[Headers][X86] Allow AVX512 masked shuffles to be used in constexpr #162301

ahmednour25689 · 2025-10-07T15:15:47Z

This PR makes AVX512 masked shuffle and broadcast intrinsics usable in constexpr

Fixes #162074

…expr

github-actions · 2025-10-07T15:16:11Z

Thank you for submitting a Pull Request (PR) to the LLVM Project!

This PR will be automatically labeled and the relevant teams will be notified.

If you wish to, you can add reviewers by using the "Reviewers" section on this page.

If this is not working for you, it is probably because you do not have write permissions for the repository. In which case you can instead tag reviewers by name in a comment by using @ followed by their GitHub username.

If you have received no comments on your PR for a week, you can request a review by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate is once a week. Please remember that you are asking for valuable time from other developers.

If you have further questions, they may be answered by the LLVM GitHub User Guide.

You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums.

llvmbot · 2025-10-07T15:16:47Z

@llvm/pr-subscribers-backend-x86

@llvm/pr-subscribers-clang

Author: Ahmed Nour (ahmednour25689)

Changes

This PR makes AVX512 masked shuffle and broadcast intrinsics usable in constexpr

Fixes #162074

Patch is 60.24 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/162301.diff

7 Files Affected:

(modified) clang/lib/Headers/avx512dqintrin.h (+24-36)
(modified) clang/lib/Headers/avx512fintrin.h (+48-73)
(modified) clang/lib/Headers/avx512vldqintrin.h (+20-30)
(modified) clang/lib/Headers/avx512vlintrin.h (+146-215)
(modified) clang/test/CodeGen/X86/avx512dq-builtins.c (+16)
(modified) clang/test/CodeGen/X86/avx512f-builtins.c (+25)
(modified) clang/test/CodeGen/X86/avx512vl-builtins.c (+70)

diff --git a/clang/lib/Headers/avx512dqintrin.h b/clang/lib/Headers/avx512dqintrin.h
index fb65bf933b8ad..3681ccab0b179 100644
--- a/clang/lib/Headers/avx512dqintrin.h
+++ b/clang/lib/Headers/avx512dqintrin.h
@@ -1083,17 +1083,15 @@ _mm512_broadcast_f32x2(__m128 __A) {
                                          0, 1, 0, 1, 0, 1, 0, 1);
 }
 
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_mask_broadcast_f32x2 (__m512 __O, __mmask16 __M, __m128 __A)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_broadcast_f32x2(__m512 __O, __mmask16 __M, __m128 __A) {
   return (__m512)__builtin_ia32_selectps_512((__mmask16)__M,
                                              (__v16sf)_mm512_broadcast_f32x2(__A),
                                              (__v16sf)__O);
 }
 
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_maskz_broadcast_f32x2 (__mmask16 __M, __m128 __A)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_broadcast_f32x2(__mmask16 __M, __m128 __A) {
   return (__m512)__builtin_ia32_selectps_512((__mmask16)__M,
                                              (__v16sf)_mm512_broadcast_f32x2(__A),
                                              (__v16sf)_mm512_setzero_ps());
@@ -1106,17 +1104,15 @@ _mm512_broadcast_f32x8(__m256 __A) {
                                          0, 1, 2, 3, 4, 5, 6, 7);
 }
 
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_mask_broadcast_f32x8(__m512 __O, __mmask16 __M, __m256 __A)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_broadcast_f32x8(__m512 __O, __mmask16 __M, __m256 __A) {
   return (__m512)__builtin_ia32_selectps_512((__mmask16)__M,
                                            (__v16sf)_mm512_broadcast_f32x8(__A),
                                            (__v16sf)__O);
 }
 
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_maskz_broadcast_f32x8(__mmask16 __M, __m256 __A)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_broadcast_f32x8(__mmask16 __M, __m256 __A) {
   return (__m512)__builtin_ia32_selectps_512((__mmask16)__M,
                                            (__v16sf)_mm512_broadcast_f32x8(__A),
                                            (__v16sf)_mm512_setzero_ps());
@@ -1128,17 +1124,15 @@ _mm512_broadcast_f64x2(__m128d __A) {
                                           0, 1, 0, 1, 0, 1, 0, 1);
 }
 
