[clang] 53ddeb4 - [X86] Add MMX/SSE/AVX PHADD/SUB & HADDPS/D intrinsics to be used in constexpr (#156822)
via cfe-commits
cfe-commits at lists.llvm.org
Tue Oct 14 03:54:15 PDT 2025
Author: why
Date: 2025-10-14T10:54:10Z
New Revision: 53ddeb493529a98d10acea77b84c73315f83d7f3
URL: https://github.com/llvm/llvm-project/commit/53ddeb493529a98d10acea77b84c73315f83d7f3
DIFF: https://github.com/llvm/llvm-project/commit/53ddeb493529a98d10acea77b84c73315f83d7f3.diff
LOG: [X86] Add MMX/SSE/AVX PHADD/SUB & HADDPS/D intrinsics to be used in constexpr (#156822)
[Clang] VectorExprEvaluator::VisitCallExpr / InterpretBuiltin - add MMX/SSE/AVX PHADD/SUB & HADDPS/D intrinsics to be used in constexpr
Fixes #155395
cover func:
_mm_hadd_pi16 _mm_hadd_epi16 _mm256_hadd_epi16
_mm_hadd_pi32 _mm_hadd_epi32 _mm256_hadd_epi32
_mm_hadds_pi16 _mm_hadds_epi16 _mm256_hadds_epi16
_mm_hsub_pi16 _mm_hsub_epi16 _mm256_hsub_epi16
_mm_hsub_pi32 _mm_hsub_epi32 _mm256_hsub_epi32
_mm_hsubs_pi16 _mm_hsubs_epi16 _mm256_hsubs_epi16
_mm_hadd_pd _mm256_hadd_pd
_mm_hadd_ps _mm256_hadd_ps
_mm_hsub_pd _mm256_hsub_pd
_mm_hsub_ps _mm256_hsub_ps
---------
Co-authored-by: whyuuwang <whyuuwang at tencent.com>
Co-authored-by: Simon Pilgrim <llvm-dev at redking.me.uk>
Co-authored-by: Simon Pilgrim <git at redking.me.uk>
Added:
Modified:
clang/include/clang/Basic/BuiltinsX86.td
clang/lib/AST/ByteCode/InterpBuiltin.cpp
clang/lib/AST/ExprConstant.cpp
clang/lib/Headers/avx2intrin.h
clang/lib/Headers/avxintrin.h
clang/lib/Headers/pmmintrin.h
clang/lib/Headers/tmmintrin.h
clang/test/CodeGen/X86/avx-builtins.c
clang/test/CodeGen/X86/avx2-builtins.c
clang/test/CodeGen/X86/mmx-builtins.c
clang/test/CodeGen/X86/sse3-builtins.c
clang/test/CodeGen/X86/ssse3-builtins.c
Removed:
################################################################################
diff --git a/clang/include/clang/Basic/BuiltinsX86.td b/clang/include/clang/Basic/BuiltinsX86.td
index b197d81e8feaa..a55dff276dd98 100644
--- a/clang/include/clang/Basic/BuiltinsX86.td
+++ b/clang/include/clang/Basic/BuiltinsX86.td
@@ -110,19 +110,20 @@ let Attributes = [Const, NoThrow, RequiredVectorWidth<128>] in {
}
let Features = "sse3" in {
- foreach Op = ["addsub", "hadd", "hsub"] in {
+ foreach Op = ["addsub"] in {
def Op#ps : X86Builtin<"_Vector<4, float>(_Vector<4, float>, _Vector<4, float>)">;
def Op#pd : X86Builtin<"_Vector<2, double>(_Vector<2, double>, _Vector<2, double>)">;
}
}
- let Features = "ssse3" in {
- foreach Op = ["phadd", "phsub"] in {
- def Op#w128 : X86Builtin<"_Vector<8, short>(_Vector<8, short>, _Vector<8, short>)">;
- def Op#sw128 : X86Builtin<"_Vector<8, short>(_Vector<8, short>, _Vector<8, short>)">;
- def Op#d128 : X86Builtin<"_Vector<4, int>(_Vector<4, int>, _Vector<4, int>)">;
+ let Features = "sse3", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<128>] in {
+ foreach Op = ["hadd", "hsub"] in {
+ def Op#ps : X86Builtin<"_Vector<4, float>(_Vector<4, float>, _Vector<4, float>)">;
