Skip to content

[Headers][X86] Group related AVX512VL FMA intrinsics together (NFC) #156794

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 1 commit into from
Sep 4, 2025
Merged

[Headers][X86] Group related AVX512VL FMA intrinsics together (NFC) #156794

merged 1 commit into from
Sep 4, 2025

Conversation

ckoparkar
Copy link
Contributor

Follow-up of #156385.

@llvmbot llvmbot added clang Clang issues not falling into any other category backend:X86 clang:headers Headers provided by Clang, e.g. for intrinsics labels Sep 4, 2025
@llvmbot
Copy link
Member

llvmbot commented Sep 4, 2025 •
edited
Loading

@llvm/pr-subscribers-clang

@llvm/pr-subscribers-backend-x86

Author: Chaitanya Koparkar (ckoparkar)

Changes

Follow-up of #156385.

Patch is 74.58 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/156794.diff

2 Files Affected:

  • (modified) clang/lib/Headers/avx512vlintrin.h (+96-96)
  • (modified) clang/test/CodeGen/X86/avx512vl-builtins.c (+265-265)
diff --git a/clang/lib/Headers/avx512vlintrin.h b/clang/lib/Headers/avx512vlintrin.h
index aa186c136bb53..ee4b7110e2ce7 100644
--- a/clang/lib/Headers/avx512vlintrin.h
+++ b/clang/lib/Headers/avx512vlintrin.h
@@ -925,6 +925,12 @@ _mm_mask_fmsub_pd(__m128d __A, __mmask8 __U, __m128d __B, __m128d __C) {
       (__mmask8)__U, (__v2df)_mm_fmsub_pd(__A, __B, __C), (__v2df)__A);
 }
 
+static __inline__ __m128d __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_mask3_fmsub_pd(__m128d __A, __m128d __B, __m128d __C, __mmask8 __U) {
+  return (__m128d)__builtin_ia32_selectpd_128(
+      (__mmask8)__U, (__v2df)_mm_fmsub_pd(__A, __B, __C), (__v2df)__C);
+}
+
 static __inline__ __m128d __DEFAULT_FN_ATTRS128_CONSTEXPR
 _mm_maskz_fmsub_pd(__mmask8 __U, __m128d __A, __m128d __B, __m128d __C) {
   return (__m128d)__builtin_ia32_selectpd_128(
@@ -932,6 +938,12 @@ _mm_maskz_fmsub_pd(__mmask8 __U, __m128d __A, __m128d __B, __m128d __C) {
       (__v2df)_mm_setzero_pd());
 }
 
+static __inline__ __m128d __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_mask_fnmadd_pd(__m128d __A, __mmask8 __U, __m128d __B, __m128d __C) {
+  return (__m128d)__builtin_ia32_selectpd_128(
+      (__mmask8)__U, (__v2df)_mm_fnmadd_pd(__A, __B, __C), (__v2df)__A);
+}
+
 static __inline__ __m128d __DEFAULT_FN_ATTRS128_CONSTEXPR
 _mm_mask3_fnmadd_pd(__m128d __A, __m128d __B, __m128d __C, __mmask8 __U) {
   return (__m128d)__builtin_ia32_selectpd_128(
@@ -945,6 +957,18 @@ _mm_maskz_fnmadd_pd(__mmask8 __U, __m128d __A, __m128d __B, __m128d __C) {
       (__v2df)_mm_setzero_pd());
 }
 
+static __inline__ __m128d __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_mask_fnmsub_pd(__m128d __A, __mmask8 __U, __m128d __B, __m128d __C) {
+  return (__m128d)__builtin_ia32_selectpd_128(
+      (__mmask8)__U, (__v2df)_mm_fnmsub_pd(__A, __B, __C), (__v2df)__A);
+}
+
+static __inline__ __m128d __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_mask3_fnmsub_pd(__m128d __A, __m128d __B, __m128d __C, __mmask8 __U) {
+  return (__m128d)__builtin_ia32_selectpd_128(
+      (__mmask8)__U, (__v2df)_mm_fnmsub_pd(__A, __B, __C), (__v2df)__C);
+}
+
 static __inline__ __m128d __DEFAULT_FN_ATTRS128_CONSTEXPR
 _mm_maskz_fnmsub_pd(__mmask8 __U, __m128d __A, __m128d __B, __m128d __C) {
   return (__m128d)__builtin_ia32_selectpd_128(
@@ -977,6 +1001,12 @@ _mm256_mask_fmsub_pd(__m256d __A, __mmask8 __U, __m256d __B, __m256d __C) {
       (__mmask8)__U, (__v4df)_mm256_fmsub_pd(__A, __B, __C), (__v4df)__A);
 }
 
