https://github.com/RKSimon updated
https://github.com/llvm/llvm-project/pull/152705
>From 9e77ca23a4e075db6896bdd66c1a5877ac90133c Mon Sep 17 00:00:00 2001
From: Simon Pilgrim <llvm-...@redking.me.uk>
Date: Fri, 8 Aug 2025 13:47:58 +0100
Subject: [PATCH] [Clang][X86] Add avx512 __builtin_ia32_select* constexpr
handling
Fixes #152321
---
clang/include/clang/Basic/BuiltinsX86.td | 48 +++++++++----------
clang/lib/AST/ExprConstant.cpp | 45 +++++++++++++++++
clang/lib/Headers/avx512fintrin.h | 4 +-
clang/lib/Headers/avx512vpopcntdqintrin.h | 16 +++----
clang/lib/Headers/avx512vpopcntdqvlintrin.h | 28 ++++++-----
clang/test/CodeGen/X86/avx512f-builtins.c | 3 ++
.../CodeGen/X86/avx512vpopcntdq-builtins.c | 4 ++
.../CodeGen/X86/avx512vpopcntdqvl-builtins.c | 8 ++++
8 files changed, 109 insertions(+), 47 deletions(-)
diff --git a/clang/include/clang/Basic/BuiltinsX86.td
b/clang/include/clang/Basic/BuiltinsX86.td
index a4acc72fdc37d..492be89a0fe89 100644
--- a/clang/include/clang/Basic/BuiltinsX86.td
+++ b/clang/include/clang/Basic/BuiltinsX86.td
@@ -4246,99 +4246,99 @@ let Features = "avx512fp16,evex512", Attributes =
[NoThrow, Const, RequiredVecto
def vfcmulcph512_mask : X86Builtin<"_Vector<16, float>(_Vector<16, float>,
_Vector<16, float>, _Vector<16, float>, unsigned short, _Constant int)">;
}
-let Features = "avx512bw,avx512vl", Attributes = [NoThrow, Const,
RequiredVectorWidth<128>] in {
+let Features = "avx512bw,avx512vl", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<128>] in {
def selectb_128 : X86Builtin<"_Vector<16, char>(unsigned short, _Vector<16,
char>, _Vector<16, char>)">;
}
-let Features = "avx512bw,avx512vl", Attributes = [NoThrow, Const,
RequiredVectorWidth<256>] in {
+let Features = "avx512bw,avx512vl", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<256>] in {
def selectb_256 : X86Builtin<"_Vector<32, char>(unsigned int, _Vector<32,
char>, _Vector<32, char>)">;
}
-let Features = "avx512bw,evex512", Attributes = [NoThrow, Const,
RequiredVectorWidth<512>] in {
+let Features = "avx512bw,evex512", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<512>] in {
def selectb_512 : X86Builtin<"_Vector<64, char>(unsigned long long int,
_Vector<64, char>, _Vector<64, char>)">;
}
-let Features = "avx512bw,avx512vl", Attributes = [NoThrow, Const,
RequiredVectorWidth<128>] in {
+let Features = "avx512bw,avx512vl", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<128>] in {
def selectw_128 : X86Builtin<"_Vector<8, short>(unsigned char, _Vector<8,
short>, _Vector<8, short>)">;
}
-let Features = "avx512bw,avx512vl", Attributes = [NoThrow, Const,
RequiredVectorWidth<256>] in {
+let Features = "avx512bw,avx512vl", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<256>] in {
def selectw_256 : X86Builtin<"_Vector<16, short>(unsigned short, _Vector<16,
short>, _Vector<16, short>)">;
}
-let Features = "avx512bw,evex512", Attributes = [NoThrow, Const,
RequiredVectorWidth<512>] in {
+let Features = "avx512bw,evex512", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<512>] in {
def selectw_512 : X86Builtin<"_Vector<32, short>(unsigned int, _Vector<32,
short>, _Vector<32, short>)">;
}
-let Features = "avx512vl", Attributes = [NoThrow, Const,
RequiredVectorWidth<128>] in {
+let Features = "avx512vl", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<128>] in {
