Skip to content

[Headers][X86] Update SSE/AVX and/andnot/or/xor intrinsics to be used in constexpr #152305

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 1 commit into from
Aug 7, 2025
Merged

[Headers][X86] Update SSE/AVX and/andnot/or/xor intrinsics to be used in constexpr #152305

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
18 changes: 14 additions & 4 deletions clang/lib/Headers/avx2intrin.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,6 +31,14 @@
__min_vector_width__(128)))
#endif

#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

/* SSE4 Multiple Packed Sums of Absolute Difference. */
/// Computes sixteen sum of absolute difference (SAD) operations on sets of
/// four unsigned 8-bit integers from the 256-bit integer vectors \a X and
Expand Down Expand Up @@ -460,7 +468,7 @@ _mm256_adds_epu16(__m256i __a, __m256i __b)
/// \param __b
/// A 256-bit integer vector.
/// \returns A 256-bit integer vector containing the result.
static __inline__ __m256i __DEFAULT_FN_ATTRS256
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_and_si256(__m256i __a, __m256i __b)
{
return (__m256i)((__v4du)__a & (__v4du)__b);
Expand All @@ -478,7 +486,7 @@ _mm256_and_si256(__m256i __a, __m256i __b)
/// \param __b
/// A 256-bit integer vector.
/// \returns A 256-bit integer vector containing the result.
static __inline__ __m256i __DEFAULT_FN_ATTRS256
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_andnot_si256(__m256i __a, __m256i __b)
{
return (__m256i)(~(__v4du)__a & (__v4du)__b);
Expand Down Expand Up @@ -1822,7 +1830,7 @@ _mm256_mul_epu32(__m256i __a, __m256i __b)
/// \param __b
/// A 256-bit integer vector.
/// \returns A 256-bit integer vector containing the result.
static __inline__ __m256i __DEFAULT_FN_ATTRS256
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_or_si256(__m256i __a, __m256i __b)
{
return (__m256i)((__v4du)__a | (__v4du)__b);
Expand Down Expand Up @@ -2974,7 +2982,7 @@ _mm256_unpacklo_epi64(__m256i __a, __m256i __b)
/// \param __b
/// A 256-bit integer vector.
/// \returns A 256-bit integer vector containing the result.
static __inline__ __m256i __DEFAULT_FN_ATTRS256
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_xor_si256(__m256i __a, __m256i __b)
{
return (__m256i)((__v4du)__a ^ (__v4du)__b);
Expand Down Expand Up @@ -5289,5 +5297,7 @@ _mm_srlv_epi64(__m128i __X, __m128i __Y)

#undef __DEFAULT_FN_ATTRS256
#undef __DEFAULT_FN_ATTRS128
#undef __DEFAULT_FN_ATTRS256_CONSTEXPR
#undef __DEFAULT_FN_ATTRS128_CONSTEXPR

#endif /* __AVX2INTRIN_H */
26 changes: 18 additions & 8 deletions clang/lib/Headers/avx512dqintrin.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,6 +20,14 @@
__attribute__((__always_inline__, __nodebug__, \
__target__("avx512dq,no-evex512")))

#if defined(__cplusplus) && (__cplusplus >= 201103L)
#define __DEFAULT_FN_ATTRS512_CONSTEXPR __DEFAULT_FN_ATTRS512 constexpr
#define __DEFAULT_FN_ATTRS_CONSTEXPR __DEFAULT_FN_ATTRS constexpr
#else
#define __DEFAULT_FN_ATTRS512_CONSTEXPR __DEFAULT_FN_ATTRS512
#define __DEFAULT_FN_ATTRS_CONSTEXPR __DEFAULT_FN_ATTRS
#endif

static __inline __mmask8 __DEFAULT_FN_ATTRS
_knot_mask8(__mmask8 __M)
{
Expand Down Expand Up @@ -167,7 +175,7 @@ _mm512_maskz_mullo_epi64(__mmask8 __U, __m512i __A, __m512i __B) {
(__v8di)_mm512_setzero_si512());
}

static __inline__ __m512d __DEFAULT_FN_ATTRS512
static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_xor_pd(__m512d __A, __m512d __B) {
return (__m512d)((__v8du)__A ^ (__v8du)__B);
}
Expand All @@ -186,7 +194,7 @@ _mm512_maskz_xor_pd(__mmask8 __U, __m512d __A, __m512d __B) {
(__v8df)_mm512_setzero_pd());
}