+static __inline__ __m256d __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_mask3_fmsub_pd(__m256d __A, __m256d __B, __m256d __C, __mmask8 __U) {
+  return (__m256d)__builtin_ia32_selectpd_256(
+      (__mmask8)__U, (__v4df)_mm256_fmsub_pd(__A, __B, __C), (__v4df)__C);
+}
+
 static __inline__ __m256d __DEFAULT_FN_ATTRS256_CONSTEXPR
 _mm256_maskz_fmsub_pd(__mmask8 __U, __m256d __A, __m256d __B, __m256d __C) {
   return (__m256d)__builtin_ia32_selectpd_256(
@@ -984,6 +1014,12 @@ _mm256_maskz_fmsub_pd(__mmask8 __U, __m256d __A, __m256d __B, __m256d __C) {
       (__v4df)_mm256_setzero_pd());
 }
 
+static __inline__ __m256d __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_mask_fnmadd_pd(__m256d __A, __mmask8 __U, __m256d __B, __m256d __C) {
+  return (__m256d)__builtin_ia32_selectpd_256(
+      (__mmask8)__U, (__v4df)_mm256_fnmadd_pd(__A, __B, __C), (__v4df)__A);
+}
+
 static __inline__ __m256d __DEFAULT_FN_ATTRS256_CONSTEXPR
 _mm256_mask3_fnmadd_pd(__m256d __A, __m256d __B, __m256d __C, __mmask8 __U) {
   return (__m256d)__builtin_ia32_selectpd_256(
@@ -997,6 +1033,18 @@ _mm256_maskz_fnmadd_pd(__mmask8 __U, __m256d __A, __m256d __B, __m256d __C) {
       (__v4df)_mm256_setzero_pd());
 }
 
+static __inline__ __m256d __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_mask_fnmsub_pd(__m256d __A, __mmask8 __U, __m256d __B, __m256d __C) {
+  return (__m256d)__builtin_ia32_selectpd_256(
+      (__mmask8)__U, (__v4df)_mm256_fnmsub_pd(__A, __B, __C), (__v4df)__A);
+}
+
+static __inline__ __m256d __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_mask3_fnmsub_pd(__m256d __A, __m256d __B, __m256d __C, __mmask8 __U) {
+  return (__m256d)__builtin_ia32_selectpd_256(
+      (__mmask8)__U, (__v4df)_mm256_fnmsub_pd(__A, __B, __C), (__v4df)__C);
+}
+
 static __inline__ __m256d __DEFAULT_FN_ATTRS256_CONSTEXPR
 _mm256_maskz_fnmsub_pd(__mmask8 __U, __m256d __A, __m256d __B, __m256d __C) {
   return (__m256d)__builtin_ia32_selectpd_256(
@@ -1029,6 +1077,12 @@ _mm_mask_fmsub_ps(__m128 __A, __mmask8 __U, __m128 __B, __m128 __C) {
       (__mmask8)__U, (__v4sf)_mm_fmsub_ps(__A, __B, __C), (__v4sf)__A);
 }
 
+static __inline__ __m128 __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_mask3_fmsub_ps(__m128 __A, __m128 __B, __m128 __C, __mmask8 __U) {
+  return (__m128)__builtin_ia32_selectps_128(
+      (__mmask8)__U, (__v4sf)_mm_fmsub_ps(__A, __B, __C), (__v4sf)__C);
+}
+
 static __inline__ __m128 __DEFAULT_FN_ATTRS128_CONSTEXPR
 _mm_maskz_fmsub_ps(__mmask8 __U, __m128 __A, __m128 __B, __m128 __C) {
   return (__m128)__builtin_ia32_selectps_128(
@@ -1036,6 +1090,12 @@ _mm_maskz_fmsub_ps(__mmask8 __U, __m128 __A, __m128 __B, __m128 __C) {
       (__v4sf)_mm_setzero_ps());
 }
 
