https://github.com/TianYe717 updated
https://github.com/llvm/llvm-project/pull/153363
>From 614af4a5e03568205e52e01007f5c84b65c520ff Mon Sep 17 00:00:00 2001
From: Ye Tian <939808...@qq.com>
Date: Wed, 13 Aug 2025 15:26:48 +0800
Subject: [PATCH] [Headers][X86] Allow AVX2/AVX512 broadcast intrinsics to be
used in constexpr
---
clang/lib/Headers/avx2intrin.h | 65 ++++++++------------
clang/lib/Headers/avx512bwintrin.h | 10 ++-
clang/lib/Headers/avx512dqintrin.h | 31 ++++------
clang/lib/Headers/avx512fintrin.h | 40 +++++-------
clang/lib/Headers/avx512vldqintrin.h | 35 ++++++-----
clang/lib/Headers/avx512vlintrin.h | 20 ++++--
clang/test/CodeGen/X86/avx2-builtins.c | 13 ++++
clang/test/CodeGen/X86/avx512bw-builtins.c | 2 +
clang/test/CodeGen/X86/avx512dq-builtins.c | 6 ++
clang/test/CodeGen/X86/avx512f-builtins.c | 8 +++
clang/test/CodeGen/X86/avx512vl-builtins.c | 3 +
clang/test/CodeGen/X86/avx512vldq-builtins.c | 6 ++
12 files changed, 130 insertions(+), 109 deletions(-)
diff --git a/clang/lib/Headers/avx2intrin.h b/clang/lib/Headers/avx2intrin.h
index 48b3cfacc0bb7..0d627488b1a07 100644
--- a/clang/lib/Headers/avx2intrin.h
+++ b/clang/lib/Headers/avx2intrin.h
@@ -2989,9 +2989,8 @@ _mm256_stream_load_si256(const void *__V)
/// \param __X
/// A 128-bit vector of [4 x float] whose low element will be broadcast.
/// \returns A 128-bit vector of [4 x float] containing the result.
-static __inline__ __m128 __DEFAULT_FN_ATTRS128
-_mm_broadcastss_ps(__m128 __X)
-{
+static __inline__ __m128 __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_broadcastss_ps(__m128 __X) {
return (__m128)__builtin_shufflevector((__v4sf)__X, (__v4sf)__X, 0, 0, 0, 0);
}
@@ -3006,9 +3005,8 @@ _mm_broadcastss_ps(__m128 __X)
/// \param __a
/// A 128-bit vector of [2 x double] whose low element will be broadcast.
/// \returns A 128-bit vector of [2 x double] containing the result.
-static __inline__ __m128d __DEFAULT_FN_ATTRS128
-_mm_broadcastsd_pd(__m128d __a)
-{
+static __inline__ __m128d __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_broadcastsd_pd(__m128d __a) {
return __builtin_shufflevector((__v2df)__a, (__v2df)__a, 0, 0);
}
@@ -3023,9 +3021,8 @@ _mm_broadcastsd_pd(__m128d __a)
/// \param __X
/// A 128-bit vector of [4 x float] whose low element will be broadcast.
/// \returns A 256-bit vector of [8 x float] containing the result.
-static __inline__ __m256 __DEFAULT_FN_ATTRS256
-_mm256_broadcastss_ps(__m128 __X)
-{
+static __inline__ __m256 __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_broadcastss_ps(__m128 __X) {
return (__m256)__builtin_shufflevector((__v4sf)__X, (__v4sf)__X, 0, 0, 0, 0,
0, 0, 0, 0);
}
@@ -3040,9 +3037,8 @@ _mm256_broadcastss_ps(__m128 __X)
/// \param __X
/// A 128-bit vector of [2 x double] whose low element will be broadcast.
/// \returns A 256-bit vector of [4 x double] containing the result.
-static __inline__ __m256d __DEFAULT_FN_ATTRS256
-_mm256_broadcastsd_pd(__m128d __X)
-{
+static __inline__ __m256d __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_broadcastsd_pd(__m128d __X) {
return (__m256d)__builtin_shufflevector((__v2df)__X, (__v2df)__X, 0, 0, 0,
0);
}
@@ -3056,9 +3052,8 @@ _mm256_broadcastsd_pd(__m128d __X)
/// \param __X
/// A 128-bit integer vector to be broadcast.
/// \returns A 256-bit integer vector containing the result.
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_broadcastsi128_si256(__m128i __X)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_broadcastsi128_si256(__m128i __X) {
return (__m256i)__builtin_shufflevector((__v2di)__X, (__v2di)__X, 0, 1, 0,
1);
}
@@ -3148,9 +3143,8 @@ _mm256_broadcastsi128_si256(__m128i __X)
/// \param __X
/// A 128-bit integer vector whose low byte will be broadcast.
/// \returns A 256-bit integer vector containing the result.
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_broadcastb_epi8(__m128i __X)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_broadcastb_epi8(__m128i __X) {
return (__m256i)__builtin_shufflevector((__v16qi)__X, (__v16qi)__X, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0);
}
@@ -3164,9 +3158,8 @@ _mm256_broadcastb_epi8(__m128i __X)
/// \param __X
/// A 128-bit vector of [8 x i16] whose low element will be broadcast.
/// \returns A 256-bit vector of [16 x i16] containing the result.
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_broadcastw_epi16(__m128i __X)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_broadcastw_epi16(__m128i __X) {
return (__m256i)__builtin_shufflevector((__v8hi)__X, (__v8hi)__X, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
}
@@ -3180,9 +3173,8 @@ _mm256_broadcastw_epi16(__m128i __X)
/// \param __X
/// A 128-bit vector of [4 x i32] whose low element will be broadcast.
/// \returns A 256-bit vector of [8 x i32] containing the result.
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_broadcastd_epi32(__m128i __X)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_broadcastd_epi32(__m128i __X) {
return (__m256i)__builtin_shufflevector((__v4si)__X, (__v4si)__X, 0, 0, 0,
0, 0, 0, 0, 0);
}
@@ -3196,9 +3188,8 @@ _mm256_broadcastd_epi32(__m128i __X)
/// \param __X
/// A 128-bit vector of [2 x i64] whose low element will be broadcast.
/// \returns A 256-bit vector of [4 x i64] containing the result.
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_broadcastq_epi64(__m128i __X)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_broadcastq_epi64(__m128i __X) {
return (__m256i)__builtin_shufflevector((__v2di)__X, (__v2di)__X, 0, 0, 0,
0);
}
@@ -3212,9 +3203,8 @@ _mm256_broadcastq_epi64(__m128i __X)
/// \param __X
/// A 128-bit integer vector whose low byte will be broadcast.