-static __inline__ __m512d __DEFAULT_FN_ATTRS512
-_mm512_mask_broadcast_f64x2(__m512d __O, __mmask8 __M, __m128d __A)
-{
+static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_broadcast_f64x2(__m512d __O, __mmask8 __M, __m128d __A) {
   return (__m512d)__builtin_ia32_selectpd_512((__mmask8)__M,
                                             (__v8df)_mm512_broadcast_f64x2(__A),
                                             (__v8df)__O);
 }
 
-static __inline__ __m512d __DEFAULT_FN_ATTRS512
-_mm512_maskz_broadcast_f64x2(__mmask8 __M, __m128d __A)
-{
+static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_broadcast_f64x2(__mmask8 __M, __m128d __A) {
   return (__m512d)__builtin_ia32_selectpd_512((__mmask8)__M,
                                             (__v8df)_mm512_broadcast_f64x2(__A),
                                             (__v8df)_mm512_setzero_pd());
@@ -1151,17 +1145,15 @@ _mm512_broadcast_i32x2(__m128i __A) {
                                           0, 1, 0, 1, 0, 1, 0, 1);
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_mask_broadcast_i32x2 (__m512i __O, __mmask16 __M, __m128i __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_broadcast_i32x2(__m512i __O, __mmask16 __M, __m128i __A) {
   return (__m512i)__builtin_ia32_selectd_512((__mmask16)__M,
                                              (__v16si)_mm512_broadcast_i32x2(__A),
                                              (__v16si)__O);
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_maskz_broadcast_i32x2 (__mmask16 __M, __m128i __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_broadcast_i32x2(__mmask16 __M, __m128i __A) {
   return (__m512i)__builtin_ia32_selectd_512((__mmask16)__M,
                                              (__v16si)_mm512_broadcast_i32x2(__A),
                                              (__v16si)_mm512_setzero_si512());
@@ -1174,17 +1166,15 @@ _mm512_broadcast_i32x8(__m256i __A) {
                                           0, 1, 2, 3, 4, 5, 6, 7);
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_mask_broadcast_i32x8(__m512i __O, __mmask16 __M, __m256i __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_broadcast_i32x8(__m512i __O, __mmask16 __M, __m256i __A) {
   return (__m512i)__builtin_ia32_selectd_512((__mmask16)__M,
                                            (__v16si)_mm512_broadcast_i32x8(__A),
                                            (__v16si)__O);
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_maskz_broadcast_i32x8(__mmask16 __M, __m256i __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_broadcast_i32x8(__mmask16 __M, __m256i __A) {
   return (__m512i)__builtin_ia32_selectd_512((__mmask16)__M,
                                            (__v16si)_mm512_broadcast_i32x8(__A),
                                            (__v16si)_mm512_setzero_si512());
@@ -1196,17 +1186,15 @@ _mm512_broadcast_i64x2(__m128i __A) {
                                           0, 1, 0, 1, 0, 1, 0, 1);
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_mask_broadcast_i64x2(__m512i __O, __mmask8 __M, __m128i __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_broadcast_i64x2(__m512i __O, __mmask8 __M, __m128i __A) {
   return (__m512i)__builtin_ia32_selectq_512((__mmask8)__M,
                                             (__v8di)_mm512_broadcast_i64x2(__A),
                                             (__v8di)__O);
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_maskz_broadcast_i64x2(__mmask8 __M, __m128i __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_broadcast_i64x2(__mmask8 __M, __m128i __A) {
   return (__m512i)__builtin_ia32_selectq_512((__mmask8)__M,
                                             (__v8di)_mm512_broadcast_i64x2(__A),
                                             (__v8di)_mm512_setzero_si512());
diff --git a/clang/lib/Headers/avx512fintrin.h b/clang/lib/Headers/avx512fintrin.h
index 80e58425cdd71..8401647ef9da5 100644
--- a/clang/lib/Headers/avx512fintrin.h
+++ b/clang/lib/Headers/avx512fintrin.h
@@ -225,17 +225,15 @@ _mm512_broadcastd_epi32(__m128i __A) {
                                           0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_mask_broadcastd_epi32 (__m512i __O, __mmask16 __M, __m128i __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_broadcastd_epi32(__m512i __O, __mmask16 __M, __m128i __A) {
   return (__m512i)__builtin_ia32_selectd_512(__M,
                                              (__v16si) _mm512_broadcastd_epi32(__A),
                                              (__v16si) __O);
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_maskz_broadcastd_epi32 (__mmask16 __M, __m128i __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_broadcastd_epi32(__mmask16 __M, __m128i __A) {
   return (__m512i)__builtin_ia32_selectd_512(__M,
                                              (__v16si) _mm512_broadcastd_epi32(__A),
                                              (__v16si) _mm512_setzero_si512());
@@ -247,18 +245,15 @@ _mm512_broadcastq_epi64(__m128i __A) {
                                           0, 0, 0, 0, 0, 0, 0, 0);
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_mask_broadcastq_epi64 (__m512i __O, __mmask8 __M, __m128i __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_broadcastq_epi64(__m512i __O, __mmask8 __M, __m128i __A) {
   return (__m512i)__builtin_ia32_selectq_512(__M,
                                              (__v8di) _mm512_broadcastq_epi64(__A),
                                              (__v8di) __O);
-
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_maskz_broadcastq_epi64 (__mmask8 __M, __m128i __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_broadcastq_epi64(__mmask8 __M, __m128i __A) {
   return (__m512i)__builtin_ia32_selectq_512(__M,
                                              (__v8di) _mm512_broadcastq_epi64(__A),
                                              (__v8di) _mm512_setzero_si512());
@@ -321,9 +316,8 @@ _mm512_set1_epi32(int __s)
     __s, __s, __s, __s, __s, __s, __s, __s };
 }
 