+ def Op#pd : X86Builtin<"_Vector<2, double>(_Vector<2, double>, _Vector<2, double>)">;
}
+ }
+ let Features = "ssse3" in {
def pmulhrsw128 : X86Builtin<"_Vector<8, short>(_Vector<8, short>, _Vector<8, short>)">;
def pshufb128 : X86Builtin<"_Vector<16, char>(_Vector<16, char>, _Vector<16, char>)">;
def psignb128 : X86Builtin<"_Vector<16, char>(_Vector<16, char>, _Vector<16, char>)">;
@@ -137,7 +138,7 @@ let Attributes = [Const, NoThrow, RequiredVectorWidth<128>] in {
// AVX
let Attributes = [Const, NoThrow, RequiredVectorWidth<256>], Features = "avx" in {
- foreach Op = ["addsub", "hadd", "hsub", "max", "min"] in {
+ foreach Op = ["addsub", "max", "min"] in {
def Op#pd256 : X86Builtin<"_Vector<4, double>(_Vector<4, double>, _Vector<4, double>)">;
def Op#ps256 : X86Builtin<"_Vector<8, float>(_Vector<8, float>, _Vector<8, float>)">;
}
@@ -316,6 +317,14 @@ let Features = "ssse3", Attributes = [NoThrow, Const, RequiredVectorWidth<128>]
def palignr128 : X86Builtin<"_Vector<16, char>(_Vector<16, char>, _Vector<16, char>, _Constant int)">;
}
+let Features = "ssse3", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<128>] in {
+ foreach Op = ["phadd", "phsub"] in {
+ def Op#w128 : X86Builtin<"_Vector<8, short>(_Vector<8, short>, _Vector<8, short>)">;
+ def Op#sw128 : X86Builtin<"_Vector<8, short>(_Vector<8, short>, _Vector<8, short>)">;
+ def Op#d128 : X86Builtin<"_Vector<4, int>(_Vector<4, int>, _Vector<4, int>)">;
+ }
+}
+
let Features = "sse4.1", Attributes = [NoThrow, Const, RequiredVectorWidth<128>] in {
def insertps128 : X86Builtin<"_Vector<4, float>(_Vector<4, float>, _Vector<4, float>, _Constant char)">;
def roundps : X86Builtin<"_Vector<4, float>(_Vector<4, float>, _Constant int)">;
@@ -515,6 +524,11 @@ let Features = "avx", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWid
def vinsertf128_pd256 : X86Builtin<"_Vector<4, double>(_Vector<4, double>, _Vector<2, double>, _Constant int)">;
def vinsertf128_ps256 : X86Builtin<"_Vector<8, float>(_Vector<8, float>, _Vector<4, float>, _Constant int)">;
def vinsertf128_si256 : X86Builtin<"_Vector<8, int>(_Vector<8, int>, _Vector<4, int>, _Constant int)">;
+
+ foreach Op = ["hadd", "hsub"] in {
+ def Op#pd256 : X86Builtin<"_Vector<4, double>(_Vector<4, double>, _Vector<4, double>)">;
+ def Op#ps256 : X86Builtin<"_Vector<8, float>(_Vector<8, float>, _Vector<8, float>)">;
+ }
}
let Features = "avx", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] in {
@@ -592,12 +606,7 @@ let Features = "avx", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWid
let Features = "avx2", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] in {
def mpsadbw256 : X86Builtin<"_Vector<32, char>(_Vector<32, char>, _Vector<32, char>, _Constant char)">;
def palignr256 : X86Builtin<"_Vector<32, char>(_Vector<32, char>, _Vector<32, char>, _Constant int)">;
- def phaddw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<16, short>)">;