+static __inline__ __m128 __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_mask_fnmadd_ps(__m128 __A, __mmask8 __U, __m128 __B, __m128 __C) {
+  return (__m128)__builtin_ia32_selectps_128(
+      (__mmask8)__U, (__v4sf)_mm_fnmadd_ps(__A, __B, __C), (__v4sf)__A);
+}
+
 static __inline__ __m128 __DEFAULT_FN_ATTRS128_CONSTEXPR
 _mm_mask3_fnmadd_ps(__m128 __A, __m128 __B, __m128 __C, __mmask8 __U) {
   return (__m128)__builtin_ia32_selectps_128(
@@ -1049,6 +1109,18 @@ _mm_maskz_fnmadd_ps(__mmask8 __U, __m128 __A, __m128 __B, __m128 __C) {
       (__v4sf)_mm_setzero_ps());
 }
 
+static __inline__ __m128 __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_mask_fnmsub_ps(__m128 __A, __mmask8 __U, __m128 __B, __m128 __C) {
+  return (__m128)__builtin_ia32_selectps_128(
+      (__mmask8)__U, (__v4sf)_mm_fnmsub_ps(__A, __B, __C), (__v4sf)__A);
+}
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_mask3_fnmsub_ps(__m128 __A, __m128 __B, __m128 __C, __mmask8 __U) {
+  return (__m128)__builtin_ia32_selectps_128(
+      (__mmask8)__U, (__v4sf)_mm_fnmsub_ps(__A, __B, __C), (__v4sf)__C);
+}
+
 static __inline__ __m128 __DEFAULT_FN_ATTRS128_CONSTEXPR
 _mm_maskz_fnmsub_ps(__mmask8 __U, __m128 __A, __m128 __B, __m128 __C) {
   return (__m128)__builtin_ia32_selectps_128(
@@ -1081,6 +1153,12 @@ _mm256_mask_fmsub_ps(__m256 __A, __mmask8 __U, __m256 __B, __m256 __C) {
       (__mmask8)__U, (__v8sf)_mm256_fmsub_ps(__A, __B, __C), (__v8sf)__A);
 }
 
+static __inline__ __m256 __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_mask3_fmsub_ps(__m256 __A, __m256 __B, __m256 __C, __mmask8 __U) {
+  return (__m256)__builtin_ia32_selectps_256(
+      (__mmask8)__U, (__v8sf)_mm256_fmsub_ps(__A, __B, __C), (__v8sf)__C);
+}
+
 static __inline__ __m256 __DEFAULT_FN_ATTRS256_CONSTEXPR
 _mm256_maskz_fmsub_ps(__mmask8 __U, __m256 __A, __m256 __B, __m256 __C) {
   return (__m256)__builtin_ia32_selectps_256(
@@ -1088,6 +1166,12 @@ _mm256_maskz_fmsub_ps(__mmask8 __U, __m256 __A, __m256 __B, __m256 __C) {
       (__v8sf)_mm256_setzero_ps());
 }
 
+static __inline__ __m256 __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_mask_fnmadd_ps(__m256 __A, __mmask8 __U, __m256 __B, __m256 __C) {
+  return (__m256)__builtin_ia32_selectps_256(
+      (__mmask8)__U, (__v8sf)_mm256_fnmadd_ps(__A, __B, __C), (__v8sf)__A);
+}
+
 static __inline__ __m256 __DEFAULT_FN_ATTRS256_CONSTEXPR
 _mm256_mask3_fnmadd_ps(__m256 __A, __m256 __B, __m256 __C, __mmask8 __U) {
   return (__m256)__builtin_ia32_selectps_256(
@@ -1101,6 +1185,18 @@ _mm256_maskz_fnmadd_ps(__mmask8 __U, __m256 __A, __m256 __B, __m256 __C) {
       (__v8sf)_mm256_setzero_ps());
 }
 
