 }
 
 __m256 test_mm256_fnmsub_ps(__m256 a, __m256 b, __m256 c) {
-  // CHECK: @llvm.x86.fma.vfnmsub.ps.256
+  // CHECK-LABEL: test_mm256_fnmsub_ps
+  // CHECK: [[NEG:%.+]] = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %{{.*}}
+  // CHECK: [[NEG2:%.+]] = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %{{.*}}
+  // CHECK: @llvm.x86.fma.vfmadd.ps.256(<8 x float> [[NEG]], <8 x float> %{{.+}}, <8 x float> [[NEG2]])
   return _mm256_fnmsub_ps(a, b, c);
 }
 
 __m256d test_mm256_fnmsub_pd(__m256d a, __m256d b, __m256d c) {
-  // CHECK: @llvm.x86.fma.vfnmsub.pd.256
+  // CHECK-LABEL: test_mm256_fnmsub_pd
+  // CHECK: [[NEG:%.+]] = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, %{{.+}}
+  // CHECK: [[NEG2:%.+]] = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmadd.pd.256(<4 x double> [[NEG]], <4 x double> %{{.+}}, <4 x double> [[NEG2]])
   return _mm256_fnmsub_pd(a, b, c);
 }
 
 __m256 test_mm256_fmaddsub_ps(__m256 a, __m256 b, __m256 c) {
+  // CHECK-LABEL: test_mm256_fmaddsub_ps
   // CHECK: @llvm.x86.fma.vfmaddsub.ps.256
   return _mm256_fmaddsub_ps(a, b, c);
 }
 
 __m256d test_mm256_fmaddsub_pd(__m256d a, __m256d b, __m256d c) {
+  // CHECK-LABEL: test_mm256_fmaddsub_pd
   // CHECK: @llvm.x86.fma.vfmaddsub.pd.256
   return _mm256_fmaddsub_pd(a, b, c);
 }
 
 __m256 test_mm256_fmsubadd_ps(__m256 a, __m256 b, __m256 c) {
-  // CHECK: @llvm.x86.fma.vfmsubadd.ps.256
+  // CHECK-LABEL: test_mm256_fmsubadd_ps
+  // CHECK: [[NEG:%.+]] = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %{{.*}}
+  // CHECK: @llvm.x86.fma.vfmaddsub.ps.256(<8 x float> %{{.*}}, <8 x float> %{{.+}}, <8 x float> [[NEG]])
   return _mm256_fmsubadd_ps(a, b, c);
 }
 
 __m256d test_mm256_fmsubadd_pd(__m256d a, __m256d b, __m256d c) {
-  // CHECK: @llvm.x86.fma.vfmsubadd.pd.256
+  // CHECK-LABEL: test_mm256_fmsubadd_pd
+  // CHECK: [[NEG:%.+]] = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmaddsub.pd.256(<4 x double> %{{.+}}, <4 x double> %{{.+}}, <4 x double> [[NEG]])
   return _mm256_fmsubadd_pd(a, b, c);
 }

diff --git a/test/CodeGen/fma4-builtins.c b/test/CodeGen/fma4-builtins.c
index 3ca5fac68892ea5446e381bbb69e8b01c47bc9ad..68a2ec6e3a99a4cfbf4fc9e3b41590106fcf7306 100644 (file)
--- a/test/CodeGen/fma4-builtins.c
+++ b/test/CodeGen/fma4-builtins.c
@@ -29,13 +29,15 @@ __m128d test_mm_macc_sd(__m128d a, __m128d b, __m128d c) {
 
 __m128 test_mm_msub_ps(__m128 a, __m128 b, __m128 c) {
   // CHECK-LABEL: test_mm_msub_ps
-  // CHECK: @llvm.x86.fma.vfmsub.ps
+  // CHECK: [[NEG:%.+]] = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmadd.ps(<4 x float> %{{.+}}, <4 x float> %{{.+}}, <4 x float> [[NEG]])
   return _mm_msub_ps(a, b, c);
 }
 
