Skip to content

[Clang] Add constexpr support for AVX512 permutexvar intrinsics #167802

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 4 commits into from
Nov 13, 2025
Merged

[Clang] Add constexpr support for AVX512 permutexvar intrinsics #167802

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
a561197
[Clang] VectorExprEvaluator::VisitCallExpr / InterpretBuiltin - Allow…
chaitanyav Nov 13, 2025
6752579
Address code review comments
chaitanyav Nov 13, 2025
3502a04
Add tests for permutevar8x32, permutevar8x32_epi32
chaitanyav Nov 13, 2025
f36d56d
Merge branch 'main' into issue_167476
RKSimon Nov 13, 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
21 changes: 11 additions & 10 deletions clang/include/clang/Basic/BuiltinsX86.td
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -617,9 +617,7 @@ let Features = "avx2", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] i
def psrlw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<8, short>)">;
def psrld256 : X86Builtin<"_Vector<8, int>(_Vector<8, int>, _Vector<4, int>)">;
def psrlq256 : X86Builtin<"_Vector<4, long long int>(_Vector<4, long long int>, _Vector<2, long long int>)">;
def permvarsi256 : X86Builtin<"_Vector<8, int>(_Vector<8, int>, _Vector<8, int>)">;
def permdf256 : X86Builtin<"_Vector<4, double>(_Vector<4, double>, _Constant int)">;
def permvarsf256 : X86Builtin<"_Vector<8, float>(_Vector<8, float>, _Vector<8, int>)">;
def permti256 : X86Builtin<"_Vector<4, long long int>(_Vector<4, long long int>, _Vector<4, long long int>, _Constant int)">;
def permdi256 : X86Builtin<"_Vector<4, long long int>(_Vector<4, long long int>, _Constant int)">;
}
Expand Down Expand Up @@ -692,6 +690,9 @@ let Features = "avx2", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWi
def pshuflw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Constant int)">;
def pshufhw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Constant int)">;
def pshufd256 : X86Builtin<"_Vector<8, int>(_Vector<8, int>, _Constant int)">;

def permvarsi256 : X86Builtin<"_Vector<8, int>(_Vector<8, int>, _Vector<8, int>)">;
def permvarsf256 : X86Builtin<"_Vector<8, float>(_Vector<8, float>, _Vector<8, int>)">;
}

let Features = "avx2", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<128>] in {
Expand Down Expand Up @@ -3042,38 +3043,38 @@ let Features = "avx512f", Attributes = [NoThrow, Const, RequiredVectorWidth<512>
def permdi512 : X86Builtin<"_Vector<8, long long int>(_Vector<8, long long int>, _Constant int)">;
}

let Features = "avx512bw", Attributes = [NoThrow, Const, RequiredVectorWidth<512>] in {
let Features = "avx512bw", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<512>] in {
def permvarhi512 : X86Builtin<"_Vector<32, short>(_Vector<32, short>, _Vector<32, short>)">;
}

let Features = "avx512f", Attributes = [NoThrow, Const, RequiredVectorWidth<512>] in {
let Features = "avx512f", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<512>] in {
def permvardf512 : X86Builtin<"_Vector<8, double>(_Vector<8, double>, _Vector<8, long long int>)">;
def permvardi512 : X86Builtin<"_Vector<8, long long int>(_Vector<8, long long int>, _Vector<8, long long int>)">;
def permvarsf512 : X86Builtin<"_Vector<16, float>(_Vector<16, float>, _Vector<16, int>)">;
def permvarsi512 : X86Builtin<"_Vector<16, int>(_Vector<16, int>, _Vector<16, int>)">;
}

let Features = "avx512vbmi", Attributes = [NoThrow, Const, RequiredVectorWidth<512>] in {
let Features = "avx512vbmi", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<512>] in {
def permvarqi512 : X86Builtin<"_Vector<64, char>(_Vector<64, char>, _Vector<64, char>)">;
}