- def phaddd256 : X86Builtin<"_Vector<8, int>(_Vector<8, int>, _Vector<8, int>)">;
- def phaddsw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<16, short>)">;
- def phsubw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<16, short>)">;
- def phsubd256 : X86Builtin<"_Vector<8, int>(_Vector<8, int>, _Vector<8, int>)">;
- def phsubsw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<16, short>)">;
+
def pmovmskb256 : X86Builtin<"int(_Vector<32, char>)">;
def pmulhrsw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<16, short>)">;
def psadbw256 : X86Builtin<"_Vector<4, long long int>(_Vector<32, char>, _Vector<32, char>)">;
@@ -666,6 +675,13 @@ let Features = "avx2", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWi
def packssdw256 : X86Builtin<"_Vector<16, short>(_Vector<8, int>, _Vector<8, int>)">;
def packuswb256 : X86Builtin<"_Vector<32, char>(_Vector<16, short>, _Vector<16, short>)">;
+ def phaddw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<16, short>)">;
+ def phaddd256 : X86Builtin<"_Vector<8, int>(_Vector<8, int>, _Vector<8, int>)">;
+ def phaddsw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<16, short>)">;
+ def phsubw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<16, short>)">;
+ def phsubd256 : X86Builtin<"_Vector<8, int>(_Vector<8, int>, _Vector<8, int>)">;
+ def phsubsw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<16, short>)">;
+
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)">;
diff --git a/clang/lib/AST/ByteCode/InterpBuiltin.cpp b/clang/lib/AST/ByteCode/InterpBuiltin.cpp
index 84c5ecc9aac1e..5f0a77c125b85 100644
--- a/clang/lib/AST/ByteCode/InterpBuiltin.cpp
+++ b/clang/lib/AST/ByteCode/InterpBuiltin.cpp
@@ -2587,6 +2587,82 @@ static bool interp__builtin_ia32_pmul(
return true;
}
+static bool interp_builtin_horizontal_int_binop(
+ InterpState &S, CodePtr OpPC, const CallExpr *Call,
+ llvm::function_ref<APInt(const APSInt &, const APSInt &)> Fn) {
+ const auto *VT = Call->getArg(0)->getType()->castAs<VectorType>();
+ PrimType ElemT = *S.getContext().classify(VT->getElementType());
+ bool DestUnsigned = Call->getType()->isUnsignedIntegerOrEnumerationType();
+
+ const Pointer &RHS = S.Stk.pop<Pointer>();
+ const Pointer &LHS = S.Stk.pop<Pointer>();
+ const Pointer &Dst = S.Stk.peek<Pointer>();
+ unsigned NumElts = VT->getNumElements();
+ unsigned EltBits = S.getASTContext().getIntWidth(VT->getElementType());
+ unsigned EltsPerLane = 128 / EltBits;
+ unsigned Lanes = NumElts * EltBits / 128;
+ unsigned DestIndex = 0;
+
+ for (unsigned Lane = 0; Lane < Lanes; ++Lane) {
+ unsigned LaneStart = Lane * EltsPerLane;
+ for (unsigned I = 0; I < EltsPerLane; I += 2) {
+ INT_TYPE_SWITCH_NO_BOOL(ElemT, {
+ APSInt Elem1 = LHS.elem<T>(LaneStart + I).toAPSInt();
+ APSInt Elem2 = LHS.elem<T>(LaneStart + I + 1).toAPSInt();
+ APSInt ResL = APSInt(Fn(Elem1, Elem2), DestUnsigned);