 __m128d test_mm_msub_pd(__m128d a, __m128d b, __m128d c) {
   // CHECK-LABEL: test_mm_msub_pd
-  // CHECK: @llvm.x86.fma.vfmsub.pd
+  // CHECK: [[NEG:%.+]] = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmadd.pd(<2 x double> %{{.+}}, <2 x double> %{{.+}}, <2 x double> [[NEG]])
   return _mm_msub_pd(a, b, c);
 }
 
@@ -53,13 +55,15 @@ __m128d test_mm_msub_sd(__m128d a, __m128d b, __m128d c) {
 
 __m128 test_mm_nmacc_ps(__m128 a, __m128 b, __m128 c) {
   // CHECK-LABEL: test_mm_nmacc_ps
-  // CHECK: @llvm.x86.fma.vfnmadd.ps
+  // CHECK: [[NEG:%.+]] = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmadd.ps(<4 x float> [[NEG]], <4 x float> %{{.+}}, <4 x float> %{{.+}})
   return _mm_nmacc_ps(a, b, c);
 }
 
 __m128d test_mm_nmacc_pd(__m128d a, __m128d b, __m128d c) {
   // CHECK-LABEL: test_mm_nmacc_pd
-  // CHECK: @llvm.x86.fma.vfnmadd.pd
+  // CHECK: [[NEG:%.+]] = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmadd.pd(<2 x double> [[NEG]], <2 x double> %{{.+}}, <2 x double> %{{.+}})
   return _mm_nmacc_pd(a, b, c);
 }
 
@@ -77,13 +81,17 @@ __m128d test_mm_nmacc_sd(__m128d a, __m128d b, __m128d c) {
 
 __m128 test_mm_nmsub_ps(__m128 a, __m128 b, __m128 c) {
   // CHECK-LABEL: test_mm_nmsub_ps
-  // CHECK: @llvm.x86.fma.vfnmsub.ps
+  // CHECK: [[NEG:%.+]] = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %{{.+}}
+  // CHECK: [[NEG2:%.+]] = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmadd.ps(<4 x float> [[NEG]], <4 x float> %{{.+}}, <4 x float> [[NEG2]])
   return _mm_nmsub_ps(a, b, c);
 }
 
 __m128d test_mm_nmsub_pd(__m128d a, __m128d b, __m128d c) {
   // CHECK-LABEL: test_mm_nmsub_pd
-  // CHECK: @llvm.x86.fma.vfnmsub.pd
+  // CHECK: [[NEG:%.+]] = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, %{{.+}}
+  // CHECK: [[NEG2:%.+]] = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmadd.pd(<2 x double> [[NEG]], <2 x double> %{{.+}}, <2 x double> [[NEG2]])
   return _mm_nmsub_pd(a, b, c);
 }
 
@@ -113,13 +121,15 @@ __m128d test_mm_maddsub_pd(__m128d a, __m128d b, __m128d c) {
 
 __m128 test_mm_msubadd_ps(__m128 a, __m128 b, __m128 c) {
   // CHECK-LABEL: test_mm_msubadd_ps
-  // CHECK: @llvm.x86.fma.vfmsubadd.ps
+  // CHECK: [[NEG:%.+]] = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmaddsub.ps(<4 x float> %{{.+}}, <4 x float> %{{.+}}, <4 x float> [[NEG]])
   return _mm_msubadd_ps(a, b, c);
 }
 
 __m128d test_mm_msubadd_pd(__m128d a, __m128d b, __m128d c) {
   // CHECK-LABEL: test_mm_msubadd_pd
-  // CHECK: @llvm.x86.fma.vfmsubadd.pd
+  // CHECK: [[NEG:%.+]] = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmaddsub.pd(<2 x double> %{{.+}}, <2 x double> %{{.+}}, <2 x double> [[NEG]])
   return _mm_msubadd_pd(a, b, c);
 }
 
@@ -137,37 +147,45 @@ __m256d test_mm256_macc_pd(__m256d a, __m256d b, __m256d c) {
 
 __m256 test_mm256_msub_ps(__m256 a, __m256 b, __m256 c) {
   // CHECK-LABEL: test_mm256_msub_ps
-  // CHECK: @llvm.x86.fma.vfmsub.ps.256
+  // CHECK: [[NEG:%.+]] = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %{{.*}}
+  // CHECK: @llvm.x86.fma.vfmadd.ps.256(<8 x float> %{{.+}}, <8 x float> %{{.+}}, <8 x float> [[NEG]])
   return _mm256_msub_ps(a, b, c);
 }
 