let Features = "avx512vbmi,avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<128>] in {
let Features = "avx512vbmi,avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<128>] in {
def permvarqi128 : X86Builtin<"_Vector<16, char>(_Vector<16, char>, _Vector<16, char>)">;
}

let Features = "avx512vbmi,avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] in {
let Features = "avx512vbmi,avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<256>] in {
def permvarqi256 : X86Builtin<"_Vector<32, char>(_Vector<32, char>, _Vector<32, char>)">;
}

let Features = "avx512bw,avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<128>] in {
let Features = "avx512bw,avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<128>] in {
def permvarhi128 : X86Builtin<"_Vector<8, short>(_Vector<8, short>, _Vector<8, short>)">;
}

let Features = "avx512bw,avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] in {
let Features = "avx512bw,avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<256>] in {
def permvarhi256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<16, short>)">;
}

let Features = "avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] in {
let Features = "avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<256>] in {
def permvardf256 : X86Builtin<"_Vector<4, double>(_Vector<4, double>, _Vector<4, long long int>)">;
def permvardi256 : X86Builtin<"_Vector<4, long long int>(_Vector<4, long long int>, _Vector<4, long long int>)">;
}
Expand Down
39 changes: 39 additions & 0 deletions clang/lib/AST/ByteCode/InterpBuiltin.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4414,6 +4414,45 @@ bool InterpretBuiltin(InterpState &S, CodePtr OpPC, const CallExpr *Call,
return std::pair<unsigned, int>{0, static_cast<int>(DstIdx)};
}
});
case X86::BI__builtin_ia32_permvarsi256:
case X86::BI__builtin_ia32_permvarsf256:
case X86::BI__builtin_ia32_permvardf512:
case X86::BI__builtin_ia32_permvardi512:
case X86::BI__builtin_ia32_permvarhi128:
return interp__builtin_ia32_shuffle_generic(
S, OpPC, Call, [](unsigned DstIdx, unsigned ShuffleMask) {
int Offset = ShuffleMask & 0x7;
return std::pair<unsigned, int>{0, Offset};
});
case X86::BI__builtin_ia32_permvarqi128:
case X86::BI__builtin_ia32_permvarhi256:
case X86::BI__builtin_ia32_permvarsi512:
case X86::BI__builtin_ia32_permvarsf512:
return interp__builtin_ia32_shuffle_generic(
S, OpPC, Call, [](unsigned DstIdx, unsigned ShuffleMask) {
int Offset = ShuffleMask & 0xF;
return std::pair<unsigned, int>{0, Offset};
});
case X86::BI__builtin_ia32_permvardi256:
case X86::BI__builtin_ia32_permvardf256:
return interp__builtin_ia32_shuffle_generic(
S, OpPC, Call, [](unsigned DstIdx, unsigned ShuffleMask) {
int Offset = ShuffleMask & 0x3;
return std::pair<unsigned, int>{0, Offset};
});
case X86::BI__builtin_ia32_permvarqi256:
case X86::BI__builtin_ia32_permvarhi512:
return interp__builtin_ia32_shuffle_generic(
S, OpPC, Call, [](unsigned DstIdx, unsigned ShuffleMask) {
int Offset = ShuffleMask & 0x1F;
return std::pair<unsigned, int>{0, Offset};
});
case X86::BI__builtin_ia32_permvarqi512:
return interp__builtin_ia32_shuffle_generic(
S, OpPC, Call, [](unsigned DstIdx, unsigned ShuffleMask) {
int Offset = ShuffleMask & 0x3F;
return std::pair<unsigned, int>{0, Offset};
});
case X86::BI__builtin_ia32_vpermi2varq128:
case X86::BI__builtin_ia32_vpermi2varpd128:
return interp__builtin_ia32_shuffle_generic(
Expand Down
59 changes: 59 additions & 0 deletions clang/lib/AST/ExprConstant.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13551,6 +13551,65 @@ bool VectorExprEvaluator::VisitCallExpr(const CallExpr *E) {
return false;
return Success(R, E);
}
case X86::BI__builtin_ia32_permvarsi256:
case X86::BI__builtin_ia32_permvarsf256:
case X86::BI__builtin_ia32_permvardf512:
case X86::BI__builtin_ia32_permvardi512:
case X86::BI__builtin_ia32_permvarhi128: {
APValue R;
if (!evalShuffleGeneric(Info, E, R,
[](unsigned DstIdx, unsigned ShuffleMask) {
int Offset = ShuffleMask & 0x7;
return std::pair<unsigned, int>{0, Offset};
}))
return false;
return Success(R, E);
}
case X86::BI__builtin_ia32_permvarqi128:
case X86::BI__builtin_ia32_permvarhi256:
case X86::BI__builtin_ia32_permvarsi512:
case X86::BI__builtin_ia32_permvarsf512: {
APValue R;
if (!evalShuffleGeneric(Info, E, R,
[](unsigned DstIdx, unsigned ShuffleMask) {
int Offset = ShuffleMask & 0xF;
return std::pair<unsigned, int>{0, Offset};
}))
return false;
return Success(R, E);
}
case X86::BI__builtin_ia32_permvardi256:
case X86::BI__builtin_ia32_permvardf256: {
APValue R;
if (!evalShuffleGeneric(Info, E, R,
[](unsigned DstIdx, unsigned ShuffleMask) {
int Offset = ShuffleMask & 0x3;
return std::pair<unsigned, int>{0, Offset};
}))
return false;
return Success(R, E);
}
case X86::BI__builtin_ia32_permvarqi256:
case X86::BI__builtin_ia32_permvarhi512: {
APValue R;
if (!evalShuffleGeneric(Info, E, R,
[](unsigned DstIdx, unsigned ShuffleMask) {
int Offset = ShuffleMask & 0x1F;
return std::pair<unsigned, int>{0, Offset};
}))
return false;
return Success(R, E);
}
case X86::BI__builtin_ia32_permvarqi512: {
APValue R;
if (!evalShuffleGeneric(Info, E, R,
[](unsigned DstIdx, unsigned ShuffleMask) {
int Offset = ShuffleMask & 0x3F;
return std::pair<unsigned, int>{0, Offset};
}))
return false;
return Success(R, E);
}
case X86::BI__builtin_ia32_vpermi2varq128:
case X86::BI__builtin_ia32_vpermi2varpd128: {
APValue R;
Expand Down
2 changes: 1 addition & 1 deletion clang/lib/Headers/avx10_2_512bf16intrin.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -179,7 +179,7 @@ _mm512_permutex2var_pbh(__m512bh __A, __m512i __I, __m512bh __B) {
(__v32hi)__B);
}