/// \returns A 128-bit integer vector containing the result.
-static __inline__ __m128i __DEFAULT_FN_ATTRS128
-_mm_broadcastb_epi8(__m128i __X)
-{
+static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_broadcastb_epi8(__m128i __X) {
return (__m128i)__builtin_shufflevector((__v16qi)__X, (__v16qi)__X, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
}
@@ -3228,9 +3218,8 @@ _mm_broadcastb_epi8(__m128i __X)
/// \param __X
/// A 128-bit vector of [8 x i16] whose low element will be broadcast.
/// \returns A 128-bit vector of [8 x i16] containing the result.
-static __inline__ __m128i __DEFAULT_FN_ATTRS128
-_mm_broadcastw_epi16(__m128i __X)
-{
+static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_broadcastw_epi16(__m128i __X) {
return (__m128i)__builtin_shufflevector((__v8hi)__X, (__v8hi)__X, 0, 0, 0,
0, 0, 0, 0, 0);
}
@@ -3244,9 +3233,8 @@ _mm_broadcastw_epi16(__m128i __X)
/// \param __X
/// A 128-bit vector of [4 x i32] whose low element will be broadcast.
/// \returns A 128-bit vector of [4 x i32] containing the result.
-static __inline__ __m128i __DEFAULT_FN_ATTRS128
-_mm_broadcastd_epi32(__m128i __X)
-{
+static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_broadcastd_epi32(__m128i __X) {
return (__m128i)__builtin_shufflevector((__v4si)__X, (__v4si)__X, 0, 0, 0,
0);
}
@@ -3260,9 +3248,8 @@ _mm_broadcastd_epi32(__m128i __X)
/// \param __X
/// A 128-bit vector of [2 x i64] whose low element will be broadcast.
/// \returns A 128-bit vector of [2 x i64] containing the result.
-static __inline__ __m128i __DEFAULT_FN_ATTRS128
-_mm_broadcastq_epi64(__m128i __X)
-{
+static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_broadcastq_epi64(__m128i __X) {
return (__m128i)__builtin_shufflevector((__v2di)__X, (__v2di)__X, 0, 0);
}
diff --git a/clang/lib/Headers/avx512bwintrin.h
b/clang/lib/Headers/avx512bwintrin.h
index ae644b3d4bad2..118e1cbc070e1 100644
--- a/clang/lib/Headers/avx512bwintrin.h
+++ b/clang/lib/Headers/avx512bwintrin.h
@@ -1881,9 +1881,8 @@ _mm512_movm_epi16 (__mmask32 __A)
return (__m512i) __builtin_ia32_cvtmask2w512 (__A);
}
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_broadcastb_epi8 (__m128i __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_broadcastb_epi8(__m128i __A) {
return (__m512i)__builtin_shufflevector((__v16qi) __A, (__v16qi) __A,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0,
@@ -1923,9 +1922,8 @@ _mm512_maskz_set1_epi16 (__mmask32 __M, short __A)
(__v32hi)
_mm512_setzero_si512());
}
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_broadcastw_epi16 (__m128i __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_broadcastw_epi16(__m128i __A) {
return (__m512i)__builtin_shufflevector((__v8hi) __A, (__v8hi) __A,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0);
diff --git a/clang/lib/Headers/avx512dqintrin.h
b/clang/lib/Headers/avx512dqintrin.h
index 62325b96f202b..87d16b474d466 100644
--- a/clang/lib/Headers/avx512dqintrin.h
+++ b/clang/lib/Headers/avx512dqintrin.h
@@ -1084,10 +1084,8 @@ _mm512_movepi64_mask (__m512i __A)
return (__mmask8) __builtin_ia32_cvtq2mask512 ((__v8di) __A);
}
-
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_broadcast_f32x2 (__m128 __A)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_broadcast_f32x2(__m128 __A) {
return (__m512)__builtin_shufflevector((__v4sf)__A, (__v4sf)__A,
0, 1, 0, 1, 0, 1, 0, 1,
0, 1, 0, 1, 0, 1, 0, 1);
@@ -1109,9 +1107,8 @@ _mm512_maskz_broadcast_f32x2 (__mmask16 __M, __m128 __A)
(__v16sf)_mm512_setzero_ps());
}
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_broadcast_f32x8(__m256 __A)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_broadcast_f32x8(__m256 __A) {
return (__m512)__builtin_shufflevector((__v8sf)__A, (__v8sf)__A,
0, 1, 2, 3, 4, 5, 6, 7,
0, 1, 2, 3, 4, 5, 6, 7);
@@ -1133,9 +1130,8 @@ _mm512_maskz_broadcast_f32x8(__mmask16 __M, __m256 __A)
(__v16sf)_mm512_setzero_ps());
}
-static __inline__ __m512d __DEFAULT_FN_ATTRS512
-_mm512_broadcast_f64x2(__m128d __A)
-{
+static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_broadcast_f64x2(__m128d __A) {
return (__m512d)__builtin_shufflevector((__v2df)__A, (__v2df)__A,
0, 1, 0, 1, 0, 1, 0, 1);
}
@@ -1156,9 +1152,8 @@ _mm512_maskz_broadcast_f64x2(__mmask8 __M, __m128d __A)
(__v8df)_mm512_setzero_pd());
}
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_broadcast_i32x2 (__m128i __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_broadcast_i32x2(__m128i __A) {
return (__m512i)__builtin_shufflevector((__v4si)__A, (__v4si)__A,
0, 1, 0, 1, 0, 1, 0, 1,
0, 1, 0, 1, 0, 1, 0, 1);
@@ -1180,9 +1175,8 @@ _mm512_maskz_broadcast_i32x2 (__mmask16 __M, __m128i __A)
(__v16si)_mm512_setzero_si512());
}
