From: Michael Zuckerman <Michael.zuckerman@intel.com>
Date: Thu, 3 Mar 2016 09:26:01 +0000 (+0000)
Subject: [CLANG][AVX512][BUILTIN] movdqa{32|64}{load|store|}{128|256|512}
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=d60afae0c62247da4d0e2ada0e271ccd3db523b5;p=clang

[CLANG][AVX512][BUILTIN] movdqa{32|64}{load|store|}{128|256|512}

Differential Revision: http://reviews.llvm.org/D17812



git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@262598 91177308-0d34-0410-b5e6-96231b3b80d8
---

diff --git a/include/clang/Basic/BuiltinsX86.def b/include/clang/Basic/BuiltinsX86.def
index a872555abe..41554248ba 100644
--- a/include/clang/Basic/BuiltinsX86.def
+++ b/include/clang/Basic/BuiltinsX86.def
@@ -1697,6 +1697,19 @@ TARGET_BUILTIN(__builtin_ia32_psrlw128_mask, "V8sV8sV8sV8sUc","","avx512bw,avx51
 TARGET_BUILTIN(__builtin_ia32_psrlw256_mask, "V16sV16sV8sV16sUs","","avx512bw,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_psrlwi128_mask, "V8sV8sIiV8sUc","","avx512bw,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_psrlwi256_mask, "V16sV16sIiV16sUs","","avx512bw,avx512vl")
+TARGET_BUILTIN(__builtin_ia32_movdqa32load512_mask, "V16ivC*V16iUs","","avx512f")
+TARGET_BUILTIN(__builtin_ia32_movdqa32store512_mask, "vV16i*V16iUs","","avx512f")
+TARGET_BUILTIN(__builtin_ia32_movdqa64_512_mask, "V8LLiV8LLiV8LLiUc","","avx512f")
+TARGET_BUILTIN(__builtin_ia32_movdqa64load512_mask, "V8LLivC*V8LLiUc","","avx512f")
+TARGET_BUILTIN(__builtin_ia32_movdqa64store512_mask, "vV8LLi*V8LLiUc","","avx512f")
+TARGET_BUILTIN(__builtin_ia32_movdqa32store128_mask, "vV4i*V4iUc","","avx512f")
+TARGET_BUILTIN(__builtin_ia32_movdqa32store256_mask, "vV8i*V8iUc","","avx512f")
+TARGET_BUILTIN(__builtin_ia32_movdqa64_128_mask, "V2LLiV2LLiV2LLiUc","","avx512vl")
+TARGET_BUILTIN(__builtin_ia32_movdqa64_256_mask, "V4LLiV4LLiV4LLiUc","","avx512vl")
+TARGET_BUILTIN(__builtin_ia32_movdqa64load128_mask, "V2LLivC*V2LLiUc","","avx512vl")
+TARGET_BUILTIN(__builtin_ia32_movdqa64load256_mask, "V4LLivC*V4LLiUc","","avx512vl")
+TARGET_BUILTIN(__builtin_ia32_movdqa64store128_mask, "vV2LLi*V2LLiUc","","avx512f")
+TARGET_BUILTIN(__builtin_ia32_movdqa64store256_mask, "vV4LLi*V4LLiUc","","avx512f")
 
 #undef BUILTIN
 #undef TARGET_BUILTIN
diff --git a/lib/Headers/avx512fintrin.h b/lib/Headers/avx512fintrin.h
index 2452c479f0..81565c4a9d 100644
--- a/lib/Headers/avx512fintrin.h
+++ b/lib/Headers/avx512fintrin.h
@@ -3602,6 +3602,72 @@ __builtin_ia32_psrlqi512_mask ((__v8di)( __A),( __B),\
               (__mmask8)( __U));\
 })
 