static __inline__ __m512 __DEFAULT_FN_ATTRS512
static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_xor_ps (__m512 __A, __m512 __B) {
return (__m512)((__v16su)__A ^ (__v16su)__B);
}
Expand All @@ -205,7 +213,7 @@ _mm512_maskz_xor_ps(__mmask16 __U, __m512 __A, __m512 __B) {
(__v16sf)_mm512_setzero_ps());
}

static __inline__ __m512d __DEFAULT_FN_ATTRS512
static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_or_pd(__m512d __A, __m512d __B) {
return (__m512d)((__v8du)__A | (__v8du)__B);
}
Expand All @@ -224,7 +232,7 @@ _mm512_maskz_or_pd(__mmask8 __U, __m512d __A, __m512d __B) {
(__v8df)_mm512_setzero_pd());
}

static __inline__ __m512 __DEFAULT_FN_ATTRS512
static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_or_ps(__m512 __A, __m512 __B) {
return (__m512)((__v16su)__A | (__v16su)__B);
}
Expand All @@ -243,7 +251,7 @@ _mm512_maskz_or_ps(__mmask16 __U, __m512 __A, __m512 __B) {
(__v16sf)_mm512_setzero_ps());
}

static __inline__ __m512d __DEFAULT_FN_ATTRS512
static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_and_pd(__m512d __A, __m512d __B) {
return (__m512d)((__v8du)__A & (__v8du)__B);
}
Expand All @@ -262,7 +270,7 @@ _mm512_maskz_and_pd(__mmask8 __U, __m512d __A, __m512d __B) {
(__v8df)_mm512_setzero_pd());
}

static __inline__ __m512 __DEFAULT_FN_ATTRS512
static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_and_ps(__m512 __A, __m512 __B) {
return (__m512)((__v16su)__A & (__v16su)__B);
}
Expand All @@ -281,7 +289,7 @@ _mm512_maskz_and_ps(__mmask16 __U, __m512 __A, __m512 __B) {
(__v16sf)_mm512_setzero_ps());
}

static __inline__ __m512d __DEFAULT_FN_ATTRS512
static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_andnot_pd(__m512d __A, __m512d __B) {
return (__m512d)(~(__v8du)__A & (__v8du)__B);
}
Expand All @@ -300,7 +308,7 @@ _mm512_maskz_andnot_pd(__mmask8 __U, __m512d __A, __m512d __B) {
(__v8df)_mm512_setzero_pd());
}

static __inline__ __m512 __DEFAULT_FN_ATTRS512
static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_andnot_ps(__m512 __A, __m512 __B) {
return (__m512)(~(__v16su)__A & (__v16su)__B);
}
Expand Down Expand Up @@ -1375,5 +1383,7 @@ _mm512_maskz_broadcast_i64x2(__mmask8 __M, __m128i __A)

#undef __DEFAULT_FN_ATTRS512
#undef __DEFAULT_FN_ATTRS
#undef __DEFAULT_FN_ATTRS512_CONSTEXPR
#undef __DEFAULT_FN_ATTRS_CONSTEXPR

#endif
24 changes: 12 additions & 12 deletions clang/lib/Headers/avx512fintrin.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -645,7 +645,7 @@ _mm512_zextsi256_si512(__m256i __a)
}