def selectd_128 : X86Builtin<"_Vector<4, int>(unsigned char, _Vector<4,
int>, _Vector<4, int>)">;
}
-let Features = "avx512vl", Attributes = [NoThrow, Const,
RequiredVectorWidth<256>] in {
+let Features = "avx512vl", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<256>] in {
def selectd_256 : X86Builtin<"_Vector<8, int>(unsigned char, _Vector<8,
int>, _Vector<8, int>)">;
}
-let Features = "avx512f,evex512", Attributes = [NoThrow, Const,
RequiredVectorWidth<512>] in {
+let Features = "avx512f,evex512", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<512>] in {
def selectd_512 : X86Builtin<"_Vector<16, int>(unsigned short, _Vector<16,
int>, _Vector<16, int>)">;
}
-let Features = "avx512fp16,avx512vl", Attributes = [NoThrow, Const,
RequiredVectorWidth<128>] in {
+let Features = "avx512fp16,avx512vl", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<128>] in {
def selectph_128 : X86Builtin<"_Vector<8, _Float16>(unsigned char,
_Vector<8, _Float16>, _Vector<8, _Float16>)">;
}
-let Features = "avx512fp16,avx512vl", Attributes = [NoThrow, Const,
RequiredVectorWidth<256>] in {
+let Features = "avx512fp16,avx512vl", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<256>] in {
def selectph_256 : X86Builtin<"_Vector<16, _Float16>(unsigned short,
_Vector<16, _Float16>, _Vector<16, _Float16>)">;
}
-let Features = "avx512fp16,evex512", Attributes = [NoThrow, Const,
RequiredVectorWidth<512>] in {
+let Features = "avx512fp16,evex512", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<512>] in {
def selectph_512 : X86Builtin<"_Vector<32, _Float16>(unsigned int,
_Vector<32, _Float16>, _Vector<32, _Float16>)">;
}
-let Features = "avx512bf16,avx512vl", Attributes = [NoThrow, Const,
RequiredVectorWidth<128>] in {
+let Features = "avx512bf16,avx512vl", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<128>] in {
def selectpbf_128 : X86Builtin<"_Vector<8, __bf16>(unsigned char, _Vector<8,
__bf16>, _Vector<8, __bf16>)">;
}
-let Features = "avx512bf16,avx512vl", Attributes = [NoThrow, Const,
RequiredVectorWidth<256>] in {
+let Features = "avx512bf16,avx512vl", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<256>] in {
def selectpbf_256 : X86Builtin<"_Vector<16, __bf16>(unsigned short,
_Vector<16, __bf16>, _Vector<16, __bf16>)">;
}
-let Features = "avx512bf16,evex512", Attributes = [NoThrow, Const,
RequiredVectorWidth<512>] in {
+let Features = "avx512bf16,evex512", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<512>] in {
def selectpbf_512 : X86Builtin<"_Vector<32, __bf16>(unsigned int,
_Vector<32, __bf16>, _Vector<32, __bf16>)">;
}
-let Features = "avx512vl", Attributes = [NoThrow, Const,
RequiredVectorWidth<128>] in {
+let Features = "avx512vl", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<128>] in {
def selectq_128 : X86Builtin<"_Vector<2, long long int>(unsigned char,
_Vector<2, long long int>, _Vector<2, long long int>)">;
}
-let Features = "avx512vl", Attributes = [NoThrow, Const,
RequiredVectorWidth<256>] in {
+let Features = "avx512vl", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<256>] in {
def selectq_256 : X86Builtin<"_Vector<4, long long int>(unsigned char,
_Vector<4, long long int>, _Vector<4, long long int>)">;
}
-let Features = "avx512f,evex512", Attributes = [NoThrow, Const,
RequiredVectorWidth<512>] in {
+let Features = "avx512f,evex512", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<512>] in {