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_broadcast_i32x8(__m256i __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_broadcast_i32x8(__m256i __A) {
return (__m512i)__builtin_shufflevector((__v8si)__A, (__v8si)__A,
0, 1, 2, 3, 4, 5, 6, 7,
0, 1, 2, 3, 4, 5, 6, 7);
@@ -1204,9 +1198,8 @@ _mm512_maskz_broadcast_i32x8(__mmask16 __M, __m256i __A)
(__v16si)_mm512_setzero_si512());
}
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_broadcast_i64x2(__m128i __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_broadcast_i64x2(__m128i __A) {
return (__m512i)__builtin_shufflevector((__v2di)__A, (__v2di)__A,
0, 1, 0, 1, 0, 1, 0, 1);
}
diff --git a/clang/lib/Headers/avx512fintrin.h
b/clang/lib/Headers/avx512fintrin.h
index 634fb78696f04..05a291ecbc664 100644
--- a/clang/lib/Headers/avx512fintrin.h
+++ b/clang/lib/Headers/avx512fintrin.h
@@ -218,9 +218,8 @@ _mm512_undefined_epi32(void)
return (__m512i)__builtin_ia32_undef512();
}
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_broadcastd_epi32 (__m128i __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_broadcastd_epi32(__m128i __A) {
return (__m512i)__builtin_shufflevector((__v4si) __A, (__v4si) __A,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0);
}
@@ -241,9 +240,8 @@ _mm512_maskz_broadcastd_epi32 (__mmask16 __M, __m128i __A)
(__v16si) _mm512_setzero_si512());
}
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_broadcastq_epi64 (__m128i __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_broadcastq_epi64(__m128i __A) {
return (__m512i)__builtin_shufflevector((__v2di) __A, (__v2di) __A,
0, 0, 0, 0, 0, 0, 0, 0);
}
@@ -344,9 +342,8 @@ _mm512_maskz_set1_epi64(__mmask8 __M, long long __A)
(__v8di)_mm512_setzero_si512());
}
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_broadcastss_ps(__m128 __A)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_broadcastss_ps(__m128 __A) {
return (__m512)__builtin_shufflevector((__v4sf) __A, (__v4sf) __A,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0);
}
@@ -389,9 +386,8 @@ _mm512_set4_ps(float __A, float __B, float __C, float __D) {
#define _mm512_setr4_ps(e0,e1,e2,e3) \
_mm512_set4_ps((e3),(e2),(e1),(e0))
-static __inline__ __m512d __DEFAULT_FN_ATTRS512
-_mm512_broadcastsd_pd(__m128d __A)
-{
+static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_broadcastsd_pd(__m128d __A) {
return (__m512d)__builtin_shufflevector((__v2df) __A, (__v2df) __A,
0, 0, 0, 0, 0, 0, 0, 0);
}
@@ -6795,9 +6791,8 @@ _mm_maskz_sqrt_ss (__mmask8 __U, __m128 __A, __m128 __B)
(__v4sf)_mm_setzero_ps(), \
(__mmask8)(U), (int)(R)))
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_broadcast_f32x4(__m128 __A)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_broadcast_f32x4(__m128 __A) {
return (__m512)__builtin_shufflevector((__v4sf)__A, (__v4sf)__A,
0, 1, 2, 3, 0, 1, 2, 3,
0, 1, 2, 3, 0, 1, 2, 3);
@@ -6819,9 +6814,8 @@ _mm512_maskz_broadcast_f32x4(__mmask16 __M, __m128 __A)
(__v16sf)_mm512_setzero_ps());
}
-static __inline__ __m512d __DEFAULT_FN_ATTRS512
-_mm512_broadcast_f64x4(__m256d __A)
-{
+static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_broadcast_f64x4(__m256d __A) {
return (__m512d)__builtin_shufflevector((__v4df)__A, (__v4df)__A,
0, 1, 2, 3, 0, 1, 2, 3);
}
@@ -6842,9 +6836,8 @@ _mm512_maskz_broadcast_f64x4(__mmask8 __M, __m256d __A)
(__v8df)_mm512_setzero_pd());
}
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_broadcast_i32x4(__m128i __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_broadcast_i32x4(__m128i __A) {
return (__m512i)__builtin_shufflevector((__v4si)__A, (__v4si)__A,
0, 1, 2, 3, 0, 1, 2, 3,
0, 1, 2, 3, 0, 1, 2, 3);
@@ -6866,9 +6859,8 @@ _mm512_maskz_broadcast_i32x4(__mmask16 __M, __m128i __A)
(__v16si)_mm512_setzero_si512());
}
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_broadcast_i64x4(__m256i __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_broadcast_i64x4(__m256i __A) {
return (__m512i)__builtin_shufflevector((__v4di)__A, (__v4di)__A,
0, 1, 2, 3, 0, 1, 2, 3);
}
diff --git a/clang/lib/Headers/avx512vldqintrin.h
b/clang/lib/Headers/avx512vldqintrin.h
index 272cdd89e2d24..ceebd09e1d3af 100644
--- a/clang/lib/Headers/avx512vldqintrin.h
+++ b/clang/lib/Headers/avx512vldqintrin.h
@@ -24,6 +24,14 @@
__target__("avx512vl,avx512dq,no-evex512"),
\
__min_vector_width__(256)))
+#if defined(__cplusplus) && (__cplusplus >= 201103L)
+#define __DEFAULT_FN_ATTRS256_CONSTEXPR __DEFAULT_FN_ATTRS256 constexpr
+#define __DEFAULT_FN_ATTRS128_CONSTEXPR __DEFAULT_FN_ATTRS128 constexpr
+#else
+#define __DEFAULT_FN_ATTRS256_CONSTEXPR __DEFAULT_FN_ATTRS256
+#define __DEFAULT_FN_ATTRS128_CONSTEXPR __DEFAULT_FN_ATTRS128
+#endif
+
static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_mullo_epi64 (__m256i __A, __m256i __B) {