-static __inline __m512i __DEFAULT_FN_ATTRS512
-_mm512_maskz_set1_epi32(__mmask16 __M, int __A)
-{
+static __inline __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_set1_epi32(__mmask16 __M, int __A) {
   return (__m512i)__builtin_ia32_selectd_512(__M,
                                              (__v16si)_mm512_set1_epi32(__A),
                                              (__v16si)_mm512_setzero_si512());
@@ -335,9 +329,8 @@ _mm512_set1_epi64(long long __d)
   return __extension__(__m512i)(__v8di){ __d, __d, __d, __d, __d, __d, __d, __d };
 }
 
-static __inline __m512i __DEFAULT_FN_ATTRS512
-_mm512_maskz_set1_epi64(__mmask8 __M, long long __A)
-{
+static __inline __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_set1_epi64(__mmask8 __M, long long __A) {
   return (__m512i)__builtin_ia32_selectq_512(__M,
                                              (__v8di)_mm512_set1_epi64(__A),
                                              (__v8di)_mm512_setzero_si512());
@@ -6552,17 +6545,15 @@ _mm512_broadcast_f32x4(__m128 __A) {
                                          0, 1, 2, 3, 0, 1, 2, 3);
 }
 
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_mask_broadcast_f32x4(__m512 __O, __mmask16 __M, __m128 __A)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_broadcast_f32x4(__m512 __O, __mmask16 __M, __m128 __A) {
   return (__m512)__builtin_ia32_selectps_512((__mmask16)__M,
                                            (__v16sf)_mm512_broadcast_f32x4(__A),
                                            (__v16sf)__O);
 }
 