+static __inline__ __m256 __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_mask_fnmsub_ps(__m256 __A, __mmask8 __U, __m256 __B, __m256 __C) {
+  return (__m256)__builtin_ia32_selectps_256(
+      (__mmask8)__U, (__v8sf)_mm256_fnmsub_ps(__A, __B, __C), (__v8sf)__A);
+}
+
+static __inline__ __m256 __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_mask3_fnmsub_ps(__m256 __A, __m256 __B, __m256 __C, __mmask8 __U) {
+  return (__m256)__builtin_ia32_selectps_256(
+      (__mmask8)__U, (__v8sf)_mm256_fnmsub_ps(__A, __B, __C), (__v8sf)__C);
+}
+
 static __inline__ __m256 __DEFAULT_FN_ATTRS256_CONSTEXPR
 _mm256_maskz_fnmsub_ps(__mmask8 __U, __m256 __A, __m256 __B, __m256 __C) {
   return (__m256)__builtin_ia32_selectps_256(
@@ -1308,30 +1404,6 @@ _mm256_maskz_fmsubadd_ps(__mmask8 __U, __m256 __A, __m256 __B, __m256 __C)
                     (__v8sf)_mm256_setzero_ps());
 }
 
-static __inline__ __m128d __DEFAULT_FN_ATTRS128_CONSTEXPR
-_mm_mask3_fmsub_pd(__m128d __A, __m128d __B, __m128d __C, __mmask8 __U) {
-  return (__m128d)__builtin_ia32_selectpd_128(
-      (__mmask8)__U, (__v2df)_mm_fmsub_pd(__A, __B, __C), (__v2df)__C);
-}
-
-static __inline__ __m256d __DEFAULT_FN_ATTRS256_CONSTEXPR
-_mm256_mask3_fmsub_pd(__m256d __A, __m256d __B, __m256d __C, __mmask8 __U) {
-  return (__m256d)__builtin_ia32_selectpd_256(
-      (__mmask8)__U, (__v4df)_mm256_fmsub_pd(__A, __B, __C), (__v4df)__C);
-}
-
-static __inline__ __m128 __DEFAULT_FN_ATTRS128_CONSTEXPR
-_mm_mask3_fmsub_ps(__m128 __A, __m128 __B, __m128 __C, __mmask8 __U) {
-  return (__m128)__builtin_ia32_selectps_128(
-      (__mmask8)__U, (__v4sf)_mm_fmsub_ps(__A, __B, __C), (__v4sf)__C);
-}
-
-static __inline__ __m256 __DEFAULT_FN_ATTRS256_CONSTEXPR
-_mm256_mask3_fmsub_ps(__m256 __A, __m256 __B, __m256 __C, __mmask8 __U) {
-  return (__m256)__builtin_ia32_selectps_256(
-      (__mmask8)__U, (__v8sf)_mm256_fmsub_ps(__A, __B, __C), (__v8sf)__C);
-}
-
 static __inline__ __m128d __DEFAULT_FN_ATTRS128
 _mm_mask3_fmsubadd_pd(__m128d __A, __m128d __B, __m128d __C, __mmask8 __U)
 {
@@ -1372,78 +1444,6 @@ _mm256_mask3_fmsubadd_ps(__m256 __A, __m256 __B, __m256 __C, __mmask8 __U)
                     (__v8sf) __C);
 }
 