def selectq_512 : X86Builtin<"_Vector<8, long long int>(unsigned char,
_Vector<8, long long int>, _Vector<8, long long int>)">;
}
-let Features = "avx512vl", Attributes = [NoThrow, Const,
RequiredVectorWidth<128>] in {
+let Features = "avx512vl", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<128>] in {
def selectps_128 : X86Builtin<"_Vector<4, float>(unsigned char, _Vector<4,
float>, _Vector<4, float>)">;
}
-let Features = "avx512vl", Attributes = [NoThrow, Const,
RequiredVectorWidth<256>] in {
+let Features = "avx512vl", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<256>] in {
def selectps_256 : X86Builtin<"_Vector<8, float>(unsigned char, _Vector<8,
float>, _Vector<8, float>)">;
}
-let Features = "avx512f,evex512", Attributes = [NoThrow, Const,
RequiredVectorWidth<512>] in {
+let Features = "avx512f,evex512", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<512>] in {
def selectps_512 : X86Builtin<"_Vector<16, float>(unsigned short,
_Vector<16, float>, _Vector<16, float>)">;
}
-let Features = "avx512vl", Attributes = [NoThrow, Const,
RequiredVectorWidth<128>] in {
+let Features = "avx512vl", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<128>] in {
def selectpd_128 : X86Builtin<"_Vector<2, double>(unsigned char, _Vector<2,
double>, _Vector<2, double>)">;
}
-let Features = "avx512vl", Attributes = [NoThrow, Const,
RequiredVectorWidth<256>] in {
+let Features = "avx512vl", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<256>] in {
def selectpd_256 : X86Builtin<"_Vector<4, double>(unsigned char, _Vector<4,
double>, _Vector<4, double>)">;
}
-let Features = "avx512f,evex512", Attributes = [NoThrow, Const,
RequiredVectorWidth<512>] in {
+let Features = "avx512f,evex512", Attributes = [NoThrow, Const, Constexpr,
RequiredVectorWidth<512>] in {
def selectpd_512 : X86Builtin<"_Vector<8, double>(unsigned char, _Vector<8,
double>, _Vector<8, double>)">;
}
diff --git a/clang/lib/AST/ExprConstant.cpp b/clang/lib/AST/ExprConstant.cpp
index 3679327da7b0c..c6d81bccf42e1 100644
--- a/clang/lib/AST/ExprConstant.cpp
+++ b/clang/lib/AST/ExprConstant.cpp
@@ -11658,6 +11658,51 @@ bool VectorExprEvaluator::VisitCallExpr(const CallExpr
*E) {
return Success(APValue(ResultElements.data(), ResultElements.size()), E);
}
+ case X86::BI__builtin_ia32_selectb_128:
+ case X86::BI__builtin_ia32_selectb_256:
+ case X86::BI__builtin_ia32_selectb_512:
+ case X86::BI__builtin_ia32_selectw_128:
+ case X86::BI__builtin_ia32_selectw_256:
+ case X86::BI__builtin_ia32_selectw_512:
+ case X86::BI__builtin_ia32_selectd_128:
+ case X86::BI__builtin_ia32_selectd_256:
+ case X86::BI__builtin_ia32_selectd_512:
+ case X86::BI__builtin_ia32_selectq_128:
+ case X86::BI__builtin_ia32_selectq_256:
+ case X86::BI__builtin_ia32_selectq_512:
+ case X86::BI__builtin_ia32_selectph_128:
+ case X86::BI__builtin_ia32_selectph_256:
+ case X86::BI__builtin_ia32_selectph_512:
+ case X86::BI__builtin_ia32_selectpbf_128:
+ case X86::BI__builtin_ia32_selectpbf_256:
+ case X86::BI__builtin_ia32_selectpbf_512:
+ case X86::BI__builtin_ia32_selectps_128:
+ case X86::BI__builtin_ia32_selectps_256:
+ case X86::BI__builtin_ia32_selectps_512:
+ case X86::BI__builtin_ia32_selectpd_128:
+ case X86::BI__builtin_ia32_selectpd_256:
+ case X86::BI__builtin_ia32_selectpd_512: {
+ // AVX512 predicated move: "Result = Mask[] ? LHS[] : RHS[]".