return (__m256i) ((__v4du) __A * (__v4du) __B);
@@ -956,9 +964,8 @@ _mm256_movepi64_mask (__m256i __A)
return (__mmask8) __builtin_ia32_cvtq2mask256 ((__v4di) __A);
}
-static __inline__ __m256 __DEFAULT_FN_ATTRS256
-_mm256_broadcast_f32x2 (__m128 __A)
-{
+static __inline__ __m256 __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_broadcast_f32x2(__m128 __A) {
return (__m256)__builtin_shufflevector((__v4sf)__A, (__v4sf)__A,
0, 1, 0, 1, 0, 1, 0, 1);
}
@@ -979,9 +986,8 @@ _mm256_maskz_broadcast_f32x2 (__mmask8 __M, __m128 __A)
(__v8sf)_mm256_setzero_ps());
}
-static __inline__ __m256d __DEFAULT_FN_ATTRS256
-_mm256_broadcast_f64x2(__m128d __A)
-{
+static __inline__ __m256d __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_broadcast_f64x2(__m128d __A) {
return (__m256d)__builtin_shufflevector((__v2df)__A, (__v2df)__A,
0, 1, 0, 1);
}
@@ -1002,9 +1008,8 @@ _mm256_maskz_broadcast_f64x2 (__mmask8 __M, __m128d __A)
(__v4df)_mm256_setzero_pd());
}
-static __inline__ __m128i __DEFAULT_FN_ATTRS128
-_mm_broadcast_i32x2 (__m128i __A)
-{
+static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_broadcast_i32x2(__m128i __A) {
return (__m128i)__builtin_shufflevector((__v4si)__A, (__v4si)__A,
0, 1, 0, 1);
}
@@ -1025,9 +1030,8 @@ _mm_maskz_broadcast_i32x2 (__mmask8 __M, __m128i __A)
(__v4si)_mm_setzero_si128());
}
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_broadcast_i32x2 (__m128i __A)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_broadcast_i32x2(__m128i __A) {
return (__m256i)__builtin_shufflevector((__v4si)__A, (__v4si)__A,
0, 1, 0, 1, 0, 1, 0, 1);
}
@@ -1048,9 +1052,8 @@ _mm256_maskz_broadcast_i32x2 (__mmask8 __M, __m128i __A)
(__v8si)_mm256_setzero_si256());
}
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_broadcast_i64x2(__m128i __A)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_broadcast_i64x2(__m128i __A) {
return (__m256i)__builtin_shufflevector((__v2di)__A, (__v2di)__A,
0, 1, 0, 1);
}
@@ -1169,5 +1172,7 @@ _mm256_maskz_broadcast_i64x2 (__mmask8 __M, __m128i __A)
#undef __DEFAULT_FN_ATTRS128
#undef __DEFAULT_FN_ATTRS256
+#undef __DEFAULT_FN_ATTRS256_CONSTEXPR
+#undef __DEFAULT_FN_ATTRS128_CONSTEXPR
#endif
diff --git a/clang/lib/Headers/avx512vlintrin.h
b/clang/lib/Headers/avx512vlintrin.h
index 366adab1fab32..09b76d46efafe 100644
--- a/clang/lib/Headers/avx512vlintrin.h
+++ b/clang/lib/Headers/avx512vlintrin.h
@@ -23,6 +23,14 @@
__target__("avx512vl,no-evex512"),
\
__min_vector_width__(256)))
+#if defined(__cplusplus) && (__cplusplus >= 201103L)
+#define __DEFAULT_FN_ATTRS256_CONSTEXPR __DEFAULT_FN_ATTRS256 constexpr
+#define __DEFAULT_FN_ATTRS128_CONSTEXPR __DEFAULT_FN_ATTRS128 constexpr
+#else
+#define __DEFAULT_FN_ATTRS256_CONSTEXPR __DEFAULT_FN_ATTRS256
+#define __DEFAULT_FN_ATTRS128_CONSTEXPR __DEFAULT_FN_ATTRS128
+#endif
+
typedef short __v2hi __attribute__((__vector_size__(4)));
typedef char __v4qi __attribute__((__vector_size__(4)));
typedef char __v2qi __attribute__((__vector_size__(2)));
@@ -6744,9 +6752,8 @@ _mm256_maskz_rsqrt14_ps (__mmask8 __U, __m256 __A)
(__mmask8) __U);
}
-static __inline__ __m256 __DEFAULT_FN_ATTRS256
-_mm256_broadcast_f32x4(__m128 __A)
-{
+static __inline__ __m256 __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_broadcast_f32x4(__m128 __A) {
return (__m256)__builtin_shufflevector((__v4sf)__A, (__v4sf)__A,
0, 1, 2, 3, 0, 1, 2, 3);
}
@@ -6767,9 +6774,8 @@ _mm256_maskz_broadcast_f32x4 (__mmask8 __M, __m128 __A)
(__v8sf)_mm256_setzero_ps());
}
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_broadcast_i32x4(__m128i __A)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_broadcast_i32x4(__m128i __A) {
return (__m256i)__builtin_shufflevector((__v4si)__A, (__v4si)__A,
0, 1, 2, 3, 0, 1, 2, 3);
}
@@ -8385,5 +8391,7 @@ _mm256_maskz_cvtph_ps (__mmask8 __U, __m128i __A)
#undef __DEFAULT_FN_ATTRS128
#undef __DEFAULT_FN_ATTRS256
+#undef __DEFAULT_FN_ATTRS256_CONSTEXPR
+#undef __DEFAULT_FN_ATTRS128_CONSTEXPR
#endif /* __AVX512VLINTRIN_H */
diff --git a/clang/test/CodeGen/X86/avx2-builtins.c
b/clang/test/CodeGen/X86/avx2-builtins.c
index 8e591397f17c9..e7f81068ab4f4 100644
--- a/clang/test/CodeGen/X86/avx2-builtins.c
+++ b/clang/test/CodeGen/X86/avx2-builtins.c
@@ -160,6 +160,7 @@ __m128i test_mm_broadcastb_epi8(__m128i a) {
// CHECK: shufflevector <16 x i8> %{{.*}}, <16 x i8> %{{.*}}, <16 x i32>
zeroinitializer
return _mm_broadcastb_epi8(a);
}
+TEST_CONSTEXPR(match_v16qi(_mm_broadcastb_epi8((__m128i)(__v16qi){42, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}), 42, 42, 42, 42, 42, 42, 42, 42, 42, 42,
42, 42, 42, 42, 42, 42));
__m256i test_mm256_broadcastb_epi8(__m128i a) {
// CHECK-LABEL: test_mm256_broadcastb_epi8