-static __inline__ __m128d __DEFAULT_FN_ATTRS128_CONSTEXPR
-_mm_mask_fnmadd_pd(__m128d __A, __mmask8 __U, __m128d __B, __m128d __C) {
-  return (__m128d)__builtin_ia32_selectpd_128(
-      (__mmask8)__U, (__v2df)_mm_fnmadd_pd(__A, __B, __C), (__v2df)__A);
-}
-
-static __inline__ __m256d __DEFAULT_FN_ATTRS256_CONSTEXPR
-_mm256_mask_fnmadd_pd(__m256d __A, __mmask8 __U, __m256d __B, __m256d __C) {
-  return (__m256d)__builtin_ia32_selectpd_256(
-      (__mmask8)__U, (__v4df)_mm256_fnmadd_pd(__A, __B, __C), (__v4df)__A);
-}
-
-static __inline__ __m128 __DEFAULT_FN_ATTRS128_CONSTEXPR
-_mm_mask_fnmadd_ps(__m128 __A, __mmask8 __U, __m128 __B, __m128 __C) {
-  return (__m128)__builtin_ia32_selectps_128(
-      (__mmask8)__U, (__v4sf)_mm_fnmadd_ps(__A, __B, __C), (__v4sf)__A);
-}
-
-static __inline__ __m256 __DEFAULT_FN_ATTRS256_CONSTEXPR
-_mm256_mask_fnmadd_ps(__m256 __A, __mmask8 __U, __m256 __B, __m256 __C) {
-  return (__m256)__builtin_ia32_selectps_256(
-      (__mmask8)__U, (__v8sf)_mm256_fnmadd_ps(__A, __B, __C), (__v8sf)__A);
-}
-
-static __inline__ __m128d __DEFAULT_FN_ATTRS128_CONSTEXPR
-_mm_mask_fnmsub_pd(__m128d __A, __mmask8 __U, __m128d __B, __m128d __C) {
-  return (__m128d)__builtin_ia32_selectpd_128(
-      (__mmask8)__U, (__v2df)_mm_fnmsub_pd(__A, __B, __C), (__v2df)__A);
-}
-
-static __inline__ __m128d __DEFAULT_FN_ATTRS128_CONSTEXPR
-_mm_mask3_fnmsub_pd(__m128d __A, __m128d __B, __m128d __C, __mmask8 __U) {
-  return (__m128d)__builtin_ia32_selectpd_128(
-      (__mmask8)__U, (__v2df)_mm_fnmsub_pd(__A, __B, __C), (__v2df)__C);
-}
-
-static __inline__ __m256d __DEFAULT_FN_ATTRS256_CONSTEXPR
-_mm256_mask_fnmsub_pd(__m256d __A, __mmask8 __U, __m256d __B, __m256d __C) {
-  return (__m256d)__builtin_ia32_selectpd_256(
-      (__mmask8)__U, (__v4df)_mm256_fnmsub_pd(__A, __B, __C), (__v4df)__A);
-}
-
-static __inline__ __m256d __DEFAULT_FN_ATTRS256_CONSTEXPR
-_mm256_mask3_fnmsub_pd(__m256d __A, __m256d __B, __m256d __C, __mmask8 __U) {
-  return (__m256d)__builtin_ia32_selectpd_256(
-      (__mmask8)__U, (__v4df)_mm256_fnmsub_pd(__A, __B, __C), (__v4df)__C);
-}
-
-static __inline__ __m128 __DEFAULT_FN_ATTRS128_CONSTEXPR
-_mm_mask_fnmsub_ps(__m128 __A, __mmask8 __U, __m128 __B, __m128 __C) {
-  return (__m128)__builtin_ia32_selectps_128(
-      (__mmask8)__U, (__v4sf)_mm_fnmsub_ps(__A, __B, __C), (__v4sf)__A);
-}
-
-static __inline__ __m128 __DEFAULT_FN_ATTRS128_CONSTEXPR
-_mm_mask3_fnmsub_ps(__m128 __A, __m128 __B, __m128 __C, __mmask8 __U) {
-  return (__m128)__builtin_ia32_selectps_128(
-      (__mmask8)__U, (__v4sf)_mm_fnmsub_ps(__A, __B, __C), (__v4sf)__C);
-}
-
-static __inline__ __m256 __DEFAULT_FN_ATTRS256_CONSTEXPR
-_mm256_mask_fnmsub_ps(__m256 __A, __mmask8 __U, __m256 __B, __m256 __C) {
-  return (__m256)__builtin_ia32_selectps_256(
-      (__mmask8)__U, (__v8sf)_mm256_fnmsub_ps(__A, __B, __C), (__v8sf)__A);
-}
-
-static __inline__ __m256 __DEFAULT_FN_ATTRS256_CONSTEXPR
-_mm256_mask3_fnmsub_ps(__m256 __A, __m256 __B, __m256 __C, __mmask8 __U) {
-  return (__m256)__builtin_ia32_selectps_256(
-      (__mmask8)__U, (__v8sf)_mm256_fnmsub_ps(__A, __B, __C), (__v8sf)__C);
-}
-
 static __inline__ __m128d __DEFAULT_FN_ATTRS128
 _mm_mask_add_pd(__m128d __W, __mmask8 __U, __m128d __A, __m128d __B) {
   return (__m128d)__builtin_ia32_selectpd_128((__mmask8)__U,
diff --git a/clang/test/CodeGen/X86/avx512vl-builtins.c b/clang/test/CodeGen/X86/avx512vl-builtins.c
index 533896938a205..39250ccefd09d 100644
--- a/clang/test/CodeGen/X86/avx512vl-builtins.c
+++ b/clang/test/CodeGen/X86/avx512vl-builtins.c
@@ -2853,6 +2853,26 @@ __m128d test_mm_mask_fmadd_pd(__m128d __A, __mmask8 __U, __m128d __B, __m128d __
 TEST_CONSTEXPR(match_m128d(_mm_mask_fmadd_pd((__m128d){1.0, 2.0}, (__mmask8)0b10, (__m128d){3.0, 4.0}, (__m128d){5.0, 6.0}), 1.0, 14.0));
 TEST_CONSTEXPR(match_m128d(_mm_mask_fmadd_pd((__m128d){1.0, 2.0}, (__mmask8)0b01, (__m128d){3.0, 4.0}, (__m128d){5.0, 6.0}), 8.0, 2.0));
 