+ APValue SourceMask, SourceLHS, SourceRHS;
+ if (!EvaluateAsRValue(Info, E->getArg(0), SourceMask) ||
+ !EvaluateAsRValue(Info, E->getArg(1), SourceLHS) ||
+ !EvaluateAsRValue(Info, E->getArg(2), SourceRHS))
+ return false;
+
+ APSInt Mask = SourceMask.getInt();
+ QualType DestEltTy = E->getType()->castAs<VectorType>()->getElementType();
+ unsigned SourceLen = SourceLHS.getVectorLength();
+ SmallVector<APValue, 4> ResultElements;
+ ResultElements.reserve(SourceLen);
+
+ for (unsigned EltNum = 0; EltNum < SourceLen; ++EltNum) {
+ const APValue &LHS = SourceLHS.getVectorElt(EltNum);
+ const APValue &RHS = SourceRHS.getVectorElt(EltNum);
+ ResultElements.push_back(Mask[EltNum] ? LHS : RHS);
+ }
+
+ return Success(APValue(ResultElements.data(), ResultElements.size()), E);
+ }
}
}
diff --git a/clang/lib/Headers/avx512fintrin.h
b/clang/lib/Headers/avx512fintrin.h
index e3bc71ed8e918..c1f61e76ed0e2 100644
--- a/clang/lib/Headers/avx512fintrin.h
+++ b/clang/lib/Headers/avx512fintrin.h
@@ -8962,7 +8962,7 @@ _mm512_mask_cvtpslo_pd (__m512d __W, __mmask8 __U, __m512
__A)
return (__m512d) _mm512_mask_cvtps_pd(__W, __U, _mm512_castps512_ps256(__A));
}
-static __inline__ __m512d __DEFAULT_FN_ATTRS512
+static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_mask_mov_pd (__m512d __W, __mmask8 __U, __m512d __A)
{
return (__m512d) __builtin_ia32_selectpd_512 ((__mmask8) __U,
@@ -8970,7 +8970,7 @@ _mm512_mask_mov_pd (__m512d __W, __mmask8 __U, __m512d
__A)
(__v8df) __W);
}
-static __inline__ __m512d __DEFAULT_FN_ATTRS512
+static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_maskz_mov_pd (__mmask8 __U, __m512d __A)
{
return (__m512d) __builtin_ia32_selectpd_512 ((__mmask8) __U,
diff --git a/clang/lib/Headers/avx512vpopcntdqintrin.h
b/clang/lib/Headers/avx512vpopcntdqintrin.h
index 79fc6e140c616..7bf0700da77c9 100644
--- a/clang/lib/Headers/avx512vpopcntdqintrin.h
+++ b/clang/lib/Headers/avx512vpopcntdqintrin.h
@@ -16,18 +16,19 @@
#define __AVX512VPOPCNTDQINTRIN_H
/* Define the default attributes for the functions in this file. */
+#if defined(__cplusplus) && (__cplusplus >= 201103L)
#define __DEFAULT_FN_ATTRS
\
__attribute__((__always_inline__, __nodebug__,
\
__target__("avx512vpopcntdq,evex512"),
\
- __min_vector_width__(512)))
-
-#if defined(__cplusplus) && (__cplusplus >= 201103L)
-#define __DEFAULT_FN_ATTRS_CONSTEXPR __DEFAULT_FN_ATTRS constexpr
+ __min_vector_width__(512))) constexpr
#else
-#define __DEFAULT_FN_ATTRS_CONSTEXPR __DEFAULT_FN_ATTRS
+#define __DEFAULT_FN_ATTRS
\
+ __attribute__((__always_inline__, __nodebug__,
\
+ __target__("avx512vpopcntdq,evex512"),
\
+ __min_vector_width__(512)))
#endif