@@ -167,6 +168,7 @@ __m256i test_mm256_broadcastb_epi8(__m128i a) {
// CHECK: shufflevector <16 x i8> %{{.*}}, <16 x i8> %{{.*}}, <32 x i32>
zeroinitializer
return _mm256_broadcastb_epi8(a);
}
+TEST_CONSTEXPR(match_v32qi(_mm256_broadcastb_epi8((__m128i)(__v16qi){42, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}), 42, 42, 42, 42, 42, 42, 42, 42, 42,
42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42,
42, 42, 42));
__m128i test_mm_broadcastd_epi32(__m128i a) {
// CHECK-LABEL: test_mm_broadcastd_epi32
@@ -174,6 +176,7 @@ __m128i test_mm_broadcastd_epi32(__m128i a) {
// CHECK: shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32>
zeroinitializer
return _mm_broadcastd_epi32(a);
}
+TEST_CONSTEXPR(match_v4si(_mm_broadcastd_epi32((__m128i)(__v4si){-42, 0, 0,
0}), -42, -42, -42, -42));
__m256i test_mm256_broadcastd_epi32(__m128i a) {
// CHECK-LABEL: test_mm256_broadcastd_epi32
@@ -181,6 +184,7 @@ __m256i test_mm256_broadcastd_epi32(__m128i a) {
// CHECK: shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <8 x i32>
zeroinitializer
return _mm256_broadcastd_epi32(a);
}
+TEST_CONSTEXPR(match_v8si(_mm256_broadcastd_epi32((__m128i)(__v4si){-42, 0, 0,
0}), -42, -42, -42, -42, -42, -42, -42, -42));
__m128i test_mm_broadcastq_epi64(__m128i a) {
// CHECK-LABEL: test_mm_broadcastq_epi64
@@ -188,6 +192,7 @@ __m128i test_mm_broadcastq_epi64(__m128i a) {
// CHECK: shufflevector <2 x i64> %{{.*}}, <2 x i64> %{{.*}}, <2 x i32>
zeroinitializer
return _mm_broadcastq_epi64(a);
}
+TEST_CONSTEXPR(match_v2di(_mm_broadcastq_epi64((__m128i)(__v2di){-42, 0}),
-42, -42));
__m256i test_mm256_broadcastq_epi64(__m128i a) {
// CHECK-LABEL: test_mm256_broadcastq_epi64
@@ -195,12 +200,14 @@ __m256i test_mm256_broadcastq_epi64(__m128i a) {
// CHECK: shufflevector <2 x i64> %{{.*}}, <2 x i64> %{{.*}}, <4 x i32>
zeroinitializer
return _mm256_broadcastq_epi64(a);
}
+TEST_CONSTEXPR(match_v4di(_mm256_broadcastq_epi64((__m128i)(__v2di){-42, 0}),
-42, -42, -42, -42));
__m128d test_mm_broadcastsd_pd(__m128d a) {
// CHECK-LABEL: test_mm_broadcastsd_pd
// CHECK: shufflevector <2 x double> %{{.*}}, <2 x double> %{{.*}}, <2 x
i32> zeroinitializer
return _mm_broadcastsd_pd(a);
}
+TEST_CONSTEXPR(match_m128d(_mm_broadcastsd_pd((__m128d){+7.0, -7.0}), +7.0,
+7.0));
__m256d test_mm256_broadcastsd_pd(__m128d a) {
// CHECK-LABEL: test_mm256_broadcastsd_pd
@@ -208,12 +215,14 @@ __m256d test_mm256_broadcastsd_pd(__m128d a) {
// CHECK: shufflevector <2 x double> %{{.*}}, <2 x double> %{{.*}}, <4 x
i32> zeroinitializer
return _mm256_broadcastsd_pd(a);
}
+TEST_CONSTEXPR(match_m256d(_mm256_broadcastsd_pd((__m128d){+7.0, -7.0}), +7.0,
+7.0, +7.0, +7.0));
__m256i test_mm256_broadcastsi128_si256(__m128i a) {
// CHECK-LABEL: test_mm256_broadcastsi128_si256
// CHECK: shufflevector <2 x i64> %{{.*}}, <2 x i64> %{{.*}}, <4 x i32> <i32
0, i32 1, i32 0, i32 1>
return _mm256_broadcastsi128_si256(a);
}
+TEST_CONSTEXPR(match_m256i(_mm256_broadcastsi128_si256((__m128i)(__v2di){3,
45}), 3, 45, 3, 45));
__m256i test_mm_broadcastsi128_si256(__m128i a) {
// CHECK-LABEL: test_mm_broadcastsi128_si256
@@ -227,6 +236,7 @@ __m128 test_mm_broadcastss_ps(__m128 a) {
// CHECK: shufflevector <4 x float> %{{.*}}, <4 x float> %{{.*}}, <4 x i32>
zeroinitializer
return _mm_broadcastss_ps(a);
}
+TEST_CONSTEXPR(match_m128(_mm_broadcastss_ps((__m128){-4.0f, +5.0f, +6.0f,
+7.0f}), -4.0f, -4.0f, -4.0f, -4.0f));
__m256 test_mm256_broadcastss_ps(__m128 a) {
// CHECK-LABEL: test_mm256_broadcastss_ps
@@ -234,6 +244,7 @@ __m256 test_mm256_broadcastss_ps(__m128 a) {
// CHECK: shufflevector <4 x float> %{{.*}}, <4 x float> %{{.*}}, <8 x i32>
zeroinitializer
return _mm256_broadcastss_ps(a);
}
+TEST_CONSTEXPR(match_m256(_mm256_broadcastss_ps((__m128){-4.0f, +5.0f, +6.0f,
+7.0f}), -4.0f, -4.0f, -4.0f, -4.0f, -4.0f, -4.0f, -4.0f, -4.0f));
__m128i test_mm_broadcastw_epi16(__m128i a) {
// CHECK-LABEL: test_mm_broadcastw_epi16
@@ -241,6 +252,7 @@ __m128i test_mm_broadcastw_epi16(__m128i a) {
// CHECK: shufflevector <8 x i16> %{{.*}}, <8 x i16> %{{.*}}, <8 x i32>
zeroinitializer
return _mm_broadcastw_epi16(a);
}
+TEST_CONSTEXPR(match_v8hi(_mm_broadcastw_epi16((__m128i)(__v8hi){42, 0, 0, 0,
0, 0, 0, 0}), 42, 42, 42, 42, 42, 42, 42, 42));
__m256i test_mm256_broadcastw_epi16(__m128i a) {
// CHECK-LABEL: test_mm256_broadcastw_epi16
@@ -248,6 +260,7 @@ __m256i test_mm256_broadcastw_epi16(__m128i a) {
// CHECK: shufflevector <8 x i16> %{{.*}}, <8 x i16> %{{.*}}, <16 x i32>
zeroinitializer
return _mm256_broadcastw_epi16(a);
}
+TEST_CONSTEXPR(match_v16hi(_mm256_broadcastw_epi16((__m128i)(__v8hi){42, 0, 0,
0, 0, 0, 0, 0}), 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42,
42));
__m256i test_mm256_bslli_epi128(__m256i a) {
// CHECK-LABEL: test_mm256_bslli_epi128