+ Dst.elem<T>(DestIndex++) = static_cast<T>(ResL);
+ });
+ }
+
+ for (unsigned I = 0; I < EltsPerLane; I += 2) {
+ INT_TYPE_SWITCH_NO_BOOL(ElemT, {
+ APSInt Elem1 = RHS.elem<T>(LaneStart + I).toAPSInt();
+ APSInt Elem2 = RHS.elem<T>(LaneStart + I + 1).toAPSInt();
+ APSInt ResR = APSInt(Fn(Elem1, Elem2), DestUnsigned);
+ Dst.elem<T>(DestIndex++) = static_cast<T>(ResR);
+ });
+ }
+ }
+ Dst.initializeAllElements();
+ return true;
+}
+
+static bool interp_builtin_horizontal_fp_binop(
+ InterpState &S, CodePtr OpPC, const CallExpr *Call,
+ llvm::function_ref<APFloat(const APFloat &, const APFloat &,
+ llvm::RoundingMode)>
+ Fn) {
+ const Pointer &RHS = S.Stk.pop<Pointer>();
+ const Pointer &LHS = S.Stk.pop<Pointer>();
+ const Pointer &Dst = S.Stk.peek<Pointer>();
+ FPOptions FPO = Call->getFPFeaturesInEffect(S.Ctx.getLangOpts());
+ llvm::RoundingMode RM = getRoundingMode(FPO);
+ const auto *VT = Call->getArg(0)->getType()->castAs<VectorType>();
+
+ unsigned NumElts = VT->getNumElements();
+ unsigned EltBits = S.getASTContext().getTypeSize(VT->getElementType());
+ unsigned NumLanes = NumElts * EltBits / 128;
+ unsigned NumElemsPerLane = NumElts / NumLanes;
+ unsigned HalfElemsPerLane = NumElemsPerLane / 2;
+
+ for (unsigned L = 0; L != NumElts; L += NumElemsPerLane) {
+ using T = PrimConv<PT_Float>::T;
+ for (unsigned E = 0; E != HalfElemsPerLane; ++E) {
+ APFloat Elem1 = LHS.elem<T>(L + (2 * E) + 0).getAPFloat();
+ APFloat Elem2 = LHS.elem<T>(L + (2 * E) + 1).getAPFloat();
+ Dst.elem<T>(L + E) = static_cast<T>(Fn(Elem1, Elem2, RM));
+ }
+ for (unsigned E = 0; E != HalfElemsPerLane; ++E) {
+ APFloat Elem1 = RHS.elem<T>(L + (2 * E) + 0).getAPFloat();
+ APFloat Elem2 = RHS.elem<T>(L + (2 * E) + 1).getAPFloat();
+ Dst.elem<T>(L + E + HalfElemsPerLane) =
+ static_cast<T>(Fn(Elem1, Elem2, RM));
+ }
+ }
+ Dst.initializeAllElements();
+ return true;
+}
+
static bool interp__builtin_elementwise_triop_fp(
InterpState &S, CodePtr OpPC, const CallExpr *Call,
llvm::function_ref<APFloat(const APFloat &, const APFloat &,
@@ -3665,6 +3741,53 @@ bool InterpretBuiltin(InterpState &S, CodePtr OpPC, const CallExpr *Call,
case Builtin::BI__builtin_elementwise_min:
return interp__builtin_elementwise_maxmin(S, OpPC, Call, BuiltinID);
+ case clang::X86::BI__builtin_ia32_phaddw128:
+ case clang::X86::BI__builtin_ia32_phaddw256:
+ case clang::X86::BI__builtin_ia32_phaddd128:
+ case clang::X86::BI__builtin_ia32_phaddd256:
+ return interp_builtin_horizontal_int_binop(
+ S, OpPC, Call,
+ [](const APSInt &LHS, const APSInt &RHS) { return LHS + RHS; });
+ case clang::X86::BI__builtin_ia32_phaddsw128:
+ case clang::X86::BI__builtin_ia32_phaddsw256:
+ return interp_builtin_horizontal_int_binop(
+ S, OpPC, Call,
+ [](const APSInt &LHS, const APSInt &RHS) { return LHS.sadd_sat(RHS); });
+ case clang::X86::BI__builtin_ia32_phsubw128:
+ case clang::X86::BI__builtin_ia32_phsubw256:
+ case clang::X86::BI__builtin_ia32_phsubd128:
+ case clang::X86::BI__builtin_ia32_phsubd256:
+ return interp_builtin_horizontal_int_binop(
+ S, OpPC, Call,
+ [](const APSInt &LHS, const APSInt &RHS) { return LHS - RHS; });
+ case clang::X86::BI__builtin_ia32_phsubsw128:
+ case clang::X86::BI__builtin_ia32_phsubsw256:
+ return interp_builtin_horizontal_int_binop(
+ S, OpPC, Call,
+ [](const APSInt &LHS, const APSInt &RHS) { return LHS.ssub_sat(RHS); });
+ case clang::X86::BI__builtin_ia32_haddpd:
+ case clang::X86::BI__builtin_ia32_haddps:
+ case clang::X86::BI__builtin_ia32_haddpd256:
+ case clang::X86::BI__builtin_ia32_haddps256:
+ return interp_builtin_horizontal_fp_binop(
+ S, OpPC, Call,
+ [](const APFloat &LHS, const APFloat &RHS, llvm::RoundingMode RM) {
+ APFloat F = LHS;
+ F.add(RHS, RM);
+ return F;
+ });
+ case clang::X86::BI__builtin_ia32_hsubpd:
+ case clang::X86::BI__builtin_ia32_hsubps:
+ case clang::X86::BI__builtin_ia32_hsubpd256:
+ case clang::X86::BI__builtin_ia32_hsubps256:
+ return interp_builtin_horizontal_fp_binop(
+ S, OpPC, Call,
+ [](const APFloat &LHS, const APFloat &RHS, llvm::RoundingMode RM) {
+ APFloat F = LHS;
+ F.subtract(RHS, RM);
+ return F;
+ });
+
case clang::X86::BI__builtin_ia32_pmuldq128:
case clang::X86::BI__builtin_ia32_pmuldq256:
case clang::X86::BI__builtin_ia32_pmuldq512:
diff --git a/clang/lib/AST/ExprConstant.cpp b/clang/lib/AST/ExprConstant.cpp
index dfdfef2a87bff..59b4f4f6b5782 100644
--- a/clang/lib/AST/ExprConstant.cpp
+++ b/clang/lib/AST/ExprConstant.cpp
@@ -12381,6 +12381,169 @@ bool VectorExprEvaluator::VisitCallExpr(const CallExpr *E) {
return Success(APValue(ResultElements.data(), ResultElements.size()), E);
}
+ case clang::X86::BI__builtin_ia32_phaddw128:
+ case clang::X86::BI__builtin_ia32_phaddw256:
+ case clang::X86::BI__builtin_ia32_phaddd128:
+ case clang::X86::BI__builtin_ia32_phaddd256:
+ case clang::X86::BI__builtin_ia32_phaddsw128:
+ case clang::X86::BI__builtin_ia32_phaddsw256:
+
+ case clang::X86::BI__builtin_ia32_phsubw128:
+ case clang::X86::BI__builtin_ia32_phsubw256:
+ case clang::X86::BI__builtin_ia32_phsubd128:
+ case clang::X86::BI__builtin_ia32_phsubd256:
+ case clang::X86::BI__builtin_ia32_phsubsw128:
+ case clang::X86::BI__builtin_ia32_phsubsw256: {
+ APValue SourceLHS, SourceRHS;
+ if (!EvaluateAsRValue(Info, E->getArg(0), SourceLHS) ||
+ !EvaluateAsRValue(Info, E->getArg(1), SourceRHS))
+ return false;
+ QualType DestEltTy = E->getType()->castAs<VectorType>()->getElementType();
+ bool DestUnsigned = DestEltTy->isUnsignedIntegerOrEnumerationType();
+
+ unsigned NumElts = SourceLHS.getVectorLength();
+ unsigned EltBits = Info.Ctx.getIntWidth(DestEltTy);
+ unsigned EltsPerLane = 128 / EltBits;
+ SmallVector<APValue, 4> ResultElements;
+ ResultElements.reserve(NumElts);
+
+ for (unsigned LaneStart = 0; LaneStart != NumElts;
+ LaneStart += EltsPerLane) {
+ for (unsigned I = 0; I != EltsPerLane; I += 2) {
+ APSInt LHSA = SourceLHS.getVectorElt(LaneStart + I).getInt();