-static __inline__ __m512i __DEFAULT_FN_ATTRS_CONSTEXPR
+static __inline__ __m512i __DEFAULT_FN_ATTRS
_mm512_popcnt_epi64(__m512i __A) {
return (__m512i)__builtin_elementwise_popcount((__v8du)__A);
}
@@ -43,7 +44,7 @@ _mm512_maskz_popcnt_epi64(__mmask8 __U, __m512i __A) {
return _mm512_mask_popcnt_epi64((__m512i)_mm512_setzero_si512(), __U, __A);
}
-static __inline__ __m512i __DEFAULT_FN_ATTRS_CONSTEXPR
+static __inline__ __m512i __DEFAULT_FN_ATTRS
_mm512_popcnt_epi32(__m512i __A) {
return (__m512i)__builtin_elementwise_popcount((__v16su)__A);
}
@@ -60,6 +61,5 @@ _mm512_maskz_popcnt_epi32(__mmask16 __U, __m512i __A) {
}
#undef __DEFAULT_FN_ATTRS
-#undef __DEFAULT_FN_ATTRS_CONSTEXPR
#endif
diff --git a/clang/lib/Headers/avx512vpopcntdqvlintrin.h
b/clang/lib/Headers/avx512vpopcntdqvlintrin.h
index d14cb1eb31f14..deb849f950357 100644
--- a/clang/lib/Headers/avx512vpopcntdqvlintrin.h
+++ b/clang/lib/Headers/avx512vpopcntdqvlintrin.h
@@ -16,6 +16,17 @@
#define __AVX512VPOPCNTDQVLINTRIN_H
/* Define the default attributes for the functions in this file. */
+
+#if defined(__cplusplus) && (__cplusplus >= 201103L)
+#define __DEFAULT_FN_ATTRS128
\
+ __attribute__((__always_inline__, __nodebug__,
\
+ __target__("avx512vpopcntdq,avx512vl,no-evex512"),
\
+ __min_vector_width__(128))) constexpr
+#define __DEFAULT_FN_ATTRS256
\
+ __attribute__((__always_inline__, __nodebug__,
\
+ __target__("avx512vpopcntdq,avx512vl,no-evex512"),
\
+ __min_vector_width__(256))) constexpr
+#else
#define __DEFAULT_FN_ATTRS128
\
__attribute__((__always_inline__, __nodebug__,
\
__target__("avx512vpopcntdq,avx512vl,no-evex512"),
\
@@ -24,16 +35,9 @@
__attribute__((__always_inline__, __nodebug__,
\
__target__("avx512vpopcntdq,avx512vl,no-evex512"),
\
__min_vector_width__(256)))
-
-#if defined(__cplusplus) && (__cplusplus >= 201103L)
-#define __DEFAULT_FN_ATTRS128_CONSTEXPR __DEFAULT_FN_ATTRS128 constexpr
-#define __DEFAULT_FN_ATTRS256_CONSTEXPR __DEFAULT_FN_ATTRS256 constexpr
-#else
-#define __DEFAULT_FN_ATTRS128_CONSTEXPR __DEFAULT_FN_ATTRS128
-#define __DEFAULT_FN_ATTRS256_CONSTEXPR __DEFAULT_FN_ATTRS256
#endif
-static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR
+static __inline__ __m128i __DEFAULT_FN_ATTRS128
_mm_popcnt_epi64(__m128i __A) {
return (__m128i)__builtin_elementwise_popcount((__v2du)__A);
}
@@ -49,7 +53,7 @@ _mm_maskz_popcnt_epi64(__mmask8 __U, __m128i __A) {
return _mm_mask_popcnt_epi64((__m128i)_mm_setzero_si128(), __U, __A);
}
-static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR
+static __inline__ __m128i __DEFAULT_FN_ATTRS128
_mm_popcnt_epi32(__m128i __A) {
return (__m128i)__builtin_elementwise_popcount((__v4su)__A);
}