+__m128d test_mm_mask3_fmadd_pd(__m128d __A, __m128d __B, __m128d __C, __mmask8 __U) {
+  // CHECK-LABEL: test_mm_mask3_fmadd_pd
+  // CHECK: call {{.*}}<2 x double> @llvm.fma.v2f64(<2 x double> %{{.*}}, <2 x double> %{{.*}}, <2 x double> %{{.*}})
+  // CHECK: shufflevector <8 x i1> %{{.*}}, <8 x i1> %{{.*}}, <2 x i32> <i32 0, i32 1>
+  // CHECK: select <2 x i1> %{{.*}}, <2 x double> %{{.*}}, <2 x double> %{{.*}}
+  return _mm_mask3_fmadd_pd(__A, __B, __C, __U);
+}
+TEST_CONSTEXPR(match_m128d(_mm_mask3_fmadd_pd((__m128d){1.0, 2.0}, (__m128d){3.0, 4.0}, (__m128d){5.0, 6.0}, (__mmask8)0b10), 5.0, 14.0));
+TEST_CONSTEXPR(match_m128d(_mm_mask3_fmadd_pd((__m128d){1.0, 2.0}, (__m128d){3.0, 4.0}, (__m128d){5.0, 6.0}, (__mmask8)0b01), 8.0, 6.0));
+
+__m128d test_mm_maskz_fmadd_pd(__mmask8 __U, __m128d __A, __m128d __B, __m128d __C) {
+  // CHECK-LABEL: test_mm_maskz_fmadd_pd
+  // CHECK: call {{.*}}<2 x double> @llvm.fma.v2f64(<2 x double> %{{.*}}, <2 x double> %{{.*}}, <2 x double> %{{.*}})
+  // CHECK: shufflevector <8 x i1> %{{.*}}, <8 x i1> %{{.*}}, <2 x i32> <i32 0, i32 1>
+  // CHECK: select <2 x i1> %{{.*}}, <2 x double> %{{.*}}, <2 x double> %{{.*}}
+  return _mm_maskz_fmadd_pd(__U, __A, __B, __C);
+}
+TEST_CONSTEXPR(match_m128d(_mm_maskz_fmadd_pd((__mmask8)0b10, (__m128d){1.0, 2.0}, (__m128d){3.0, 4.0}, (__m128d){5.0, 6.0}), 0.0, 14.0));
+TEST_CONSTEXPR(match_m128d(_mm_maskz_fmadd_pd((__mmask8)0b01, (__m128d){1.0, 2.0}, (__m128d){3.0, 4.0}, (__m128d){5.0, 6.0}), 8.0, 0.0));
+
 __m128d test_mm_mask_fmsub_pd(__m128d __A, __mmask8 __U, __m128d __B, __m128d __C) {
   // CHECK-LABEL: test_mm_mask_fmsub_pd
   // CHECK: fneg <2 x double> %{{.*}}
@@ -2864,47 +2884,49 @@ __m128d test_mm_mask_fmsub_pd(__m128d __A, __mmask8 __U, __m128d __B, __m128d __
 TEST_CONSTEXPR(match_m128d(_mm_mask_fmsub_pd((__m128d){1.0, 2.0}, (__mmask8)0b10, (__m128d){3.0, 4.0}, (__m128d){5.0, 6.0}), 1.0, 2.0));
 TEST_CONSTEXPR(match_m128d(_mm_mask_fmsub_pd((__m128d){1.0, 2.0}, (__mmask8)0b01, (__m128d){3.0, 4.0}, (__m128d){5.0, 6.0}), -2.0, 2.0));
 