+static __inline__ __m512i __DEFAULT_FN_ATTRS
+_mm512_mask_load_epi32 (__m512i __W, __mmask16 __U, void const *__P)
+{
+  return (__m512i) __builtin_ia32_movdqa32load512_mask ((const __v16si *) __P,
+              (__v16si) __W,
+              (__mmask16) __U);
+}
+
+static __inline__ __m512i __DEFAULT_FN_ATTRS
+_mm512_maskz_load_epi32 (__mmask16 __U, void const *__P)
+{
+  return (__m512i) __builtin_ia32_movdqa32load512_mask ((const __v16si *) __P,
+              (__v16si)
+              _mm512_setzero_si512 (),
+              (__mmask16) __U);
+}
+
+static __inline__ void __DEFAULT_FN_ATTRS
+_mm512_mask_store_epi32 (void *__P, __mmask16 __U, __m512i __A)
+{
+  __builtin_ia32_movdqa32store512_mask ((__v16si *) __P, (__v16si) __A,
+          (__mmask16) __U);
+}
+
+static __inline__ __m512i __DEFAULT_FN_ATTRS
+_mm512_mask_mov_epi64 (__m512i __W, __mmask8 __U, __m512i __A)
+{
+  return (__m512i) __builtin_ia32_movdqa64_512_mask ((__v8di) __A,
+                 (__v8di) __W,
+                 (__mmask8) __U);
+}
+
+static __inline__ __m512i __DEFAULT_FN_ATTRS
+_mm512_maskz_mov_epi64 (__mmask8 __U, __m512i __A)
+{
+  return (__m512i) __builtin_ia32_movdqa64_512_mask ((__v8di) __A,
+                 (__v8di)
+                 _mm512_setzero_si512 (),
+                 (__mmask8) __U);
+}
+
+static __inline__ __m512i __DEFAULT_FN_ATTRS
+_mm512_mask_load_epi64 (__m512i __W, __mmask8 __U, void const *__P)
+{
+  return (__m512i) __builtin_ia32_movdqa64load512_mask ((const __v8di *) __P,
+              (__v8di) __W,
+              (__mmask8) __U);
+}
+
+static __inline__ __m512i __DEFAULT_FN_ATTRS
+_mm512_maskz_load_epi64 (__mmask8 __U, void const *__P)
+{
+  return (__m512i) __builtin_ia32_movdqa64load512_mask ((const __v8di *) __P,
+              (__v8di)
+              _mm512_setzero_si512 (),
+              (__mmask8) __U);
+}
+
+static __inline__ void __DEFAULT_FN_ATTRS
+_mm512_mask_store_epi64 (void *__P, __mmask8 __U, __m512i __A)
+{
+  __builtin_ia32_movdqa64store512_mask ((__v8di *) __P, (__v8di) __A,
+          (__mmask8) __U);
+}
+
+
 #undef __DEFAULT_FN_ATTRS
 
 #endif // __AVX512FINTRIN_H
diff --git a/lib/Headers/avx512vlintrin.h b/lib/Headers/avx512vlintrin.h
index 047ec9ca39..873f02d019 100644
--- a/lib/Headers/avx512vlintrin.h
+++ b/lib/Headers/avx512vlintrin.h
@@ -5871,6 +5871,114 @@ _mm256_maskz_srav_epi64 (__mmask8 __U, __m256i __X, __m256i __Y)
               _mm256_setzero_si256 (),
               (__mmask8) __U);
 }
+
+
+
+static __inline__ void __DEFAULT_FN_ATTRS
+_mm_mask_store_epi32 (void *__P, __mmask8 __U, __m128i __A)
+{
+  __builtin_ia32_movdqa32store128_mask ((__v4si *) __P,
+          (__v4si) __A,
+          (__mmask8) __U);
+}
+
+static __inline__ void __DEFAULT_FN_ATTRS
+_mm256_mask_store_epi32 (void *__P, __mmask8 __U, __m256i __A)
+{
+  __builtin_ia32_movdqa32store256_mask ((__v8si *) __P,
+          (__v8si) __A,
+          (__mmask8) __U);
+}
+
+static __inline__ __m128i __DEFAULT_FN_ATTRS
+_mm_mask_mov_epi64 (__m128i __W, __mmask8 __U, __m128i __A)
+{
+  return (__m128i) __builtin_ia32_movdqa64_128_mask ((__v2di) __A,
+                 (__v2di) __W,
+                 (__mmask8) __U);
+}
+
+static __inline__ __m128i __DEFAULT_FN_ATTRS
+_mm_maskz_mov_epi64 (__mmask8 __U, __m128i __A)
+{
+  return (__m128i) __builtin_ia32_movdqa64_128_mask ((__v2di) __A,
+                 (__v2di)
+                 _mm_setzero_di (),
+                 (__mmask8) __U);
+}
+
+static __inline__ __m256i __DEFAULT_FN_ATTRS
+_mm256_mask_mov_epi64 (__m256i __W, __mmask8 __U, __m256i __A)
+{
+  return (__m256i) __builtin_ia32_movdqa64_256_mask ((__v4di) __A,
+                 (__v4di) __W,
+                 (__mmask8) __U);
+}
+
+static __inline__ __m256i __DEFAULT_FN_ATTRS
+_mm256_maskz_mov_epi64 (__mmask8 __U, __m256i __A)
+{
+  return (__m256i) __builtin_ia32_movdqa64_256_mask ((__v4di) __A,
+                 (__v4di)
+                 _mm256_setzero_si256 (),
+                 (__mmask8) __U);
+}
+
+static __inline__ __m128i __DEFAULT_FN_ATTRS
+_mm_mask_load_epi64 (__m128i __W, __mmask8 __U, void const *__P)
+{
+  return (__m128i) __builtin_ia32_movdqa64load128_mask ((__v2di *) __P,
+              (__v2di) __W,
+              (__mmask8)
+              __U);
+}
+
+static __inline__ __m128i __DEFAULT_FN_ATTRS
+_mm_maskz_load_epi64 (__mmask8 __U, void const *__P)
+{
+  return (__m128i) __builtin_ia32_movdqa64load128_mask ((__v2di *) __P,
+              (__v2di)
+              _mm_setzero_di (),
+              (__mmask8)
+              __U);
+}
+
+static __inline__ __m256i __DEFAULT_FN_ATTRS
+_mm256_mask_load_epi64 (__m256i __W, __mmask8 __U, void const *__P)
+{
+  return (__m256i) __builtin_ia32_movdqa64load256_mask ((__v4di *) __P,
+              (__v4di) __W,
+              (__mmask8)
+              __U);
+}
+
+static __inline__ __m256i __DEFAULT_FN_ATTRS
+_mm256_maskz_load_epi64 (__mmask8 __U, void const *__P)
+{
+  return (__m256i) __builtin_ia32_movdqa64load256_mask ((__v4di *) __P,
+              (__v4di)
+              _mm256_setzero_si256 (),
+              (__mmask8)
+              __U);
+}
+
+static __inline__ void __DEFAULT_FN_ATTRS
+_mm_mask_store_epi64 (void *__P, __mmask8 __U, __m128i __A)
+{
+  __builtin_ia32_movdqa64store128_mask ((__v2di *) __P,
+          (__v2di) __A,
+          (__mmask8) __U);
+}
+
+static __inline__ void __DEFAULT_FN_ATTRS
+_mm256_mask_store_epi64 (void *__P, __mmask8 __U, __m256i __A)
+{
+  __builtin_ia32_movdqa64store256_mask ((__v4di *) __P,
+          (__v4di) __A,
+          (__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 ef8c158e10..f9e7ea7f86 100644
--- a/test/CodeGen/avx512f-builtins.c
+++ b/test/CodeGen/avx512f-builtins.c
@@ -2298,3 +2298,51 @@ __m512i test_mm512_maskz_srli_epi64(__mmask8 __U, __m512i __A) {
   return _mm512_maskz_srli_epi64(__U, __A, 5); 
 }
 