@@ -65,7 +69,7 @@ _mm_maskz_popcnt_epi32(__mmask8 __U, __m128i __A) {
return _mm_mask_popcnt_epi32((__m128i)_mm_setzero_si128(), __U, __A);
}
-static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_popcnt_epi64(__m256i __A) {
return (__m256i)__builtin_elementwise_popcount((__v4du)__A);
}
@@ -81,7 +85,7 @@ _mm256_maskz_popcnt_epi64(__mmask8 __U, __m256i __A) {
return _mm256_mask_popcnt_epi64((__m256i)_mm256_setzero_si256(), __U, __A);
}
-static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_popcnt_epi32(__m256i __A) {
return (__m256i)__builtin_elementwise_popcount((__v8su)__A);
}
@@ -99,7 +103,5 @@ _mm256_maskz_popcnt_epi32(__mmask8 __U, __m256i __A) {
#undef __DEFAULT_FN_ATTRS128
#undef __DEFAULT_FN_ATTRS256
-#undef __DEFAULT_FN_ATTRS128_CONSTEXPR
-#undef __DEFAULT_FN_ATTRS256_CONSTEXPR
#endif
diff --git a/clang/test/CodeGen/X86/avx512f-builtins.c
b/clang/test/CodeGen/X86/avx512f-builtins.c
index d59799e2c24e5..f939b5ac01a8c 100644
--- a/clang/test/CodeGen/X86/avx512f-builtins.c
+++ b/clang/test/CodeGen/X86/avx512f-builtins.c
@@ -8905,17 +8905,20 @@ __m512d test_mm512_maskz_cvtps_pd(__mmask8 __U, __m256
__A) {
// CHECK: select <8 x i1> %{{.*}}, <8 x double> %{{.*}}, <8 x double> %{{.*}}
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: select <8 x i1> %{{.*}}, <8 x double> %{{.*}}, <8 x double> %{{.*}}
return _mm512_mask_mov_pd(__W, __U, __A);
}
+TEST_CONSTEXPR(match_m512d(_mm512_mask_mov_pd((__m512d){-8.0, -7.0, -6.0,
-5.0, -4.0, -3.0, -2.0, -1.0}, 0xC3, (__m512d){+1.0, +2.0, +3.0, +4.0, +5.0,
+6.0, +7.0, +8.0}), +1.0, +2.0, -6.0, -5.0, -4.0, -3.0, +7.0, +8.0));
__m512d test_mm512_maskz_mov_pd(__mmask8 __U, __m512d __A) {
// CHECK-LABEL: test_mm512_maskz_mov_pd
// CHECK: select <8 x i1> %{{.*}}, <8 x double> %{{.*}}, <8 x double> %{{.*}}
return _mm512_maskz_mov_pd(__U, __A);
}
+TEST_CONSTEXPR(match_m512d(_mm512_maskz_mov_pd(0xC3, (__m512d){+1.0, +2.0,
+3.0, +4.0, +5.0, +6.0, +7.0, +8.0}), +1.0, +2.0, +0.0, +0.0, +0.0, +0.0, +7.0,
+8.0));
__m512 test_mm512_mask_mov_ps(__m512 __W, __mmask16 __U, __m512 __A) {
// CHECK-LABEL: test_mm512_mask_mov_ps
diff --git a/clang/test/CodeGen/X86/avx512vpopcntdq-builtins.c
b/clang/test/CodeGen/X86/avx512vpopcntdq-builtins.c
index 8927ae273d29b..b80ffdf0dcc7c 100644
--- a/clang/test/CodeGen/X86/avx512vpopcntdq-builtins.c
+++ b/clang/test/CodeGen/X86/avx512vpopcntdq-builtins.c
@@ -19,6 +19,7 @@ __m512i test_mm512_mask_popcnt_epi64(__m512i __W, __mmask8
__U, __m512i __A) {
// CHECK: select <8 x i1> %{{[0-9]+}}, <8 x i64> %{{.*}}, <8 x i64> %{{.*}}
return _mm512_mask_popcnt_epi64(__W, __U, __A);
}
+TEST_CONSTEXPR(match_v8di(_mm512_mask_popcnt_epi64(_mm512_set1_epi64(-1),