static __inline__ __m512bh __DEFAULT_FN_ATTRS512
static __inline__ __m512bh __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_permutexvar_pbh(__m512i __A, __m512bh __B) {
return (__m512bh)__builtin_ia32_permvarhi512((__v32hi)__B, (__v32hi)__A);
}
Expand Down
4 changes: 2 additions & 2 deletions clang/lib/Headers/avx10_2bf16intrin.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -307,12 +307,12 @@ _mm256_permutex2var_pbh(__m256bh __A, __m256i __I, __m256bh __B) {
(__v16hi)__B);
}

static __inline__ __m128bh __DEFAULT_FN_ATTRS128
static __inline__ __m128bh __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_permutexvar_pbh(__m128i __A, __m128bh __B) {
return (__m128bh)__builtin_ia32_permvarhi128((__v8hi)__B, (__v8hi)__A);
}

static __inline__ __m256bh __DEFAULT_FN_ATTRS256
static __inline__ __m256bh __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_permutexvar_pbh(__m256i __A, __m256bh __B) {
return (__m256bh)__builtin_ia32_permvarhi256((__v16hi)__B, (__v16hi)__A);
}
Expand Down
10 changes: 4 additions & 6 deletions clang/lib/Headers/avx2intrin.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3214,9 +3214,8 @@ _mm_broadcastq_epi64(__m128i __X) {
/// A 256-bit vector of [8 x i32] containing indexes of values to use from
/// \a __a.
/// \returns A 256-bit vector of [8 x i32] containing the result.
static __inline__ __m256i __DEFAULT_FN_ATTRS256
_mm256_permutevar8x32_epi32(__m256i __a, __m256i __b)
{
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_permutevar8x32_epi32(__m256i __a, __m256i __b) {
Copy link
Collaborator