-__m128d test_mm_mask3_fmadd_pd(__m128d __A, __m128d __B, __m128d __C, __mmask8 __U) {
-  // CHECK-LABEL: test_mm_mask3_fmadd_pd
+__m128d test_mm_mask3_fmsub_pd(__m128d __A, __m128d __B, __m128d __C, __mmask8 __U) {
+  // CHECK-LABEL: test_mm_mask3_fmsub_pd
+  // CHECK: fneg <2 x double> %{{.*}}
   // CHECK: call {{.*}}<2 x double> @llvm.fma.v2f64(<2 x double> %{{.*}}, <2 x double> %{{.*}}, <2 x double> %{{.*}})
   // CHECK: shufflevector <8 x i1> %{{.*}}, <8 x i1> %{{.*}}, <2 x i32> <i32 0, i32 1>
   // CHECK: select <2 x i1> %{{.*}}, <2 x double> %{{.*}}, <2 x double> %{{.*}}
-  return _mm_mask3_fmadd_pd(__A, __B, __C, __U);
+  return _mm_mask3_fmsub_pd(__A, __B, __C, __U);
 }
-TEST_CONSTEXPR(match_m128d(_mm_mask3_fmadd_pd((__m128d){1.0, 2.0}, (__m128d){3.0, 4.0}, (__m128d){5.0, 6.0}, (__mmask8)0b10), 5.0, 14.0));
-TEST_CONSTEXPR(match_m128d(_mm_mask3_fmadd_pd((__m128d){1.0, 2.0}, (__m128d){3.0, 4.0}, (__m128d){5.0, 6.0}, (__mmask8)0b01), 8.0, 6.0));
+TEST_CONSTEXPR(match_m128d(_mm_mask3_fmsub_pd((__m128d){1.0, 2.0}, (__m128d){3.0, 4.0}, (__m128d){5.0, 6.0}, (__mmask8)0b10), 5.0, 2.0));
+TEST_CONSTEXPR(match_m128d(_mm_mask3_fmsub_pd((__m128d){1.0, 2.0}, (__m128d){3.0, 4.0}, (__m128d){5.0, 6.0}, (__mmask8)0b01), -2.0, 6.0));
 
-__m128d test_mm_mask3_fnmadd_pd(__m128d __A, __m128d __B, __m128d __C, __mmask8 __U) {
-  // CHECK-LABEL: test_mm_mask3_fnmadd_pd
+__m128d test_mm_maskz_fmsub_pd(__mmask8 __U, __m128d __A, __m128d __B, __m128d __C) {
+  // CHECK-LABEL: test_mm_maskz_fmsub_pd
   // CHECK: fneg <2 x double> %{{.*}}
   // CHECK: call {{.*}}<2 x double> @llvm.fma.v2f64(<2 x double> %{{.*}}, <2 x double> %{{.*}}, <2 x double> %{{.*}})
   // CHECK: shufflevector <8 x i1> %{{.*}}, <8 x i1> %{{.*}}, <2 x i32> <i32 0, i32 1>
   // CHECK: select <2 x i1> %{{.*}}, <2 x double> %{{.*}}, <2 x double> %{{.*}}
-  return _mm_mask3_fnmadd_pd(__A, __B, __C, __U);
+  return _mm_maskz_fmsub_pd(__U, __A, __B, __C);
 }
-TEST_CONSTEXPR(match_m128d(_mm_mask3_fnmadd_pd((__m128d){1.0, 2.0}, (__m128d){3.0, 4.0}, (__m128d){5.0, 6.0}, (__mmask8)0b10), 5.0, -2.0));
-TEST_CONSTEXPR(match_m128d(_mm_mask3_fnmadd_pd((__m128d){1.0, 2.0}, (__m128d){3.0, 4.0}, (__m128d){5.0, 6.0}, (__mmask8)0b01), 2.0, 6.0));
+TEST_CONSTEXPR(match_m128d(_mm_maskz_fmsub_pd((__mmask8)0b10, (__m128d){1.0, 2.0}, (__m128d){3.0, 4.0}, (__m128d){5.0, 6.0}), 0.0, 2.0));
+TEST_CONSTEXPR(match_m128d(_mm_maskz_fmsub_pd((__mmask8)0b01, (__m128d){1.0, 2.0}, (__m128d){3.0, 4.0}, (__m128d){5.0, 6.0}), -2.0, 0.0));
 