0x81, (__m512i)(__v8di){+5, -3, -10, +8, 0, -256, +256, -128}), 2, -1, -1, -1,
-1, -1, -1, 57));
__m512i test_mm512_maskz_popcnt_epi64(__mmask8 __U, __m512i __A) {
// CHECK-LABEL: test_mm512_maskz_popcnt_epi64
@@ -26,6 +27,7 @@ __m512i test_mm512_maskz_popcnt_epi64(__mmask8 __U, __m512i
__A) {
// CHECK: select <8 x i1> %{{[0-9]+}}, <8 x i64> %{{.*}}, <8 x i64> %{{.*}}
return _mm512_maskz_popcnt_epi64(__U, __A);
}
+TEST_CONSTEXPR(match_v8di(_mm512_maskz_popcnt_epi64(0x42,
(__m512i)(__v8di){+5, -3, -10, +8, 0, -256, +256, -128}), 0, 63, 0, 0, 0, 0, 1,
0));
__m512i test_mm512_popcnt_epi32(__m512i __A) {
// CHECK-LABEL: test_mm512_popcnt_epi32
@@ -40,6 +42,7 @@ __m512i test_mm512_mask_popcnt_epi32(__m512i __W, __mmask16
__U, __m512i __A) {
// CHECK: select <16 x i1> %{{[0-9]+}}, <16 x i32> %{{.*}}, <16 x i32>
%{{.*}}
return _mm512_mask_popcnt_epi32(__W, __U, __A);
}
+TEST_CONSTEXPR(match_v16si(_mm512_mask_popcnt_epi32(_mm512_set1_epi32(-1),
0x0F81, (__m512i)(__v16si){+5, -3, -10, +8, 0, -256, +256, -128, +3, +9, +15,
+33, +63, +129, +511, +1025}), 2, -1, -1, -1, -1, -1, -1, 25, 2, 2, 4, 2, -1,
-1, -1, -1));
__m512i test_mm512_maskz_popcnt_epi32(__mmask16 __U, __m512i __A) {
// CHECK-LABEL: test_mm512_maskz_popcnt_epi32
@@ -47,3 +50,4 @@ __m512i test_mm512_maskz_popcnt_epi32(__mmask16 __U, __m512i
__A) {
// CHECK: select <16 x i1> %{{[0-9]+}}, <16 x i32> %{{.*}}, <16 x i32>
%{{.*}}
return _mm512_maskz_popcnt_epi32(__U, __A);
}
+TEST_CONSTEXPR(match_v16si(_mm512_maskz_popcnt_epi32(0xF042,
(__m512i)(__v16si){+5, -3, -10, +8, 0, -256, +256, -128, +3, +9, +15, +33, +63,
+129, +511, +1025}), 0, 31, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 6, 2, 9, 2));
diff --git a/clang/test/CodeGen/X86/avx512vpopcntdqvl-builtins.c
b/clang/test/CodeGen/X86/avx512vpopcntdqvl-builtins.c
index d9fbd7628142f..7258034c57adc 100644
--- a/clang/test/CodeGen/X86/avx512vpopcntdqvl-builtins.c
+++ b/clang/test/CodeGen/X86/avx512vpopcntdqvl-builtins.c
@@ -19,6 +19,7 @@ __m128i test_mm_mask_popcnt_epi64(__m128i __W, __mmask8 __U,
__m128i __A) {
// CHECK: select <2 x i1> %{{.+}}, <2 x i64> %{{.*}}, <2 x i64> %{{.*}}
return _mm_mask_popcnt_epi64(__W, __U, __A);
}
+TEST_CONSTEXPR(match_v2di(_mm_mask_popcnt_epi64(_mm_set1_epi64x(-1), 0x2,
(__m128i)(__v2di){+5, -3}), -1, 63));
__m128i test_mm_maskz_popcnt_epi64(__mmask8 __U, __m128i __A) {
// CHECK-LABEL: test_mm_maskz_popcnt_epi64
@@ -26,6 +27,7 @@ __m128i test_mm_maskz_popcnt_epi64(__mmask8 __U, __m128i __A)
{
// CHECK: select <2 x i1> %{{.+}}, <2 x i64> %{{.*}}, <2 x i64> %{{.*}}
return _mm_maskz_popcnt_epi64(__U, __A);
}
+TEST_CONSTEXPR(match_v2di(_mm_maskz_popcnt_epi64(0x1, (__m128i)(__v2di){+5,
-3}), 2, 0));
