From: Michael Zuckerman <Michael.zuckerman@intel.com>
Date: Mon, 2 May 2016 14:02:01 +0000 (+0000)
Subject: [CLANG][AVX512][BUILTIN]movap{d|s}{128|256|512}
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=a10fe99ac8f5f63a9ec2227c9c20b6a28de9785e;p=clang

[CLANG][AVX512][BUILTIN]movap{d|s}{128|256|512}

Differential Revision: http://reviews.llvm.org/D17818


git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@268230 91177308-0d34-0410-b5e6-96231b3b80d8
---

diff --git a/include/clang/Basic/BuiltinsX86.def b/include/clang/Basic/BuiltinsX86.def
index 11881eff65..e2192f8f05 100644
--- a/include/clang/Basic/BuiltinsX86.def
+++ b/include/clang/Basic/BuiltinsX86.def
@@ -2242,6 +2242,12 @@ TARGET_BUILTIN(__builtin_ia32_expandloadsi512_mask, "V16ivC*V16iUs","","avx512f"
 TARGET_BUILTIN(__builtin_ia32_expandsf512_mask, "V16fV16fV16fUs","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_expandsi512_mask, "V16iV16iV16iUs","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_cvtps2pd512_mask, "V8dV8fV8dUcIi","","avx512f")
+TARGET_BUILTIN(__builtin_ia32_movapd512_mask, "V8dV8dV8dUc","","avx512f")
+TARGET_BUILTIN(__builtin_ia32_movaps512_mask, "V16fV16fV16fUs","","avx512f")
+TARGET_BUILTIN(__builtin_ia32_movapd128_mask, "V2dV2dV2dUc","","avx512vl")
+TARGET_BUILTIN(__builtin_ia32_movapd256_mask, "V4dV4dV4dUc","","avx512vl")
+TARGET_BUILTIN(__builtin_ia32_movaps128_mask, "V4fV4fV4fUc","","avx512vl")
+TARGET_BUILTIN(__builtin_ia32_movaps256_mask, "V8fV8fV8fUc","","avx512vl")
 
 #undef BUILTIN
 #undef TARGET_BUILTIN
diff --git a/lib/Headers/avx512fintrin.h b/lib/Headers/avx512fintrin.h
index 2a1a81c3e5..3970bdbbdd 100644
--- a/lib/Headers/avx512fintrin.h
+++ b/lib/Headers/avx512fintrin.h
@@ -7931,6 +7931,41 @@ _mm512_maskz_cvtps_pd (__mmask8 __U, __m256 __A)
                 _MM_FROUND_CUR_DIRECTION);
 }
 
+static __inline__ __m512d __DEFAULT_FN_ATTRS
+_mm512_mask_mov_pd (__m512d __W, __mmask8 __U, __m512d __A)
+{
+  return (__m512d) __builtin_ia32_movapd512_mask ((__v8df) __A,
+              (__v8df) __W,
+              (__mmask8) __U);
+}
+
+static __inline__ __m512d __DEFAULT_FN_ATTRS
+_mm512_maskz_mov_pd (__mmask8 __U, __m512d __A)
+{
+  return (__m512d) __builtin_ia32_movapd512_mask ((__v8df) __A,
+              (__v8df)
+              _mm512_setzero_pd (),
+              (__mmask8) __U);
+}
+
+static __inline__ __m512 __DEFAULT_FN_ATTRS
+_mm512_mask_mov_ps (__m512 __W, __mmask16 __U, __m512 __A)
+{
+  return (__m512) __builtin_ia32_movaps512_mask ((__v16sf) __A,
+             (__v16sf) __W,
+             (__mmask16) __U);
+}
+
+static __inline__ __m512 __DEFAULT_FN_ATTRS
+_mm512_maskz_mov_ps (__mmask16 __U, __m512 __A)
+{
+  return (__m512) __builtin_ia32_movaps512_mask ((__v16sf) __A,
+             (__v16sf)
+             _mm512_setzero_ps (),
+             (__mmask16) __U);
+}
+
+
 #undef __DEFAULT_FN_ATTRS
 