diff --git a/clang/test/CodeGen/X86/avx512bw-builtins.c
b/clang/test/CodeGen/X86/avx512bw-builtins.c
index e91a3b5779786..0bd9718974c7d 100644
--- a/clang/test/CodeGen/X86/avx512bw-builtins.c
+++ b/clang/test/CodeGen/X86/avx512bw-builtins.c
@@ -2223,6 +2223,7 @@ __m512i test_mm512_broadcastb_epi8(__m128i __A) {
// CHECK: shufflevector <16 x i8> %{{.*}}, <16 x i8> %{{.*}}, <64 x i32>
zeroinitializer
return _mm512_broadcastb_epi8(__A);
}
+TEST_CONSTEXPR(match_v64qi(_mm512_broadcastb_epi8((__m128i)(__v16qi){42, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}), 42, 42, 42, 42, 42, 42, 42, 42, 42,
42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42,
42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42,
42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42));
__m512i test_mm512_mask_broadcastb_epi8(__m512i __O, __mmask64 __M, __m128i
__A) {
// CHECK-LABEL: test_mm512_mask_broadcastb_epi8
@@ -2243,6 +2244,7 @@ __m512i test_mm512_broadcastw_epi16(__m128i __A) {
// CHECK: shufflevector <8 x i16> %{{.*}}, <8 x i16> %{{.*}}, <32 x i32>
zeroinitializer
return _mm512_broadcastw_epi16(__A);
}
+TEST_CONSTEXPR(match_v32hi(_mm512_broadcastw_epi16((__m128i)(__v8hi){42, 3,
10, 8, 0, 256, 256, 128}), 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42,
42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42));
__m512i test_mm512_mask_broadcastw_epi16(__m512i __O, __mmask32 __M, __m128i
__A) {
// CHECK-LABEL: test_mm512_mask_broadcastw_epi16
diff --git a/clang/test/CodeGen/X86/avx512dq-builtins.c
b/clang/test/CodeGen/X86/avx512dq-builtins.c
index ad5a940686533..47e5a3ffeeff4 100644
--- a/clang/test/CodeGen/X86/avx512dq-builtins.c
+++ b/clang/test/CodeGen/X86/avx512dq-builtins.c
@@ -1211,6 +1211,7 @@ __m512 test_mm512_broadcast_f32x2(__m128 __A) {
// CHECK: shufflevector <4 x float> %{{.*}}, <4 x float> %{{.*}}, <16 x i32>
<i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0,
i32 1, i32 0, i32 1, i32 0, i32 1>
return _mm512_broadcast_f32x2(__A);
}
+TEST_CONSTEXPR(match_m512(_mm512_broadcast_f32x2((__m128)(__v4sf){1.0, -2.0,
3.0, -4.0}), 1.0, -2.0, 1.0, -2.0, 1.0, -2.0, 1.0, -2.0, 1.0, -2.0, 1.0, -2.0,
1.0, -2.0, 1.0, -2.0));
__m512 test_mm512_mask_broadcast_f32x2(__m512 __O, __mmask16 __M, __m128 __A) {
// CHECK-LABEL: test_mm512_mask_broadcast_f32x2
@@ -1231,6 +1232,7 @@ __m512 test_mm512_broadcast_f32x8(float const* __A) {
// CHECK: shufflevector <8 x float> %{{.*}}, <8 x float> %{{.*}}, <16 x i32>
<i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2,
i32 3, i32 4, i32 5, i32 6, i32 7>
return _mm512_broadcast_f32x8(_mm256_loadu_ps(__A));
}
+TEST_CONSTEXPR(match_m512(_mm512_broadcast_f32x8((__m256)(__v8sf){1.0f, 2.0f,
3.0f, 4.0f, -5.0f, -6.0f, -7.0f, -8.0f}), 1.0f, 2.0f, 3.0f, 4.0f, -5.0f, -6.0f,
-7.0f, -8.0f, 1.0f, 2.0f, 3.0f, 4.0f, -5.0f, -6.0f, -7.0f, -8.0f));
__m512 test_mm512_mask_broadcast_f32x8(__m512 __O, __mmask16 __M, float const*
__A) {
// CHECK-LABEL: test_mm512_mask_broadcast_f32x8
@@ -1251,6 +1253,7 @@ __m512d test_mm512_broadcast_f64x2(double const* __A) {
// CHECK: shufflevector <2 x double> %{{.*}}, <2 x double> %{{.*}}, <8 x
i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
return _mm512_broadcast_f64x2(_mm_loadu_pd(__A));
}
+TEST_CONSTEXPR(match_m512d(_mm512_broadcast_f64x2((__m128d)(__v2df){1.0,
-2.0}), 1.0, -2.0, 1.0, -2.0, 1.0, -2.0, 1.0, -2.0));
__m512d test_mm512_mask_broadcast_f64x2(__m512d __O, __mmask8 __M, double
const* __A) {
// CHECK-LABEL: test_mm512_mask_broadcast_f64x2
@@ -1271,6 +1274,7 @@ __m512i test_mm512_broadcast_i32x2(__m128i __A) {
// CHECK: shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <16 x i32>
<i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0,
i32 1, i32 0, i32 1, i32 0, i32 1>
return _mm512_broadcast_i32x2(__A);
}
+TEST_CONSTEXPR(match_v16si(_mm512_broadcast_i32x2((__m128i)(__v4si){1, -2, 3,
-4}), 1, -2, 1, -2, 1, -2, 1, -2, 1, -2, 1, -2, 1, -2, 1, -2));
__m512i test_mm512_mask_broadcast_i32x2(__m512i __O, __mmask16 __M, __m128i
__A) {
// CHECK-LABEL: test_mm512_mask_broadcast_i32x2
@@ -1291,6 +1295,7 @@ __m512i test_mm512_broadcast_i32x8(__m256i const* __A) {
// CHECK: shufflevector <8 x i32> %{{.*}}, <8 x i32> %{{.*}}, <16 x i32>
<i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2,
i32 3, i32 4, i32 5, i32 6, i32 7>
return _mm512_broadcast_i32x8(_mm256_loadu_si256(__A));
}
+TEST_CONSTEXPR(match_v16si(_mm512_broadcast_i32x8((__m256i)(__v8si){1, 2, 3,
4, -5, -6, -7, -8}), 1, 2, 3, 4, -5, -6, -7, -8, 1, 2, 3, 4, -5, -6, -7, -8));
__m512i test_mm512_mask_broadcast_i32x8(__m512i __O, __mmask16 __M, __m256i
const* __A) {
// CHECK-LABEL: test_mm512_mask_broadcast_i32x8
@@ -1311,6 +1316,7 @@ __m512i test_mm512_broadcast_i64x2(__m128i const* __A) {
