Skip to content

[Headers][X86] Allow SSE2/AVX2/AVX512F/AVX512BW/AVX512DQ integer arithmetic intrinsics to be used in constexpr #157582

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 16 commits into from
Sep 22, 2025
Merged

[Headers][X86] Allow SSE2/AVX2/AVX512F/AVX512BW/AVX512DQ integer arithmetic intrinsics to be used in constexpr #157582

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
16 commits
Select commit Hold shift + click to select a range
f0f9927
[clang][x86][headers] Make SSE2 add/sub intrinsics constexpr
donneypr Sep 8, 2025
590c0e1
[clang][x86][headers] Make AVX2 add/sub intrinsics constexpr
donneypr Sep 9, 2025
dca479e
[clang][x86][headers] Make AVX-512 epi64 add/sub intrinsics constexpr
donneypr Sep 9, 2025
0d4b800
[clang][x86][headers] Make AVX-512F masked epi32 add/sub constexpr Fi…
donneypr Sep 9, 2025
c70762a
[clang][x86][headers] Make AVX-512BW masked add/sub (epi8/epi16) cons…
donneypr Sep 9, 2025
8b22bd7
clang-format: touched lines in x86 add/sub headers for #152490
donneypr Sep 9, 2025
287254a
formatted only touched lines
donneypr Sep 15, 2025
25a4077
made _mm_sub_epi8 CONSTEXPR
donneypr Sep 15, 2025
873f6a0
added tests for all add arithmitics llvm#152490
donneypr Sep 19, 2025
01a003a
added tests for all sub arithmitics llvm#152490
donneypr Sep 19, 2025
1df56c5
mask and maskz (sub) in CONSTEXPR llvm#152490
donneypr Sep 19, 2025
bb2c2e3
mask and maskz (add) in CONSTEXPR llvm#152490
donneypr Sep 19, 2025
ce5d3b6
clang-formatted and also converted __v16qi/__v32qi/__v64qi to __v16qs…
donneypr Sep 21, 2025
1e40f32
clang-format avx512vlbwintrin.h
donneypr Sep 21, 2025
c01e2a3
clang-format again, missed avx512vlintrin.h
donneypr Sep 21, 2025
fc0fc3f
Merge branch 'main' into constexpr-addsub
RKSimon Sep 22, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
40 changes: 16 additions & 24 deletions clang/lib/Headers/avx2intrin.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -279,9 +279,8 @@ _mm256_packus_epi32(__m256i __V1, __m256i __V2)
/// \param __b
/// A 256-bit integer vector containing one of the source operands.
/// \returns A 256-bit integer vector containing the sums.
static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_add_epi8(__m256i __a, __m256i __b)
{
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_add_epi8(__m256i __a, __m256i __b) {
return (__m256i)((__v32qu)__a + (__v32qu)__b);
}

Expand All @@ -298,9 +297,8 @@ _mm256_add_epi8(__m256i __a, __m256i __b)
/// \param __b
/// A 256-bit vector of [16 x i16] containing one of the source operands.
/// \returns A 256-bit vector of [16 x i16] containing the sums.
static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_add_epi16(__m256i __a, __m256i __b)
{
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_add_epi16(__m256i __a, __m256i __b) {
return (__m256i)((__v16hu)__a + (__v16hu)__b);
}

Expand All @@ -317,9 +315,8 @@ _mm256_add_epi16(__m256i __a, __m256i __b)
/// \param __b
/// A 256-bit vector of [8 x i32] containing one of the source operands.
/// \returns A 256-bit vector of [8 x i32] containing the sums.
static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_add_epi32(__m256i __a, __m256i __b)
{
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_add_epi32(__m256i __a, __m256i __b) {
return (__m256i)((__v8su)__a + (__v8su)__b);
}

Expand All @@ -336,9 +333,8 @@ _mm256_add_epi32(__m256i __a, __m256i __b)
/// \param __b
/// A 256-bit vector of [4 x i64] containing one of the source operands.
/// \returns A 256-bit vector of [4 x i64] containing the sums.
static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_add_epi64(__m256i __a, __m256i __b)
{
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_add_epi64(__m256i __a, __m256i __b) {
return (__m256i)((__v4du)__a + (__v4du)__b);
}

Expand Down Expand Up @@ -2464,9 +2460,8 @@ _mm256_srl_epi64(__m256i __a, __m128i __count)
/// \param __b
/// A 256-bit integer vector containing the subtrahends.
/// \returns A 256-bit integer vector containing the differences.
static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_sub_epi8(__m256i __a, __m256i __b)
{
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_sub_epi8(__m256i __a, __m256i __b) {
return (__m256i)((__v32qu)__a - (__v32qu)__b);
}