+__m512i test_mm512_mask_load_epi32(__m512i __W, __mmask16 __U, void const *__P) {
+  // CHECK-LABEL: @test_mm512_mask_load_epi32
+  // CHECK: @llvm.x86.avx512.mask.load.d.512
+  return _mm512_mask_load_epi32(__W, __U, __P); 
+}
+
+__m512i test_mm512_maskz_load_epi32(__mmask16 __U, void const *__P) {
+  // CHECK-LABEL: @test_mm512_maskz_load_epi32
+  // CHECK: @llvm.x86.avx512.mask.load.d.512
+  return _mm512_maskz_load_epi32(__U, __P); 
+}
+
+__m512i test_mm512_mask_mov_epi64(__m512i __W, __mmask8 __U, __m512i __A) {
+  // CHECK-LABEL: @test_mm512_mask_mov_epi64
+  // CHECK: @llvm.x86.avx512.mask.mov
+  return _mm512_mask_mov_epi64(__W, __U, __A); 
+}
+
+__m512i test_mm512_maskz_mov_epi64(__mmask8 __U, __m512i __A) {
+  // CHECK-LABEL: @test_mm512_maskz_mov_epi64
+  // CHECK: @llvm.x86.avx512.mask.mov
+  return _mm512_maskz_mov_epi64(__U, __A); 
+}
+
+__m512i test_mm512_mask_load_epi64(__m512i __W, __mmask8 __U, void const *__P) {
+  // CHECK-LABEL: @test_mm512_mask_load_epi64
+  // CHECK: @llvm.x86.avx512.mask.load.q.512
+  return _mm512_mask_load_epi64(__W, __U, __P); 
+}
+
+__m512i test_mm512_maskz_load_epi64(__mmask8 __U, void const *__P) {
+  // CHECK-LABEL: @test_mm512_maskz_load_epi64
+  // CHECK: @llvm.x86.avx512.mask.load.q.512
+  return _mm512_maskz_load_epi64(__U, __P); 
+}
+
+void test_mm512_mask_store_epi32(void *__P, __mmask16 __U, __m512i __A) {
+  // CHECK-LABEL: @test_mm512_mask_store_epi32
+  // CHECK: @llvm.x86.avx512.mask.store.d.512
+  return _mm512_mask_store_epi32(__P, __U, __A); 
+}
+
+void test_mm512_mask_store_epi64(void *__P, __mmask8 __U, __m512i __A) {
+  // CHECK-LABEL: @test_mm512_mask_store_epi64
+  // CHECK: @llvm.x86.avx512.mask.store.q.512
+  return _mm512_mask_store_epi64(__P, __U, __A); 
+}
+
diff --git a/test/CodeGen/avx512vl-builtins.c b/test/CodeGen/avx512vl-builtins.c
index 7061369940..c8c27322a3 100644
--- a/test/CodeGen/avx512vl-builtins.c
+++ b/test/CodeGen/avx512vl-builtins.c
@@ -3936,3 +3936,75 @@ __m256i test_mm256_maskz_srav_epi64(__mmask8 __U, __m256i __X, __m256i __Y) {
   return _mm256_maskz_srav_epi64(__U, __X, __Y); 
 }
 