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_maskz_broadcast_f32x4(__mmask16 __M, __m128 __A)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_broadcast_f32x4(__mmask16 __M, __m128 __A) {
   return (__m512)__builtin_ia32_selectps_512((__mmask16)__M,
                                            (__v16sf)_mm512_broadcast_f32x4(__A),
                                            (__v16sf)_mm512_setzero_ps());
@@ -6597,17 +6588,15 @@ _mm512_broadcast_i32x4(__m128i __A) {
                                           0, 1, 2, 3, 0, 1, 2, 3);
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_mask_broadcast_i32x4(__m512i __O, __mmask16 __M, __m128i __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_broadcast_i32x4(__m512i __O, __mmask16 __M, __m128i __A) {
   return (__m512i)__builtin_ia32_selectd_512((__mmask16)__M,
                                            (__v16si)_mm512_broadcast_i32x4(__A),
                                            (__v16si)__O);
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_maskz_broadcast_i32x4(__mmask16 __M, __m128i __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_broadcast_i32x4(__mmask16 __M, __m128i __A) {
   return (__m512i)__builtin_ia32_selectd_512((__mmask16)__M,
                                            (__v16si)_mm512_broadcast_i32x4(__A),
                                            (__v16si)_mm512_setzero_si512());
@@ -6635,33 +6624,29 @@ _mm512_maskz_broadcast_i64x4(__mmask8 __M, __m256i __A)
                                             (__v8di)_mm512_setzero_si512());
 }
 
-static __inline__ __m512d __DEFAULT_FN_ATTRS512
-_mm512_mask_broadcastsd_pd (__m512d __O, __mmask8 __M, __m128d __A)
-{
+static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_broadcastsd_pd(__m512d __O, __mmask8 __M, __m128d __A) {
   return (__m512d)__builtin_ia32_selectpd_512(__M,
                                               (__v8df) _mm512_broadcastsd_pd(__A),
                                               (__v8df) __O);
 }
 
-static __inline__ __m512d __DEFAULT_FN_ATTRS512
-_mm512_maskz_broadcastsd_pd (__mmask8 __M, __m128d __A)
-{
+static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_broadcastsd_pd(__mmask8 __M, __m128d __A) {
   return (__m512d)__builtin_ia32_selectpd_512(__M,
                                               (__v8df) _mm512_broadcastsd_pd(__A),
                                               (__v8df) _mm512_setzero_pd());
 }
 
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_mask_broadcastss_ps (__m512 __O, __mmask16 __M, __m128 __A)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_broadcastss_ps(__m512 __O, __mmask16 __M, __m128 __A) {
   return (__m512)__builtin_ia32_selectps_512(__M,
                                              (__v16sf) _mm512_broadcastss_ps(__A),
                                              (__v16sf) __O);
 }
 
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_maskz_broadcastss_ps (__mmask16 __M, __m128 __A)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_broadcastss_ps(__mmask16 __M, __m128 __A) {
   return (__m512)__builtin_ia32_selectps_512(__M,
                                              (__v16sf) _mm512_broadcastss_ps(__A),
                                              (__v16sf) _mm512_setzero_ps());
@@ -8381,17 +8366,15 @@ _mm512_movehdup_ps (__m512 __A)
                          1, 1, 3, 3, 5, 5, 7, 7, 9, 9, 11, 11, 13, 13, 15, 15);
 }
 
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_mask_movehdup_ps (__m512 __W, __mmask16 __U, __m512 __A)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_movehdup_ps(__m512 __W, __mmask16 __U, __m512 __A) {
   return (__m512)__builtin_ia32_selectps_512((__mmask16)__U,
                                              (__v16sf)_mm512_movehdup_ps(__A),
                                              (__v16sf)__W);
 }
 
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_maskz_movehdup_ps (__mmask16 __U, __m512 __A)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_movehdup_ps(__mmask16 __U, __m512 __A) {
   return (__m512)__builtin_ia32_selectps_512((__mmask16)__U,
                                              (__v16sf)_mm512_movehdup_ps(__A),
                                              (__v16sf)_mm512_setzero_ps());
@@ -8404,44 +8387,38 @@ _mm512_moveldup_ps (__m512 __A)
                          0, 0, 2, 2, 4, 4, 6, 6, 8, 8, 10, 10, 12, 12, 14, 14);
 }
 
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_mask_moveldup_ps (__m512 __W, __mmask16 __U, __m512 __A)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_moveldup_ps(__m512 __W, __mmask16 __U, __m512 __A) {
   return (__m512)__builtin_ia32_selectps_512((__mmask16)__U,
                                              (__v16sf)_mm512_moveldup_ps(__A),
                                              (__v16sf)__W);
 }
 