 #endif // __AVX512FINTRIN_H
diff --git a/lib/Headers/avx512vlintrin.h b/lib/Headers/avx512vlintrin.h
index 3aacc68b0a..37a4fa11a2 100644
--- a/lib/Headers/avx512vlintrin.h
+++ b/lib/Headers/avx512vlintrin.h
@@ -9384,6 +9384,76 @@ __builtin_ia32_pshufd256_mask((__v8si) (__A), (__I),\
               (__mmask8) (__U));\
 })
 
+
+static __inline__ __m128d __DEFAULT_FN_ATTRS
+_mm_mask_mov_pd (__m128d __W, __mmask8 __U, __m128d __A)
+{
+  return (__m128d) __builtin_ia32_movapd128_mask ((__v2df) __A,
+              (__v2df) __W,
+              (__mmask8) __U);
+}
+
+static __inline__ __m128d __DEFAULT_FN_ATTRS
+_mm_maskz_mov_pd (__mmask8 __U, __m128d __A)
+{
+  return (__m128d) __builtin_ia32_movapd128_mask ((__v2df) __A,
+              (__v2df)
+              _mm_setzero_pd (),
+              (__mmask8) __U);
+}
+
+static __inline__ __m256d __DEFAULT_FN_ATTRS
+_mm256_mask_mov_pd (__m256d __W, __mmask8 __U, __m256d __A)
+{
+  return (__m256d) __builtin_ia32_movapd256_mask ((__v4df) __A,
+              (__v4df) __W,
+              (__mmask8) __U);
+}
+
+static __inline__ __m256d __DEFAULT_FN_ATTRS
+_mm256_maskz_mov_pd (__mmask8 __U, __m256d __A)
+{
+  return (__m256d) __builtin_ia32_movapd256_mask ((__v4df) __A,
+              (__v4df)
+              _mm256_setzero_pd (),
+              (__mmask8) __U);
+}
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_mask_mov_ps (__m128 __W, __mmask8 __U, __m128 __A)
+{
+  return (__m128) __builtin_ia32_movaps128_mask ((__v4sf) __A,
+             (__v4sf) __W,
+             (__mmask8) __U);
+}
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_maskz_mov_ps (__mmask8 __U, __m128 __A)
+{
+  return (__m128) __builtin_ia32_movaps128_mask ((__v4sf) __A,
+             (__v4sf)
+             _mm_setzero_ps (),
+             (__mmask8) __U);
+}
+
+static __inline__ __m256 __DEFAULT_FN_ATTRS
+_mm256_mask_mov_ps (__m256 __W, __mmask8 __U, __m256 __A)
+{
+  return (__m256) __builtin_ia32_movaps256_mask ((__v8sf) __A,
+             (__v8sf) __W,
+             (__mmask8) __U);
+}
+
+static __inline__ __m256 __DEFAULT_FN_ATTRS
+_mm256_maskz_mov_ps (__mmask8 __U, __m256 __A)
+{
+  return (__m256) __builtin_ia32_movaps256_mask ((__v8sf) __A,
+             (__v8sf)
+             _mm256_setzero_ps (),
+             (__mmask8) __U);
+}
+
+
 #undef __DEFAULT_FN_ATTRS
 #undef __DEFAULT_FN_ATTRS_BOTH
 