__m128i test_mm_popcnt_epi32(__m128i __A) {
// CHECK-LABEL: test_mm_popcnt_epi32
@@ -40,6 +42,7 @@ __m128i test_mm_mask_popcnt_epi32(__m128i __W, __mmask8 __U,
__m128i __A) {
// CHECK: select <4 x i1> %{{.+}}, <4 x i32> %{{.*}}, <4 x i32> %{{.*}}
return _mm_mask_popcnt_epi32(__W, __U, __A);
}
+TEST_CONSTEXPR(match_v4si(_mm_mask_popcnt_epi32(_mm_set1_epi32(-1), 0x3,
(__m128i)(__v4si){+5, -3, -10, +8}), 2, 31, -1, -1));
__m128i test_mm_maskz_popcnt_epi32(__mmask8 __U, __m128i __A) {
// CHECK-LABEL: test_mm_maskz_popcnt_epi32
@@ -47,6 +50,7 @@ __m128i test_mm_maskz_popcnt_epi32(__mmask8 __U, __m128i __A)
{
// CHECK: select <4 x i1> %{{.+}}, <4 x i32> %{{.*}}, <4 x i32> %{{.*}}
return _mm_maskz_popcnt_epi32(__U, __A);
}
+TEST_CONSTEXPR(match_v4si(_mm_maskz_popcnt_epi32(0x5, (__m128i)(__v4si){+5,
-3, -10, +8}), 2, 0, 30, 0));
__m256i test_mm256_popcnt_epi64(__m256i __A) {
// CHECK-LABEL: test_mm256_popcnt_epi64
@@ -61,6 +65,7 @@ __m256i test_mm256_mask_popcnt_epi64(__m256i __W, __mmask8
__U, __m256i __A) {
// CHECK: select <4 x i1> %{{.+}}, <4 x i64> %{{.*}}, <4 x i64> %{{.*}}
return _mm256_mask_popcnt_epi64(__W, __U, __A);
}
+TEST_CONSTEXPR(match_v4di(_mm256_mask_popcnt_epi64(_mm256_set1_epi64x(-1),
0x3, (__m256i)(__v4di){+5, -3, -10, +8}), 2, 63, -1, -1));
__m256i test_mm256_maskz_popcnt_epi64(__mmask8 __U, __m256i __A) {
// CHECK-LABEL: test_mm256_maskz_popcnt_epi64
@@ -68,6 +73,7 @@ __m256i test_mm256_maskz_popcnt_epi64(__mmask8 __U, __m256i
__A) {
// CHECK: select <4 x i1> %{{.+}}, <4 x i64> %{{.*}}, <4 x i64> %{{.*}}
return _mm256_maskz_popcnt_epi64(__U, __A);
}
+TEST_CONSTEXPR(match_v4di(_mm256_maskz_popcnt_epi64(0x5, (__m256i)(__v4di){+5,
-3, -10, +8}), 2, 0, 62, 0));
__m256i test_mm256_popcnt_epi32(__m256i __A) {
// CHECK-LABEL: test_mm256_popcnt_epi32
@@ -82,6 +88,7 @@ __m256i test_mm256_mask_popcnt_epi32(__m256i __W, __mmask8
__U, __m256i __A) {
// CHECK: select <8 x i1> %{{.+}}, <8 x i32> %{{.*}}, <8 x i32> %{{.*}}
return _mm256_mask_popcnt_epi32(__W, __U, __A);
}
+TEST_CONSTEXPR(match_v8si(_mm256_mask_popcnt_epi32(_mm256_set1_epi32(-1),
0x37, (__m256i)(__v8si){+5, -3, -10, +8, 0, -256, +256, -128}), 2, 31, 30, -1,
0, 24, -1, -1));
__m256i test_mm256_maskz_popcnt_epi32(__mmask8 __U, __m256i __A) {
// CHECK-LABEL: test_mm256_maskz_popcnt_epi32
@@ -89,3 +96,4 @@ __m256i test_mm256_maskz_popcnt_epi32(__mmask8 __U, __m256i
__A) {
// CHECK: select <8 x i1> %{{.+}}, <8 x i32> %{{.*}}, <8 x i32> %{{.*}}
return _mm256_maskz_popcnt_epi32(__U, __A);
}
+TEST_CONSTEXPR(match_v8si(_mm256_maskz_popcnt_epi32(0x8C,
(__m256i)(__v8si){+5, -3, -10, +8, 0, -256, +256, -128}), 0, 0, 30, 1, 0, 0, 0,
25));
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