// CHECK: shufflevector <2 x i64> %{{.*}}, <2 x i64> %{{.*}}, <8 x i32> <i32
0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
return _mm512_broadcast_i64x2(_mm_loadu_si128(__A));
}
+TEST_CONSTEXPR(match_v8di(_mm512_broadcast_i64x2((__m128i)(__v2di){1, -2}), 1,
-2, 1, -2, 1, -2, 1, -2));
__m512i test_mm512_mask_broadcast_i64x2(__m512i __O, __mmask8 __M, __m128i
const* __A) {
// CHECK-LABEL: test_mm512_mask_broadcast_i64x2
diff --git a/clang/test/CodeGen/X86/avx512f-builtins.c
b/clang/test/CodeGen/X86/avx512f-builtins.c
index a92227d5aaae1..b8eadc42c78f3 100644
--- a/clang/test/CodeGen/X86/avx512f-builtins.c
+++ b/clang/test/CodeGen/X86/avx512f-builtins.c
@@ -6311,6 +6311,7 @@ __m512 test_mm512_broadcast_f32x4(float const* __A) {
// CHECK: shufflevector <4 x float> %{{.*}}, <4 x float> %{{.*}}, <16 x i32>
<i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2,
i32 3, i32 0, i32 1, i32 2, i32 3>
return _mm512_broadcast_f32x4(_mm_loadu_ps(__A));
}
+TEST_CONSTEXPR(match_m512(_mm512_broadcast_f32x4((__m128)(__v4sf){1.0f, 2.0f,
-3.0f, -4.0f}), 1.0f, 2.0f, -3.0f, -4.0f, 1.0f, 2.0f, -3.0f, -4.0f, 1.0f, 2.0f,
-3.0f, -4.0f, 1.0f, 2.0f, -3.0f, -4.0f));
__m512 test_mm512_mask_broadcast_f32x4(__m512 __O, __mmask16 __M, float const*
__A) {
// CHECK-LABEL: test_mm512_mask_broadcast_f32x4
@@ -6331,6 +6332,7 @@ __m512d test_mm512_broadcast_f64x4(double const* __A) {
// CHECK: shufflevector <4 x double> %{{.*}}, <4 x double> %{{.*}}, <8 x
i32> <i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3>
return _mm512_broadcast_f64x4(_mm256_loadu_pd(__A));
}
+TEST_CONSTEXPR(match_m512d(_mm512_broadcast_f64x4((__m256d)(__v4df){1.0, 2.0,
-3.0, -4.0}), 1.0, 2.0, -3.0, -4.0, 1.0, 2.0, -3.0, -4.0));
__m512d test_mm512_mask_broadcast_f64x4(__m512d __O, __mmask8 __M, double
const* __A) {
// CHECK-LABEL: test_mm512_mask_broadcast_f64x4
@@ -6351,6 +6353,7 @@ __m512i test_mm512_broadcast_i32x4(__m128i const* __A) {
// CHECK: shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <16 x i32>
<i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2,
i32 3, i32 0, i32 1, i32 2, i32 3>
return _mm512_broadcast_i32x4(_mm_loadu_si128(__A));
}
+TEST_CONSTEXPR(match_v16si(_mm512_broadcast_i32x4((__m128i)(__v4si){1, 2, -3,
-4}), 1, 2, -3, -4, 1, 2, -3, -4, 1, 2, -3, -4, 1, 2, -3, -4));
__m512i test_mm512_mask_broadcast_i32x4(__m512i __O, __mmask16 __M, __m128i
const* __A) {
// CHECK-LABEL: test_mm512_mask_broadcast_i32x4
@@ -6371,6 +6374,7 @@ __m512i test_mm512_broadcast_i64x4(__m256i const* __A) {
// CHECK: shufflevector <4 x i64> %{{.*}}, <4 x i64> %{{.*}}, <8 x i32> <i32
0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3>
return _mm512_broadcast_i64x4(_mm256_loadu_si256(__A));
}
+TEST_CONSTEXPR(match_v8di(_mm512_broadcast_i64x4((__m256i)(__v4di){1, 2, -3,
-4}), 1, 2, -3, -4, 1, 2, -3, -4));
__m512i test_mm512_mask_broadcast_i64x4(__m512i __O, __mmask8 __M, __m256i
const* __A) {
// CHECK-LABEL: test_mm512_mask_broadcast_i64x4
@@ -6391,6 +6395,7 @@ __m512d test_mm512_broadcastsd_pd(__m128d __A) {
// CHECK: shufflevector <2 x double> %{{.*}}, <2 x double> %{{.*}}, <8 x
i32> zeroinitializer
return _mm512_broadcastsd_pd(__A);
}
+TEST_CONSTEXPR(match_m512d(_mm512_broadcastsd_pd((__m128d)(__v2df){1.0, 2.0}),
1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0));
__m512d test_mm512_mask_broadcastsd_pd(__m512d __O, __mmask8 __M, __m128d __A)
{
// CHECK-LABEL: test_mm512_mask_broadcastsd_pd
@@ -6411,6 +6416,7 @@ __m512 test_mm512_broadcastss_ps(__m128 __A) {
// CHECK: shufflevector <4 x float> %{{.*}}, <4 x float> %{{.*}}, <16 x i32>
zeroinitializer
return _mm512_broadcastss_ps(__A);
}
+TEST_CONSTEXPR(match_m512(_mm512_broadcastss_ps((__m128)(__v4sf){1.0f, 2.0f,
-3.0f, -4.0f}), 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f,
1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f));
__m512 test_mm512_mask_broadcastss_ps(__m512 __O, __mmask16 __M, __m128 __A) {
// CHECK-LABEL: test_mm512_mask_broadcastss_ps
@@ -6431,6 +6437,7 @@ __m512i test_mm512_broadcastd_epi32(__m128i __A) {
// CHECK: shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <16 x i32>
zeroinitializer
return _mm512_broadcastd_epi32(__A);
}
+TEST_CONSTEXPR(match_v16si(_mm512_broadcastd_epi32((__m128i)(__v4si){-42, 0,
0, 0}), -42, -42, -42, -42, -42, -42, -42, -42, -42, -42, -42, -42, -42, -42,
-42, -42));
__m512i test_mm512_mask_broadcastd_epi32(__m512i __O, __mmask16 __M, __m128i
__A) {
// CHECK-LABEL: test_mm512_mask_broadcastd_epi32
@@ -6451,6 +6458,7 @@ __m512i test_mm512_broadcastq_epi64(__m128i __A) {
// CHECK: shufflevector <2 x i64> %{{.*}}, <2 x i64> %{{.*}}, <8 x i32>
zeroinitializer
return _mm512_broadcastq_epi64(__A);
}
+TEST_CONSTEXPR(match_v8di(_mm512_broadcastq_epi64((__m128i)(__v2di){-42, 0}),