 __m256d test_mm256_msub_pd(__m256d a, __m256d b, __m256d c) {
   // CHECK-LABEL: test_mm256_msub_pd
-  // CHECK: @llvm.x86.fma.vfmsub.pd.256
+  // CHECK: [[NEG:%.+]] = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmadd.pd.256(<4 x double> %{{.+}}, <4 x double> %{{.+}}, <4 x double> [[NEG]])
   return _mm256_msub_pd(a, b, c);
 }
 
 __m256 test_mm256_nmacc_ps(__m256 a, __m256 b, __m256 c) {
   // CHECK-LABEL: test_mm256_nmacc_ps
-  // CHECK: @llvm.x86.fma.vfnmadd.ps.256
+  // CHECK: [[NEG:%.+]] = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %{{.*}}
+  // CHECK: @llvm.x86.fma.vfmadd.ps.256(<8 x float> [[NEG]], <8 x float> %{{.+}}, <8 x float> %{{.+}})
   return _mm256_nmacc_ps(a, b, c);
 }
 
 __m256d test_mm256_nmacc_pd(__m256d a, __m256d b, __m256d c) {
   // CHECK-LABEL: test_mm256_nmacc_pd
-  // CHECK: @llvm.x86.fma.vfnmadd.pd.256
+  // CHECK: [[NEG:%.+]] = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmadd.pd.256(<4 x double> [[NEG]], <4 x double> %{{.+}}, <4 x double> %{{.+}})
   return _mm256_nmacc_pd(a, b, c);
 }
 
 __m256 test_mm256_nmsub_ps(__m256 a, __m256 b, __m256 c) {
   // CHECK-LABEL: test_mm256_nmsub_ps
-  // CHECK: @llvm.x86.fma.vfnmsub.ps.256
+  // CHECK: [[NEG:%.+]] = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %{{.*}}
+  // CHECK: [[NEG2:%.+]] = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %{{.*}}
+  // CHECK: @llvm.x86.fma.vfmadd.ps.256(<8 x float> [[NEG]], <8 x float> %{{.+}}, <8 x float> [[NEG2]])
   return _mm256_nmsub_ps(a, b, c);
 }
 
 __m256d test_mm256_nmsub_pd(__m256d a, __m256d b, __m256d c) {
   // CHECK-LABEL: test_mm256_nmsub_pd
-  // CHECK: @llvm.x86.fma.vfnmsub.pd.256
+  // CHECK: [[NEG:%.+]] = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, %{{.+}}
+  // CHECK: [[NEG2:%.+]] = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmadd.pd.256(<4 x double> [[NEG]], <4 x double> %{{.+}}, <4 x double> [[NEG2]])
   return _mm256_nmsub_pd(a, b, c);
 }
 
@@ -185,12 +203,14 @@ __m256d test_mm256_maddsub_pd(__m256d a, __m256d b, __m256d c) {
 
 __m256 test_mm256_msubadd_ps(__m256 a, __m256 b, __m256 c) {
   // CHECK-LABEL: test_mm256_msubadd_ps
-  // CHECK: @llvm.x86.fma.vfmsubadd.ps.256
+  // CHECK: [[NEG:%.+]] = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %{{.*}}
+  // CHECK: @llvm.x86.fma.vfmaddsub.ps.256(<8 x float> %{{.*}}, <8 x float> %{{.+}}, <8 x float> [[NEG]])
   return _mm256_msubadd_ps(a, b, c);
 }
 
 __m256d test_mm256_msubadd_pd(__m256d a, __m256d b, __m256d c) {
   // CHECK-LABEL: test_mm256_msubadd_pd
-  // CHECK: @llvm.x86.fma.vfmsubadd.pd.256
+  // CHECK: [[NEG:%.+]] = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, %{{.+}}
+  // CHECK: @llvm.x86.fma.vfmaddsub.pd.256(<4 x double> %{{.+}}, <4 x double> %{{.+}}, <4 x double> [[NEG]])
   return _mm256_msubadd_pd(a, b, c);
 }