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_maskz_moveldup_ps (__mmask16 __U, __m512 __A)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_moveldup_ps(__mmask16 __U, __m512 __A) {
   return (__m512)__builtin_ia32_selectps_512((__mmask16)__U,
                                              (__v16sf)_mm512_moveldup_ps(__A),
                                              (__v16sf)_mm512_setzero_ps());
 }
 
-static __inline__ __m128 __DEFAULT_FN_ATTRS128
-_mm_mask_move_ss (__m128 __W, __mmask8 __U, __m128 __A, __m128 __B)
-{
+static __inline__ __m128 __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_mask_move_ss(__m128 __W, __mmask8 __U, __m128 __A, __m128 __B) {
   return __builtin_ia32_selectss_128(__U, _mm_move_ss(__A, __B), __W);
 }
 
-static __inline__ __m128 __DEFAULT_FN_ATTRS128
-_mm_maskz_move_ss (__mmask8 __U, __m128 __A, __m128 __B)
-{
+static __inline__ __m128 __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_maskz_move_ss(__mmask8 __U, __m128 __A, __m128 __B) {
   return __builtin_ia32_selectss_128(__U, _mm_move_ss(__A, __B),
                                      _mm_setzero_ps());
 }
 
-static __inline__ __m128d __DEFAULT_FN_ATTRS128
-_mm_mask_move_sd (__m128d __W, __mmask8 __U, __m128d __A, __m128d __B)
-{
+static __inline__ __m128d __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_mask_move_sd(__m128d __W, __mmask8 __U, __m128d __A, __m128d __B) {
   return __builtin_ia32_selectsd_128(__U, _mm_move_sd(__A, __B), __W);
 }
 
-static __inline__ __m128d __DEFAULT_FN_ATTRS128
-_mm_maskz_move_sd (__mmask8 __U, __m128d __A, __m128d __B)
-{
+static __inline__ __m128d __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_maskz_move_sd(__mmask8 __U, __m128d __A, __m128d __B) {
   return __builtin_ia32_selectsd_128(__U, _mm_move_sd(__A, __B),
                                      _mm_setzero_pd());
 }
@@ -8884,17 +8861,15 @@ _mm_cvtu64_ss (__m128 __A, unsigned long long __B)
 }
 #endif
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_mask_set1_epi32 (__m512i __O, __mmask16 __M, int __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_set1_epi32(__m512i __O, __mmask16 __M, int __A) {
   return (__m512i) __builtin_ia32_selectd_512(__M,
                                               (__v16si) _mm512_set1_epi32(__A),
                                               (__v16si) __O);
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_mask_set1_epi64 (__m512i __O, __mmask8 __M, long long __A)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_set1_epi64(__m512i __O, __mmask8 __M, long long __A) {
   return (__m512i) __builtin_ia32_selectq_512(__M,
                                               (__v8di) _mm512_set1_epi64(__A),
                                               (__v8di) __O);
diff --git a/clang/lib/Headers/avx512vldqintrin.h b/clang/lib/Headers/avx512vldqintrin.h
index 68bd52e43981a..ee7974e924afe 100644
--- a/clang/lib/Headers/avx512vldqintrin.h
+++ b/clang/lib/Headers/avx512vldqintrin.h
@@ -968,17 +968,15 @@ _mm256_broadcast_f32x2(__m128 __A) {
                                          0, 1, 0, 1, 0, 1, 0, 1);
 }
 
-static __inline__ __m256 __DEFAULT_FN_ATTRS256
-_mm256_mask_broadcast_f32x2 (__m256 __O, __mmask8 __M, __m128 __A)
-{
+static __inline__ __m256 __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_mask_broadcast_f32x2(__m256 __O, __mmask8 __M, __m128 __A) {
   return (__m256)__builtin_ia32_selectps_256((__mmask8)__M,
                                              (__v8sf)_mm256_broadcast_f32x2(__A),
                                              (__v8sf)__O);
 }
 
-static __inline__ __m256 __DEFAULT_FN_ATTRS256
-_mm256_maskz_broadcast_f32x2 (__mmask8 __M, __m128 __A)
-{
+static __inline__ __m256 __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_maskz_broadcast_f32x2(__mmask8 __M, __m128 __A) {
   return (__m256)__builtin_ia32_selectps_256((__mmask8)__M,
                                              (__v8sf)_mm256_broadcast_f32x2(__A),
                                              (__v8sf)_mm256_setzero_ps());
@@ -990,17 +988,15 @@ _mm256_broadcast_f64x2(__m128d __A) {
                                           0, 1, 0, 1);
 }
 