/* Bitwise operators */
static __inline__ __m512i __DEFAULT_FN_ATTRS512
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_and_epi32(__m512i __a, __m512i __b)
{
return (__m512i)((__v16su)__a & (__v16su)__b);
Expand All @@ -666,7 +666,7 @@ _mm512_maskz_and_epi32(__mmask16 __k, __m512i __a, __m512i __b)
__k, __a, __b);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_and_epi64(__m512i __a, __m512i __b)
{
return (__m512i)((__v8du)__a & (__v8du)__b);
Expand All @@ -687,13 +687,13 @@ _mm512_maskz_and_epi64(__mmask8 __k, __m512i __a, __m512i __b)
__k, __a, __b);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_andnot_si512 (__m512i __A, __m512i __B)
{
return (__m512i)(~(__v8du)__A & (__v8du)__B);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_andnot_epi32 (__m512i __A, __m512i __B)
{
return (__m512i)(~(__v16su)__A & (__v16su)__B);
Expand All @@ -714,7 +714,7 @@ _mm512_maskz_andnot_epi32(__mmask16 __U, __m512i __A, __m512i __B)
__U, __A, __B);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_andnot_epi64(__m512i __A, __m512i __B)
{
return (__m512i)(~(__v8du)__A & (__v8du)__B);
Expand All @@ -735,7 +735,7 @@ _mm512_maskz_andnot_epi64(__mmask8 __U, __m512i __A, __m512i __B)
__U, __A, __B);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_or_epi32(__m512i __a, __m512i __b)
{
return (__m512i)((__v16su)__a | (__v16su)__b);
Expand All @@ -755,7 +755,7 @@ _mm512_maskz_or_epi32(__mmask16 __k, __m512i __a, __m512i __b)
return (__m512i)_mm512_mask_or_epi32(_mm512_setzero_si512(), __k, __a, __b);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_or_epi64(__m512i __a, __m512i __b)
{
return (__m512i)((__v8du)__a | (__v8du)__b);
Expand All @@ -775,7 +775,7 @@ _mm512_maskz_or_epi64(__mmask8 __k, __m512i __a, __m512i __b)
return (__m512i)_mm512_mask_or_epi64(_mm512_setzero_si512(), __k, __a, __b);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_xor_epi32(__m512i __a, __m512i __b)
{
return (__m512i)((__v16su)__a ^ (__v16su)__b);
Expand All @@ -795,7 +795,7 @@ _mm512_maskz_xor_epi32(__mmask16 __k, __m512i __a, __m512i __b)
return (__m512i)_mm512_mask_xor_epi32(_mm512_setzero_si512(), __k, __a, __b);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_xor_epi64(__m512i __a, __m512i __b)
{
return (__m512i)((__v8du)__a ^ (__v8du)__b);
Expand All @@ -815,19 +815,19 @@ _mm512_maskz_xor_epi64(__mmask8 __k, __m512i __a, __m512i __b)
return (__m512i)_mm512_mask_xor_epi64(_mm512_setzero_si512(), __k, __a, __b);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_and_si512(__m512i __a, __m512i __b)
{
return (__m512i)((__v8du)__a & (__v8du)__b);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_or_si512(__m512i __a, __m512i __b)
{
return (__m512i)((__v8du)__a | (__v8du)__b);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_xor_si512(__m512i __a, __m512i __b)
{
return (__m512i)((__v8du)__a ^ (__v8du)__b);
Expand Down
16 changes: 8 additions & 8 deletions clang/lib/Headers/avxintrin.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -555,7 +555,7 @@ _mm256_rcp_ps(__m256 __a)
/// A 256-bit vector of [4 x double] containing one of the source operands.
/// \returns A 256-bit vector of [4 x double] containing the bitwise AND of the
/// values between both operands.
static __inline __m256d __DEFAULT_FN_ATTRS
static __inline __m256d __DEFAULT_FN_ATTRS_CONSTEXPR
_mm256_and_pd(__m256d __a, __m256d __b)
{
return (__m256d)((__v4du)__a & (__v4du)__b);
Expand All @@ -573,7 +573,7 @@ _mm256_and_pd(__m256d __a, __m256d __b)
/// A 256-bit vector of [8 x float] containing one of the source operands.
/// \returns A 256-bit vector of [8 x float] containing the bitwise AND of the
/// values between both operands.
static __inline __m256 __DEFAULT_FN_ATTRS
static __inline __m256 __DEFAULT_FN_ATTRS_CONSTEXPR
_mm256_and_ps(__m256 __a, __m256 __b)
{
return (__m256)((__v8su)__a & (__v8su)__b);
Expand All @@ -594,7 +594,7 @@ _mm256_and_ps(__m256 __a, __m256 __b)
/// \returns A 256-bit vector of [4 x double] containing the bitwise AND of the
/// values of the second operand and the one's complement of the first
/// operand.
static __inline __m256d __DEFAULT_FN_ATTRS
static __inline __m256d __DEFAULT_FN_ATTRS_CONSTEXPR
_mm256_andnot_pd(__m256d __a, __m256d __b)
{
return (__m256d)(~(__v4du)__a & (__v4du)__b);
Expand All @@ -615,7 +615,7 @@ _mm256_andnot_pd(__m256d __a, __m256d __b)
/// \returns A 256-bit vector of [8 x float] containing the bitwise AND of the
/// values of the second operand and the one's complement of the first
/// operand.
static __inline __m256 __DEFAULT_FN_ATTRS
static __inline __m256 __DEFAULT_FN_ATTRS_CONSTEXPR
_mm256_andnot_ps(__m256 __a, __m256 __b)
{
return (__m256)(~(__v8su)__a & (__v8su)__b);
Expand All @@ -633,7 +633,7 @@ _mm256_andnot_ps(__m256 __a, __m256 __b)
/// A 256-bit vector of [4 x double] containing one of the source operands.
/// \returns A 256-bit vector of [4 x double] containing the bitwise OR of the
/// values between both operands.
static __inline __m256d __DEFAULT_FN_ATTRS
static __inline __m256d __DEFAULT_FN_ATTRS_CONSTEXPR
_mm256_or_pd(__m256d __a, __m256d __b)
{
return (__m256d)((__v4du)__a | (__v4du)__b);
Expand All @@ -651,7 +651,7 @@ _mm256_or_pd(__m256d __a, __m256d __b)
/// A 256-bit vector of [8 x float] containing one of the source operands.
/// \returns A 256-bit vector of [8 x float] containing the bitwise OR of the
/// values between both operands.
static __inline __m256 __DEFAULT_FN_ATTRS
static __inline __m256 __DEFAULT_FN_ATTRS_CONSTEXPR
_mm256_or_ps(__m256 __a, __m256 __b)
{
return (__m256)((__v8su)__a | (__v8su)__b);
Expand All @@ -669,7 +669,7 @@ _mm256_or_ps(__m256 __a, __m256 __b)
/// A 256-bit vector of [4 x double] containing one of the source operands.
/// \returns A 256-bit vector of [4 x double] containing the bitwise XOR of the
/// values between both operands.
static __inline __m256d __DEFAULT_FN_ATTRS
static __inline __m256d __DEFAULT_FN_ATTRS_CONSTEXPR
_mm256_xor_pd(__m256d __a, __m256d __b)
{
return (__m256d)((__v4du)__a ^ (__v4du)__b);
Expand All @@ -687,7 +687,7 @@ _mm256_xor_pd(__m256d __a, __m256d __b)
/// A 256-bit vector of [8 x float] containing one of the source operands.
/// \returns A 256-bit vector of [8 x float] containing the bitwise XOR of the
/// values between both operands.
static __inline __m256 __DEFAULT_FN_ATTRS
static __inline __m256 __DEFAULT_FN_ATTRS_CONSTEXPR
_mm256_xor_ps(__m256 __a, __m256 __b)
{
return (__m256)((__v8su)__a ^ (__v8su)__b);
Expand Down
16 changes: 8 additions & 8 deletions clang/lib/Headers/emmintrin.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2676,8 +2676,8 @@ static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_subs_epu16(__m128i __a,
/// A 128-bit integer vector containing one of the source operands.
/// \returns A 128-bit integer vector containing the bitwise AND of the values
/// in both operands.
static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_and_si128(__m128i __a,
__m128i __b) {
static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR
_mm_and_si128(__m128i __a, __m128i __b) {
return (__m128i)((__v2du)__a & (__v2du)__b);
}