-__m128d test_mm_maskz_fmadd_pd(__mmask8 __U, __m128d __A, __m128d __B, __m128d __C) {
-  // CHECK-LABEL: test_mm_maskz_fmadd_pd
+__m128d test_mm_mask_fnmadd_pd(__m128d __A, __mmask8 __U, __m128d __B, __m128d __C) {
+  // CHECK-LABEL: test_mm_mask_fnmadd_pd
+  // CHECK: fneg <2 x double> %{{.*}}
   // CHECK: call {{.*}}<2 x double> @llvm.fma.v2f64(<2 x double> %{{.*}}, <2 x double> %{{.*}}, <2 x double> %{{.*}})
   // CHECK: shufflevector <8 x i1> %{{.*}}, <8 x i1> %{{.*}}, <2 x i32> <i32 0, i32 1>
   // CHECK: select <2 x i1> %{{.*}}, <2 x double> %{{.*}}, <2 x double> %{{.*}}
-  return _mm_maskz_fmadd_pd(__U, __A, __B, __C);
+  return _mm_mask_fnmadd_pd(__A, __U, __B, __C);
 }
-TEST_CONSTEXPR(match_m128d(_mm_maskz_fmadd_pd((__mmask8)0b10, (__m128d){1.0, 2.0}, (__m128d){3.0, 4.0}, (__m128d){5.0, 6.0}), 0.0, 14.0));
-TEST_CONSTEXPR(match_m128d(_mm_maskz_fmadd_pd((__mmask8)0b01, (__m128d){1.0, 2.0}, (__m128d){3.0, 4.0}, (__m128d){5.0, 6.0}), 8.0, 0.0));
+TEST_CONSTEXPR(match_m128d(_mm_mask_fnmadd_pd((__m128d){1.0, 2.0}, (__mmask8)0b10, (__m128d){3.0, 4.0}, (__m128d){5.0, 6.0}), 1.0, -2.0));
+TEST_CONSTEXPR(match_m128d(_mm_mask_fnmadd_pd((__m128d){1.0, 2.0}, (__mmask8)0b01, (__m128d){3.0, 4.0}, (__m128d){5.0, 6.0}), 2.0, 2.0));
 
-__m128d test_mm_maskz_fmsub_pd(__mmask8 __U, __m128d __A, __m128d __B, __m128d __C) {
-  // CHECK-LABEL: test_mm_maskz_fmsub_pd
+__m128d test_mm_mask3_fnmadd_pd(__m128d __A, __m128d __B, __m128d __C, __mmask8 __U) {
+  // CHECK-LABEL: test_mm_mask3_fnmadd_pd
   // CHECK: fneg <2 x double> %{{.*}}
   // CHECK: call {{.*}}<2 x double> @llvm.fma.v2f64(<2 x double> %{{.*}}, <2 x double> %{{.*}}, <2 x double> %{{.*}})
   // CHECK: shufflevector <8 x i1> %{{.*}}, <8 x i1> %{{.*}}, <2 x i32> <i32 0, i32 1>
   // CHECK: select <2 x i1> %{{.*}}, <2 x double> %{{.*}}, <2 x double> %{{.*}}
-  return _mm_maskz_fmsub_pd(__U, __A, __B, __C);
+  return _mm_mask3_fnmadd_pd(__A, __B, __C, __U);
 }
-TEST_CONSTEXPR(match_m128d(_mm_maskz_fmsub_pd((__mmask8)0b10, (__m128d){1.0, 2.0}, (__m128d){3.0, 4.0}, (__m128d){5.0, 6.0}), 0.0, 2.0));
-TEST_CONSTEXPR(match_m128d(_mm_maskz_fmsub_pd((__mmask8)0b01, (__m...
[truncated]

@RKSimon RKSimon self-requested a review September 4, 2025 13:07
RKSimon
RKSimon approved these changes Sep 4, 2025
View reviewed changes
Copy link
Collaborator

@RKSimon RKSimon left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM - cheers

@RKSimon RKSimon merged commit faca9dd into llvm:main Sep 4, 2025
13 checks passed
@ckoparkar ckoparkar deleted the ckoparkar/fmaintrins branch September 4, 2025 13:26
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@RKSimon RKSimon RKSimon approved these changes

Assignees
No one assigned
Labels
backend:X86 clang:headers Headers provided by Clang, e.g. for intrinsics clang Clang issues not falling into any other category
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@ckoparkar @llvmbot @RKSimon