-42, -42, -42, -42, -42, -42, -42, -42));
__m512i test_mm512_mask_broadcastq_epi64(__m512i __O, __mmask8 __M, __m128i
__A) {
// CHECK-LABEL: test_mm512_mask_broadcastq_epi64
diff --git a/clang/test/CodeGen/X86/avx512vl-builtins.c
b/clang/test/CodeGen/X86/avx512vl-builtins.c
index ac7aa3eedbb99..f1ff210b0b1da 100644
--- a/clang/test/CodeGen/X86/avx512vl-builtins.c
+++ b/clang/test/CodeGen/X86/avx512vl-builtins.c
@@ -2,6 +2,7 @@
// RUN: %clang_cc1 -x c++ -flax-vector-conversions=none -ffreestanding %s
-triple=x86_64-apple-darwin -target-feature +avx512f -target-feature +avx512vl
-emit-llvm -o - -Wall -Werror -Wsign-conversion | FileCheck %s
#include <immintrin.h>
+#include "builtin_test_helpers.h"
__mmask8 test_mm_cmpeq_epu32_mask(__m128i __a, __m128i __b) {
// CHECK-LABEL: test_mm_cmpeq_epu32_mask
@@ -8121,6 +8122,7 @@ __m256 test_mm256_broadcast_f32x4(__m128 __A) {
// CHECK: shufflevector <4 x float> %{{.*}}, <4 x float> %{{.*}}, <8 x i32>
<i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3>
return _mm256_broadcast_f32x4(__A);
}
+TEST_CONSTEXPR(match_m256(_mm256_broadcast_f32x4((__m128)(__v4sf){1.0f, 3.0f,
-5.0f, -8.0f}), 1.0f, 3.0f, -5.0f, -8.0f, 1.0f, 3.0f, -5.0f, -8.0f));
__m256 test_mm256_mask_broadcast_f32x4(__m256 __O, __mmask8 __M, __m128 __A) {
// CHECK-LABEL: test_mm256_mask_broadcast_f32x4
@@ -8141,6 +8143,7 @@ __m256i test_mm256_broadcast_i32x4(__m128i const* __A) {
// CHECK: shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <8 x i32> <i32
0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3>
return _mm256_broadcast_i32x4(_mm_loadu_si128(__A));
}
+TEST_CONSTEXPR(match_v8si(_mm256_broadcast_i32x4((__m128i)(__v4si){1, 3, -5,
-8}), 1, 3, -5, -8, 1, 3, -5, -8));
__m256i test_mm256_mask_broadcast_i32x4(__m256i __O, __mmask8 __M, __m128i
const* __A) {
// CHECK-LABEL: test_mm256_mask_broadcast_i32x4
diff --git a/clang/test/CodeGen/X86/avx512vldq-builtins.c
b/clang/test/CodeGen/X86/avx512vldq-builtins.c
index fbe780ca5bd42..66ba0c704668c 100644
--- a/clang/test/CodeGen/X86/avx512vldq-builtins.c
+++ b/clang/test/CodeGen/X86/avx512vldq-builtins.c
@@ -5,6 +5,7 @@
#include <immintrin.h>
+#include "builtin_test_helpers.h"
__m256i test_mm256_mullo_epi64 (__m256i __A, __m256i __B) {
// CHECK-LABEL: test_mm256_mullo_epi64
@@ -930,6 +931,7 @@ __m256 test_mm256_broadcast_f32x2(__m128 __A) {
// CHECK: shufflevector <4 x float> %{{.*}}, <4 x float> %{{.*}}, <8 x i32>
<i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
return _mm256_broadcast_f32x2(__A);
}
+TEST_CONSTEXPR(match_m256(_mm256_broadcast_f32x2((__m128)(__v4sf){1.0f, -2.0f,
3.0f, -4.0f}), 1.0f, -2.0f, 1.0f, -2.0f, 1.0f, -2.0f, 1.0f, -2.0f));
__m256 test_mm256_mask_broadcast_f32x2(__m256 __O, __mmask8 __M, __m128 __A) {
// CHECK-LABEL: test_mm256_mask_broadcast_f32x2
@@ -950,6 +952,7 @@ __m256d test_mm256_broadcast_f64x2(double const* __A) {
// CHECK: shufflevector <2 x double> %{{.*}}, <2 x double> %{{.*}}, <4 x
i32> <i32 0, i32 1, i32 0, i32 1>
return _mm256_broadcast_f64x2(_mm_loadu_pd(__A));
}
+TEST_CONSTEXPR(match_m256d(_mm256_broadcast_f64x2((__m128d)(__v2df){1.0,
-2.0}), 1.0, -2.0, 1.0, -2.0));
__m256d test_mm256_mask_broadcast_f64x2(__m256d __O, __mmask8 __M, double
const* __A) {
// CHECK-LABEL: test_mm256_mask_broadcast_f64x2
@@ -970,6 +973,7 @@ __m128i test_mm_broadcast_i32x2(__m128i __A) {
// CHECK: shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32
0, i32 1, i32 0, i32 1>
return _mm_broadcast_i32x2(__A);
}
+TEST_CONSTEXPR(match_v4si(_mm_broadcast_i32x2((__m128i)(__v4si){1, -2, 3,
-4}), 1, -2, 1, -2));
__m128i test_mm_mask_broadcast_i32x2(__m128i __O, __mmask8 __M, __m128i __A) {
// CHECK-LABEL: test_mm_mask_broadcast_i32x2
@@ -990,6 +994,7 @@ __m256i test_mm256_broadcast_i32x2(__m128i __A) {
// CHECK: shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <8 x i32> <i32
0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
return _mm256_broadcast_i32x2(__A);
}
+TEST_CONSTEXPR(match_v8si(_mm256_broadcast_i32x2((__m128i)(__v4si){1, -2, 3,
-4}), 1, -2, 1, -2, 1, -2, 1, -2));
__m256i test_mm256_mask_broadcast_i32x2(__m256i __O, __mmask8 __M, __m128i
__A) {
// CHECK-LABEL: test_mm256_mask_broadcast_i32x2
@@ -1010,6 +1015,7 @@ __m256i test_mm256_broadcast_i64x2(__m128i const* __A) {
// CHECK: shufflevector <2 x i64> %{{.*}}, <2 x i64> %{{.*}}, <4 x i32> <i32
0, i32 1, i32 0, i32 1>
return _mm256_broadcast_i64x2(_mm_loadu_si128(__A));
}
+TEST_CONSTEXPR(match_v4di(_mm256_broadcast_i64x2((__m128i)(__v2di){1, -2}), 1,
-2, 1, -2));
__m256i test_mm256_mask_broadcast_i64x2(__m256i __O, __mmask8 __M, __m128i
const* __A) {
// CHECK-LABEL: test_mm256_mask_broadcast_i64x2
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