@RKSimon RKSimon Nov 13, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

test coverage?

return (__m256i)__builtin_ia32_permvarsi256((__v8si)__a, (__v8si)__b);
}

Expand Down Expand Up @@ -3272,9 +3271,8 @@ _mm256_permutevar8x32_epi32(__m256i __a, __m256i __b)
/// A 256-bit vector of [8 x i32] containing indexes of values to use from
/// \a __a.
/// \returns A 256-bit vector of [8 x float] containing the result.
static __inline__ __m256 __DEFAULT_FN_ATTRS256
_mm256_permutevar8x32_ps(__m256 __a, __m256i __b)
{
static __inline__ __m256 __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_permutevar8x32_ps(__m256 __a, __m256i __b) {
return (__m256)__builtin_ia32_permvarsf256((__v8sf)__a, (__v8si)__b);
}

Expand Down
18 changes: 7 additions & 11 deletions clang/lib/Headers/avx512bwintrin.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1846,25 +1846,21 @@ _mm512_maskz_broadcastw_epi16 (__mmask32 __M, __m128i __A)
(__v32hi) _mm512_setzero_si512());
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_permutexvar_epi16 (__m512i __A, __m512i __B)
{
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_permutexvar_epi16(__m512i __A, __m512i __B) {
return (__m512i)__builtin_ia32_permvarhi512((__v32hi)__B, (__v32hi)__A);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_maskz_permutexvar_epi16 (__mmask32 __M, __m512i __A,
__m512i __B)
{
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_maskz_permutexvar_epi16(__mmask32 __M, __m512i __A, __m512i __B) {
return (__m512i)__builtin_ia32_selectw_512((__mmask32)__M,
(__v32hi)_mm512_permutexvar_epi16(__A, __B),
(__v32hi)_mm512_setzero_si512());
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_mask_permutexvar_epi16 (__m512i __W, __mmask32 __M, __m512i __A,
__m512i __B)
{
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_mask_permutexvar_epi16(__m512i __W, __mmask32 __M, __m512i __A,
__m512i __B) {
return (__m512i)__builtin_ia32_selectw_512((__mmask32)__M,
(__v32hi)_mm512_permutexvar_epi16(__A, __B),
(__v32hi)__W);
Expand Down
65 changes: 27 additions & 38 deletions clang/lib/Headers/avx512fintrin.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7959,93 +7959,82 @@ _mm_mask3_fnmsub_sd (__m128d __W, __m128d __X, __m128d __Y, __mmask8 __U)
(__v8di)_mm512_permutex_epi64((X), (C)), \
(__v8di)_mm512_setzero_si512()))

static __inline__ __m512d __DEFAULT_FN_ATTRS512
_mm512_permutexvar_pd (__m512i __X, __m512d __Y)
{
static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_permutexvar_pd(__m512i __X, __m512d __Y) {
return (__m512d)__builtin_ia32_permvardf512((__v8df) __Y, (__v8di) __X);
}

static __inline__ __m512d __DEFAULT_FN_ATTRS512
_mm512_mask_permutexvar_pd (__m512d __W, __mmask8 __U, __m512i __X, __m512d __Y)
{
static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_mask_permutexvar_pd(__m512d __W, __mmask8 __U, __m512i __X,
__m512d __Y) {
return (__m512d)__builtin_ia32_selectpd_512((__mmask8)__U,
(__v8df)_mm512_permutexvar_pd(__X, __Y),
(__v8df)__W);
}