diff --git a/test/CodeGen/avx512f-builtins.c b/test/CodeGen/avx512f-builtins.c
index da1dd526b8..70bc645c89 100644
--- a/test/CodeGen/avx512f-builtins.c
+++ b/test/CodeGen/avx512f-builtins.c
@@ -5505,3 +5505,26 @@ __m512d test_mm512_maskz_cvtps_pd(__mmask8 __U, __m256 __A) {
   // CHECK: @llvm.x86.avx512.mask.cvtps2pd.512
   return _mm512_maskz_cvtps_pd(__U, __A); 
 }
+__m512d test_mm512_mask_mov_pd(__m512d __W, __mmask8 __U, __m512d __A) {
+  // CHECK-LABEL: @test_mm512_mask_mov_pd
+  // CHECK: @llvm.x86.avx512.mask.mova.pd.512
+  return _mm512_mask_mov_pd(__W, __U, __A); 
+}
+
+__m512d test_mm512_maskz_mov_pd(__mmask8 __U, __m512d __A) {
+  // CHECK-LABEL: @test_mm512_maskz_mov_pd
+  // CHECK: @llvm.x86.avx512.mask.mova.pd.512
+  return _mm512_maskz_mov_pd(__U, __A); 
+}
+
+__m512 test_mm512_mask_mov_ps(__m512 __W, __mmask16 __U, __m512 __A) {
+  // CHECK-LABEL: @test_mm512_mask_mov_ps
+  // CHECK: @llvm.x86.avx512.mask.mova.ps.512
+  return _mm512_mask_mov_ps(__W, __U, __A); 
+}
+
+__m512 test_mm512_maskz_mov_ps(__mmask16 __U, __m512 __A) {
+  // CHECK-LABEL: @test_mm512_maskz_mov_ps
+  // CHECK: @llvm.x86.avx512.mask.mova.ps.512
+  return _mm512_maskz_mov_ps(__U, __A); 
+}
diff --git a/test/CodeGen/avx512vl-builtins.c b/test/CodeGen/avx512vl-builtins.c
index 3ab3d94299..72e9ba2de6 100644
--- a/test/CodeGen/avx512vl-builtins.c
+++ b/test/CodeGen/avx512vl-builtins.c
@@ -6606,3 +6606,51 @@ __m256i test_mm256_maskz_shuffle_epi32(__mmask8 __U, __m256i __A) {
   return _mm256_maskz_shuffle_epi32(__U, __A, 2); 
 }
 
+__m128d test_mm_mask_mov_pd(__m128d __W, __mmask8 __U, __m128d __A) {
+  // CHECK-LABEL: @test_mm_mask_mov_pd
+  // CHECK: @llvm.x86.avx512.mask.mova.pd.128
+  return _mm_mask_mov_pd(__W, __U, __A); 
+}
+
+__m128d test_mm_maskz_mov_pd(__mmask8 __U, __m128d __A) {
+  // CHECK-LABEL: @test_mm_maskz_mov_pd
+  // CHECK: @llvm.x86.avx512.mask.mova.pd.128
+  return _mm_maskz_mov_pd(__U, __A); 
+}
+
+__m256d test_mm256_mask_mov_pd(__m256d __W, __mmask8 __U, __m256d __A) {
+  // CHECK-LABEL: @test_mm256_mask_mov_pd
+  // CHECK: @llvm.x86.avx512.mask.mova.pd.256
+  return _mm256_mask_mov_pd(__W, __U, __A); 
+}
+
+__m256d test_mm256_maskz_mov_pd(__mmask8 __U, __m256d __A) {
+  // CHECK-LABEL: @test_mm256_maskz_mov_pd
+  // CHECK: @llvm.x86.avx512.mask.mova.pd.256
+  return _mm256_maskz_mov_pd(__U, __A); 
+}
+
+__m128 test_mm_mask_mov_ps(__m128 __W, __mmask8 __U, __m128 __A) {
+  // CHECK-LABEL: @test_mm_mask_mov_ps
+  // CHECK: @llvm.x86.avx512.mask.mova.ps.128
+  return _mm_mask_mov_ps(__W, __U, __A); 
+}
+
+__m128 test_mm_maskz_mov_ps(__mmask8 __U, __m128 __A) {
+  // CHECK-LABEL: @test_mm_maskz_mov_ps
+  // CHECK: @llvm.x86.avx512.mask.mova.ps.128
+  return _mm_maskz_mov_ps(__U, __A); 
+}
+
+__m256 test_mm256_mask_mov_ps(__m256 __W, __mmask8 __U, __m256 __A) {
+  // CHECK-LABEL: @test_mm256_mask_mov_ps
+  // CHECK: @llvm.x86.avx512.mask.mova.ps.256
+  return _mm256_mask_mov_ps(__W, __U, __A); 
+}
+
+__m256 test_mm256_maskz_mov_ps(__mmask8 __U, __m256 __A) {
+  // CHECK-LABEL: @test_mm256_maskz_mov_ps
+  // CHECK: @llvm.x86.avx512.mask.mova.ps.256
+  return _mm256_maskz_mov_ps(__U, __A); 
+}
+