+void test_mm_mask_store_epi32(void *__P, __mmask8 __U, __m128i __A) {
+  // CHECK-LABEL: @test_mm_mask_store_epi32
+  // CHECK: @llvm.x86.avx512.mask.store.d.128
+  return _mm_mask_store_epi32(__P, __U, __A); 
+}
+
+void test_mm256_mask_store_epi32(void *__P, __mmask8 __U, __m256i __A) {
+  // CHECK-LABEL: @test_mm256_mask_store_epi32
+  // CHECK: @llvm.x86.avx512.mask.store.d.256
+  return _mm256_mask_store_epi32(__P, __U, __A); 
+}
+
+__m128i test_mm_mask_mov_epi64(__m128i __W, __mmask8 __U, __m128i __A) {
+  // CHECK-LABEL: @test_mm_mask_mov_epi64
+  // CHECK: @llvm.x86.avx512.mask.mov
+  return _mm_mask_mov_epi64(__W, __U, __A); 
+}
+
+__m128i test_mm_maskz_mov_epi64(__mmask8 __U, __m128i __A) {
+  // CHECK-LABEL: @test_mm_maskz_mov_epi64
+  // CHECK: @llvm.x86.avx512.mask.mov
+  return _mm_maskz_mov_epi64(__U, __A); 
+}
+
+__m256i test_mm256_mask_mov_epi64(__m256i __W, __mmask8 __U, __m256i __A) {
+  // CHECK-LABEL: @test_mm256_mask_mov_epi64
+  // CHECK: @llvm.x86.avx512.mask.mov
+  return _mm256_mask_mov_epi64(__W, __U, __A); 
+}
+
+__m256i test_mm256_maskz_mov_epi64(__mmask8 __U, __m256i __A) {
+  // CHECK-LABEL: @test_mm256_maskz_mov_epi64
+  // CHECK: @llvm.x86.avx512.mask.mov
+  return _mm256_maskz_mov_epi64(__U, __A); 
+}
+
+__m128i test_mm_mask_load_epi64(__m128i __W, __mmask8 __U, void const *__P) {
+  // CHECK-LABEL: @test_mm_mask_load_epi64
+  // CHECK: @llvm.x86.avx512.mask.load.q.128
+  return _mm_mask_load_epi64(__W, __U, __P); 
+}
+
+__m128i test_mm_maskz_load_epi64(__mmask8 __U, void const *__P) {
+  // CHECK-LABEL: @test_mm_maskz_load_epi64
+  // CHECK: @llvm.x86.avx512.mask.load.q.128
+  return _mm_maskz_load_epi64(__U, __P); 
+}
+
+__m256i test_mm256_mask_load_epi64(__m256i __W, __mmask8 __U, void const *__P) {
+  // CHECK-LABEL: @test_mm256_mask_load_epi64
+  // CHECK: @llvm.x86.avx512.mask.load.q.256
+  return _mm256_mask_load_epi64(__W, __U, __P); 
+}
+
+__m256i test_mm256_maskz_load_epi64(__mmask8 __U, void const *__P) {
+  // CHECK-LABEL: @test_mm256_maskz_load_epi64
+  // CHECK: @llvm.x86.avx512.mask.load.q.256
+  return _mm256_maskz_load_epi64(__U, __P); 
+}
+
+void test_mm_mask_store_epi64(void *__P, __mmask8 __U, __m128i __A) {
+  // CHECK-LABEL: @test_mm_mask_store_epi64
+  // CHECK: @llvm.x86.avx512.mask.store.q.128
+  return _mm_mask_store_epi64(__P, __U, __A); 
+}
+
+void test_mm256_mask_store_epi64(void *__P, __mmask8 __U, __m256i __A) {
+  // CHECK-LABEL: @test_mm256_mask_store_epi64
+  // CHECK: @llvm.x86.avx512.mask.store.q.256
+  return _mm256_mask_store_epi64(__P, __U, __A); 
+}
+