static __inline__ __m512d __DEFAULT_FN_ATTRS512
_mm512_maskz_permutexvar_pd (__mmask8 __U, __m512i __X, __m512d __Y)
{
static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_maskz_permutexvar_pd(__mmask8 __U, __m512i __X, __m512d __Y) {
return (__m512d)__builtin_ia32_selectpd_512((__mmask8)__U,
(__v8df)_mm512_permutexvar_pd(__X, __Y),
(__v8df)_mm512_setzero_pd());
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_permutexvar_epi64 (__m512i __X, __m512i __Y)
{
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_permutexvar_epi64(__m512i __X, __m512i __Y) {
return (__m512i)__builtin_ia32_permvardi512((__v8di)__Y, (__v8di)__X);
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_maskz_permutexvar_epi64 (__mmask8 __M, __m512i __X, __m512i __Y)
{
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_maskz_permutexvar_epi64(__mmask8 __M, __m512i __X, __m512i __Y) {
return (__m512i)__builtin_ia32_selectq_512((__mmask8)__M,
(__v8di)_mm512_permutexvar_epi64(__X, __Y),
(__v8di)_mm512_setzero_si512());
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_mask_permutexvar_epi64 (__m512i __W, __mmask8 __M, __m512i __X,
__m512i __Y)
{
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_mask_permutexvar_epi64(__m512i __W, __mmask8 __M, __m512i __X,
__m512i __Y) {
return (__m512i)__builtin_ia32_selectq_512((__mmask8)__M,
(__v8di)_mm512_permutexvar_epi64(__X, __Y),
(__v8di)__W);
}

static __inline__ __m512 __DEFAULT_FN_ATTRS512
_mm512_permutexvar_ps (__m512i __X, __m512 __Y)
{
static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_permutexvar_ps(__m512i __X, __m512 __Y) {
return (__m512)__builtin_ia32_permvarsf512((__v16sf)__Y, (__v16si)__X);
}

static __inline__ __m512 __DEFAULT_FN_ATTRS512
_mm512_mask_permutexvar_ps (__m512 __W, __mmask16 __U, __m512i __X, __m512 __Y)
{
static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_mask_permutexvar_ps(__m512 __W, __mmask16 __U, __m512i __X, __m512 __Y) {
return (__m512)__builtin_ia32_selectps_512((__mmask16)__U,
(__v16sf)_mm512_permutexvar_ps(__X, __Y),
(__v16sf)__W);
}

static __inline__ __m512 __DEFAULT_FN_ATTRS512
_mm512_maskz_permutexvar_ps (__mmask16 __U, __m512i __X, __m512 __Y)
{
static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_maskz_permutexvar_ps(__mmask16 __U, __m512i __X, __m512 __Y) {
return (__m512)__builtin_ia32_selectps_512((__mmask16)__U,
(__v16sf)_mm512_permutexvar_ps(__X, __Y),
(__v16sf)_mm512_setzero_ps());
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_permutexvar_epi32 (__m512i __X, __m512i __Y)
{
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_permutexvar_epi32(__m512i __X, __m512i __Y) {
return (__m512i)__builtin_ia32_permvarsi512((__v16si)__Y, (__v16si)__X);
}

#define _mm512_permutevar_epi32 _mm512_permutexvar_epi32

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_maskz_permutexvar_epi32 (__mmask16 __M, __m512i __X, __m512i __Y)
{
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_maskz_permutexvar_epi32(__mmask16 __M, __m512i __X, __m512i __Y) {
return (__m512i)__builtin_ia32_selectd_512((__mmask16)__M,
(__v16si)_mm512_permutexvar_epi32(__X, __Y),
(__v16si)_mm512_setzero_si512());
}

static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_mask_permutexvar_epi32 (__m512i __W, __mmask16 __M, __m512i __X,
__m512i __Y)
{
static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
_mm512_mask_permutexvar_epi32(__m512i __W, __mmask16 __M, __m512i __X,
__m512i __Y) {
return (__m512i)__builtin_ia32_selectd_512((__mmask16)__M,
(__v16si)_mm512_permutexvar_epi32(__X, __Y),
(__v16si)__W);
Expand Down
Loading
Loading