Expand All @@ -2695,8 +2695,8 @@ static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_and_si128(__m128i __a,
/// A 128-bit vector containing the right source operand.
/// \returns A 128-bit integer vector containing the bitwise AND of the one's
/// complement of the first operand and the values in the second operand.
static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_andnot_si128(__m128i __a,
__m128i __b) {
static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR
_mm_andnot_si128(__m128i __a, __m128i __b) {
return (__m128i)(~(__v2du)__a & (__v2du)__b);
}
/// Performs a bitwise OR of two 128-bit integer vectors.
Expand All @@ -2711,8 +2711,8 @@ static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_andnot_si128(__m128i __a,
/// A 128-bit integer vector containing one of the source operands.
/// \returns A 128-bit integer vector containing the bitwise OR of the values
/// in both operands.
static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_or_si128(__m128i __a,
__m128i __b) {
static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR
_mm_or_si128(__m128i __a, __m128i __b) {
return (__m128i)((__v2du)__a | (__v2du)__b);
}

Expand All @@ -2728,8 +2728,8 @@ static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_or_si128(__m128i __a,
/// A 128-bit integer vector containing one of the source operands.
/// \returns A 128-bit integer vector containing the bitwise exclusive OR of the
/// values in both operands.
static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_xor_si128(__m128i __a,
__m128i __b) {
static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR
_mm_xor_si128(__m128i __a, __m128i __b) {
return (__m128i)((__v2du)__a ^ (__v2du)__b);
}

Expand Down
Loading
Loading