Expand All @@ -2491,9 +2486,8 @@ _mm256_sub_epi8(__m256i __a, __m256i __b)
/// \param __b
/// A 256-bit vector of [16 x i16] containing the subtrahends.
/// \returns A 256-bit vector of [16 x i16] containing the differences.
static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_sub_epi16(__m256i __a, __m256i __b)
{
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_sub_epi16(__m256i __a, __m256i __b) {
return (__m256i)((__v16hu)__a - (__v16hu)__b);
}

Expand All @@ -2517,9 +2511,8 @@ _mm256_sub_epi16(__m256i __a, __m256i __b)
/// \param __b
/// A 256-bit vector of [8 x i32] containing the subtrahends.
/// \returns A 256-bit vector of [8 x i32] containing the differences.
static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_sub_epi32(__m256i __a, __m256i __b)
{
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_sub_epi32(__m256i __a, __m256i __b) {
return (__m256i)((__v8su)__a - (__v8su)__b);
}

Expand All @@ -2543,9 +2536,8 @@ _mm256_sub_epi32(__m256i __a, __m256i __b)
/// \param __b
/// A 256-bit vector of [4 x i64] containing the subtrahends.
/// \returns A 256-bit vector of [4 x i64] containing the differences.
static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_sub_epi64(__m256i __a, __m256i __b)
{
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_sub_epi64(__m256i __a, __m256i __b) {
return (__m256i)((__v4du)__a - (__v4du)__b);
}

Expand Down
32 changes: 16 additions & 16 deletions clang/lib/Headers/avx512bwintrin.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -369,76 +369,76 @@ static __inline__ void __DEFAULT_FN_ATTRS _store_mask64(__mmask64 *__A,
#define _mm512_mask_cmpneq_epu16_mask(k, A, B) \
_mm512_mask_cmp_epu16_mask((k), (A), (B), _MM_CMPINT_NE)