-static __inline__ __m256d __DEFAULT_FN_ATTRS256
-_mm256_mask_broadcast_f64x2(__m256d __O, __mmask8 __M, __m128d __A)
-{
+static __inline__ __m256d __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_mask_broadcast_f64x2(__m256d __O, __mmask8 __M, __m128d __A) {
   return (__m256d)__builtin_ia32_selectpd_256((__mmask8)__M,
                                             (__v4df)_mm256_broadcast_f64x2(__A),
                                             (__v4df)__O);
 }
 
-static __inline__ __m256d __DEFAULT_FN_ATTRS256
-_mm256_maskz_broadcast_f64x2 (__mmask8 __M, __m128d __A)
-{
+static __inline__ __m256d __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_maskz_broadcast_f64x2(__mmask8 __M, __m128d __A) {
   return (__m256d)__builtin_ia32_selectpd_256((__mmask8)__M,
                                             (__v4df)_mm256_broadcast_f64x2(__A),
...
[truncated]

RKSimon · 2025-10-07T16:47:22Z

clang/test/CodeGen/X86/avx512dq-builtins.c

+// constexpr coverage for i32x8/f32x8 broadcasts (DQ)
+
+
+  TEST_CONSTEXPR(match_v16si(_mm512_mask_broadcast_i32x8(_mm512_setzero_si512(), 0xFFFF, _mm256_set1_epi32(8)),


(style) Please move these next to the corresponding test_* function - we want all test coverage for _mm512_mask_broadcast_i32x8 etc. to be close together

@RKSimon Thanks for the Feedback, I have updated the test files

github-actions · 2025-10-07T16:50:02Z

✅ With the latest revision this PR passed the C/C++ code formatter.

ahmednour25689 · 2025-10-08T20:35:57Z

@RKSimon the PR is ready for review again and all feedback addressed
please have a look when you have time

clang/test/CodeGen/X86/avx512vl-builtins.c

clang/test/CodeGen/X86/avx512f-builtins.c

ahmednour25689 · 2025-10-09T12:46:33Z

@RKSimon this is ready for review again when you have time

ahmednour25689 · 2025-10-13T08:10:56Z

Hey @RKSimon please review when you have some time

RKSimon

A few minors - sorry for the short review I'm a bit snowed under with other things at the moment

clang/test/CodeGen/X86/avx512dq-builtins.c

clang/test/CodeGen/X86/avx512vl-builtins.c

RKSimon

still quite a few missing tests - and need to avoid using "0xFF" allbits" style masks

ahmednour25689 · 2025-10-16T09:15:38Z

Hey @RKSimon i have addressed your feedback please review when you have some time

RKSimon

LGTM - cheers

github-actions · 2025-10-16T16:09:25Z

@ahmednour25689 Congratulations on having your first Pull Request (PR) merged into the LLVM Project!

Your changes will be combined with recent changes from other authors, then tested by our build bots. If there is a problem with a build, you may receive a report in an email or a comment on this PR.

Please check whether problems have been caused by your change specifically, as the builds can include changes from many authors. It is not uncommon for your change to be included in a build that fails due to someone else's changes, or infrastructure issues.

How to do this, and the rest of the post-merge process, is covered in detail here.

If your change does cause a problem, it may be reverted, or you can revert it yourself. This is a normal part of LLVM development. You can fix your changes and open a new PR to merge them again.