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_add_epi8 (__m512i __A, __m512i __B) {
static __inline__ __m512i
__DEFAULT_FN_ATTRS512_CONSTEXPR _mm512_add_epi8(__m512i __A, __m512i __B) {
return (__m512i) ((__v64qu) __A + (__v64qu) __B);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_mask_add_epi8(__m512i __W, __mmask64 __U, __m512i __A, __m512i __B) {
return (__m512i)__builtin_ia32_selectb_512((__mmask64)__U,
(__v64qi)_mm512_add_epi8(__A, __B),
(__v64qi)__W);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_maskz_add_epi8(__mmask64 __U, __m512i __A, __m512i __B) {
return (__m512i)__builtin_ia32_selectb_512((__mmask64)__U,
(__v64qi)_mm512_add_epi8(__A, __B),
(__v64qi)_mm512_setzero_si512());
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_sub_epi8 (__m512i __A, __m512i __B) {
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_sub_epi8(__m512i __A, __m512i __B) {
return (__m512i) ((__v64qu) __A - (__v64qu) __B);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_mask_sub_epi8(__m512i __W, __mmask64 __U, __m512i __A, __m512i __B) {
return (__m512i)__builtin_ia32_selectb_512((__mmask64)__U,
(__v64qi)_mm512_sub_epi8(__A, __B),
(__v64qi)__W);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_maskz_sub_epi8(__mmask64 __U, __m512i __A, __m512i __B) {
return (__m512i)__builtin_ia32_selectb_512((__mmask64)__U,
(__v64qi)_mm512_sub_epi8(__A, __B),
(__v64qi)_mm512_setzero_si512());
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_add_epi16 (__m512i __A, __m512i __B) {
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_add_epi16(__m512i __A, __m512i __B) {
return (__m512i) ((__v32hu) __A + (__v32hu) __B);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_mask_add_epi16(__m512i __W, __mmask32 __U, __m512i __A, __m512i __B) {
return (__m512i)__builtin_ia32_selectw_512((__mmask32)__U,
(__v32hi)_mm512_add_epi16(__A, __B),
(__v32hi)__W);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_maskz_add_epi16(__mmask32 __U, __m512i __A, __m512i __B) {
return (__m512i)__builtin_ia32_selectw_512((__mmask32)__U,
(__v32hi)_mm512_add_epi16(__A, __B),
(__v32hi)_mm512_setzero_si512());
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_sub_epi16 (__m512i __A, __m512i __B) {
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_sub_epi16(__m512i __A, __m512i __B) {
return (__m512i) ((__v32hu) __A - (__v32hu) __B);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_mask_sub_epi16(__m512i __W, __mmask32 __U, __m512i __A, __m512i __B) {
return (__m512i)__builtin_ia32_selectw_512((__mmask32)__U,
(__v32hi)_mm512_sub_epi16(__A, __B),
(__v32hi)__W);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_maskz_sub_epi16(__mmask32 __U, __m512i __A, __m512i __B) {
return (__m512i)__builtin_ia32_selectw_512((__mmask32)__U,
(__v32hi)_mm512_sub_epi16(__A, __B),
Expand Down
55 changes: 22 additions & 33 deletions clang/lib/Headers/avx512fintrin.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -859,83 +859,72 @@ _mm512_add_epi64(__m512i __A, __m512i __B) {
return (__m512i) ((__v8du) __A + (__v8du) __B);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_mask_add_epi64(__m512i __W, __mmask8 __U, __m512i __A, __m512i __B)
{
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_mask_add_epi64(__m512i __W, __mmask8 __U, __m512i __A, __m512i __B) {
return (__m512i)__builtin_ia32_selectq_512((__mmask8)__U,
(__v8di)_mm512_add_epi64(__A, __B),
(__v8di)__W);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_maskz_add_epi64(__mmask8 __U, __m512i __A, __m512i __B)
{
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_maskz_add_epi64(__mmask8 __U, __m512i __A, __m512i __B) {
return (__m512i)__builtin_ia32_selectq_512((__mmask8)__U,
(__v8di)_mm512_add_epi64(__A, __B),
(__v8di)_mm512_setzero_si512());
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_sub_epi64 (__m512i __A, __m512i __B)
{
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_sub_epi64(__m512i __A, __m512i __B) {
return (__m512i) ((__v8du) __A - (__v8du) __B);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_mask_sub_epi64(__m512i __W, __mmask8 __U, __m512i __A, __m512i __B)
{
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_mask_sub_epi64(__m512i __W, __mmask8 __U, __m512i __A, __m512i __B) {
return (__m512i)__builtin_ia32_selectq_512((__mmask8)__U,
(__v8di)_mm512_sub_epi64(__A, __B),
(__v8di)__W);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_maskz_sub_epi64(__mmask8 __U, __m512i __A, __m512i __B)
{
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_maskz_sub_epi64(__mmask8 __U, __m512i __A, __m512i __B) {
return (__m512i)__builtin_ia32_selectq_512((__mmask8)__U,
(__v8di)_mm512_sub_epi64(__A, __B),
(__v8di)_mm512_setzero_si512());
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_add_epi32 (__m512i __A, __m512i __B)
{
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_add_epi32(__m512i __A, __m512i __B) {
return (__m512i) ((__v16su) __A + (__v16su) __B);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_mask_add_epi32(__m512i __W, __mmask16 __U, __m512i __A, __m512i __B)
{
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_mask_add_epi32(__m512i __W, __mmask16 __U, __m512i __A, __m512i __B) {
return (__m512i)__builtin_ia32_selectd_512((__mmask16)__U,
(__v16si)_mm512_add_epi32(__A, __B),
(__v16si)__W);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_maskz_add_epi32 (__mmask16 __U, __m512i __A, __m512i __B)
{
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_maskz_add_epi32(__mmask16 __U, __m512i __A, __m512i __B) {
return (__m512i)__builtin_ia32_selectd_512((__mmask16)__U,
(__v16si)_mm512_add_epi32(__A, __B),
(__v16si)_mm512_setzero_si512());
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_sub_epi32 (__m512i __A, __m512i __B)
{
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_sub_epi32(__m512i __A, __m512i __B) {
return (__m512i) ((__v16su) __A - (__v16su) __B);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_mask_sub_epi32(__m512i __W, __mmask16 __U, __m512i __A, __m512i __B)
{
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_mask_sub_epi32(__m512i __W, __mmask16 __U, __m512i __A, __m512i __B) {
return (__m512i)__builtin_ia32_selectd_512((__mmask16)__U,
(__v16si)_mm512_sub_epi32(__A, __B),
(__v16si)__W);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_maskz_sub_epi32(__mmask16 __U, __m512i __A, __m512i __B)
{
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_maskz_sub_epi32(__mmask16 __U, __m512i __A, __m512i __B) {
return (__m512i)__builtin_ia32_selectd_512((__mmask16)__U,
(__v16si)_mm512_sub_epi32(__A, __B),
(__v16si)_mm512_setzero_si512());
Expand Down
Loading