If you don't get any reports, no action is required from you. Your changes are working as expected, well done!

ahmednour25689 added 4 commits October 7, 2025 13:03

feat: [Headers][X86] Allow AVX512 masked shuffles to be used in const…

0f0effd

…expr

chor: apply formatting

facdc7e

refactor: update tests

d9488e5

refactor: update test

c9472b6

llvmbot added clang Clang issues not falling into any other category backend:X86 clang:headers Headers provided by Clang, e.g. for intrinsics labels Oct 7, 2025

RKSimon self-requested a review October 7, 2025 15:50

RKSimon reviewed Oct 7, 2025

View reviewed changes

ahmednour25689 added 3 commits October 7, 2025 21:52

refactor: PR Feedback

1fcea1b

chore: fix formatting

39619bf

Merge branch 'main' into an/avx-masked-shuffles

27759e2

ahmednour25689 requested a review from RKSimon October 7, 2025 18:58

Merge branch 'main' into an/avx-masked-shuffles

fcf5e8b

Merge branch 'main' into an/avx-masked-shuffles

b4101ff

RKSimon requested changes Oct 9, 2025

View reviewed changes

clang/test/CodeGen/X86/avx512vl-builtins.c Outdated Show resolved Hide resolved

clang/test/CodeGen/X86/avx512f-builtins.c Outdated Show resolved Hide resolved

ahmednour25689 added 2 commits October 9, 2025 13:09

chore: Move tests from avx512vl-builtins

539b86e

chore: Move test func from avx512f-builtins to corresponding test file

ccc654a

ahmednour25689 requested a review from RKSimon October 9, 2025 10:14

Merge branch 'main' into an/avx-masked-shuffles

3c88c39

ahmednour25689 added 2 commits October 10, 2025 21:49

Merge branch 'main' into an/avx-masked-shuffles

c2df2f4

Merge branch 'main' into an/avx-masked-shuffles

467f6ef

Merge branch 'main' into an/avx-masked-shuffles

1e989b0

RKSimon requested changes Oct 14, 2025

View reviewed changes

clang/test/CodeGen/X86/avx512dq-builtins.c Outdated Show resolved Hide resolved

clang/test/CodeGen/X86/avx512vl-builtins.c Show resolved Hide resolved

chore: PR Feedback

f0f7f3f

ahmednour25689 requested a review from RKSimon October 15, 2025 09:08

Merge branch 'main' into an/avx-masked-shuffles

505dfd3

RKSimon requested changes Oct 15, 2025

View reviewed changes

ahmednour25689 added 4 commits October 16, 2025 11:40

chore: add missing tests

df5211e

refactor: update tests

89ad0c3

feat: Use partial masks

cfea576

Merge branch 'main' into an/avx-masked-shuffles

1c01f28

ahmednour25689 requested a review from RKSimon October 16, 2025 09:15

RKSimon approved these changes Oct 16, 2025

View reviewed changes

Merge branch 'main' into an/avx-masked-shuffles

b5d9eea

RKSimon enabled auto-merge (squash) October 16, 2025 15:33

RKSimon merged commit 7fe0691 into llvm:main Oct 16, 2025
7 of 9 checks passed

		// constexpr coverage for i32x8/f32x8 broadcasts (DQ)


		TEST_CONSTEXPR(match_v16si(_mm512_mask_broadcast_i32x8(_mm512_setzero_si512(), 0xFFFF, _mm256_set1_epi32(8)),

[Headers][X86] Allow AVX512 masked shuffles to be used in constexpr #162301

[Headers][X86] Allow AVX512 masked shuffles to be used in constexpr #162301

Uh oh!

Conversation

ahmednour25689 commented Oct 7, 2025

Uh oh!

github-actions bot commented Oct 7, 2025

Uh oh!

llvmbot commented Oct 7, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

RKSimon Oct 7, 2025

Choose a reason for hiding this comment

Uh oh!

ahmednour25689 Oct 7, 2025

Choose a reason for hiding this comment

Uh oh!

github-actions bot commented Oct 7, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ahmednour25689 commented Oct 8, 2025

Uh oh!

Uh oh!

Uh oh!

ahmednour25689 commented Oct 9, 2025

Uh oh!

ahmednour25689 commented Oct 13, 2025

Uh oh!

RKSimon left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

RKSimon left a comment

Choose a reason for hiding this comment

Uh oh!

ahmednour25689 commented Oct 16, 2025

Uh oh!

RKSimon left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

github-actions bot commented Oct 16, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

llvmbot commented Oct 7, 2025 •

edited

Loading

github-actions bot commented Oct 7, 2025 •

edited

Loading