+ APSInt LHSB = SourceLHS.getVectorElt(LaneStart + I + 1).getInt();
+ switch (E->getBuiltinCallee()) {
+ case clang::X86::BI__builtin_ia32_phaddw128:
+ case clang::X86::BI__builtin_ia32_phaddw256:
+ case clang::X86::BI__builtin_ia32_phaddd128:
+ case clang::X86::BI__builtin_ia32_phaddd256: {
+ APSInt Res(LHSA + LHSB, DestUnsigned);
+ ResultElements.push_back(APValue(Res));
+ break;
+ }
+ case clang::X86::BI__builtin_ia32_phaddsw128:
+ case clang::X86::BI__builtin_ia32_phaddsw256: {
+ APSInt Res(LHSA.sadd_sat(LHSB));
+ ResultElements.push_back(APValue(Res));
+ break;
+ }
+ case clang::X86::BI__builtin_ia32_phsubw128:
+ case clang::X86::BI__builtin_ia32_phsubw256:
+ case clang::X86::BI__builtin_ia32_phsubd128:
+ case clang::X86::BI__builtin_ia32_phsubd256: {
+ APSInt Res(LHSA - LHSB, DestUnsigned);
+ ResultElements.push_back(APValue(Res));
+ break;
+ }
+ case clang::X86::BI__builtin_ia32_phsubsw128:
+ case clang::X86::BI__builtin_ia32_phsubsw256: {
+ APSInt Res(LHSA.ssub_sat(LHSB));
+ ResultElements.push_back(APValue(Res));
+ break;
+ }
+ }
+ }
+ for (unsigned I = 0; I != EltsPerLane; I += 2) {
+ APSInt RHSA = SourceRHS.getVectorElt(LaneStart + I).getInt();
+ APSInt RHSB = SourceRHS.getVectorElt(LaneStart + I + 1).getInt();
+ switch (E->getBuiltinCallee()) {
+ case clang::X86::BI__builtin_ia32_phaddw128:
+ case clang::X86::BI__builtin_ia32_phaddw256:
+ case clang::X86::BI__builtin_ia32_phaddd128:
+ case clang::X86::BI__builtin_ia32_phaddd256: {
+ APSInt Res(RHSA + RHSB, DestUnsigned);
+ ResultElements.push_back(APValue(Res));
+ break;
+ }
+ case clang::X86::BI__builtin_ia32_phaddsw128:
+ case clang::X86::BI__builtin_ia32_phaddsw256: {
+ APSInt Res(RHSA.sadd_sat(RHSB));
+ ResultElements.push_back(APValue(Res));
+ break;
+ }
+ case clang::X86::BI__builtin_ia32_phsubw128:
+ case clang::X86::BI__builtin_ia32_phsubw256:
+ case clang::X86::BI__builtin_ia32_phsubd128:
+ case clang::X86::BI__builtin_ia32_phsubd256: {
+ APSInt Res(RHSA - RHSB, DestUnsigned);
+ ResultElements.push_back(APValue(Res));
+ break;
+ }
+ case clang::X86::BI__builtin_ia32_phsubsw128:
+ case clang::X86::BI__builtin_ia32_phsubsw256: {
+ APSInt Res(RHSA.ssub_sat(RHSB));
+ ResultElements.push_back(APValue(Res));
+ break;
+ }
+ }
+ }
+ }
+ return Success(APValue(ResultElements.data(), ResultElements.size()), E);
+ }
+ case clang::X86::BI__builtin_ia32_haddpd:
+ case clang::X86::BI__builtin_ia32_haddps:
+ case clang::X86::BI__builtin_ia32_haddps256:
+ case clang::X86::BI__builtin_ia32_haddpd256:
+ case clang::X86::BI__builtin_ia32_hsubpd:
+ case clang::X86::BI__builtin_ia32_hsubps:
+ case clang::X86::BI__builtin_ia32_hsubps256:
+ case clang::X86::BI__builtin_ia32_hsubpd256: {
+ APValue SourceLHS, SourceRHS;
+ if (!EvaluateAsRValue(Info, E->getArg(0), SourceLHS) ||
+ !EvaluateAsRValue(Info, E->getArg(1), SourceRHS))
+ return false;
+ unsigned NumElts = SourceLHS.getVectorLength();
+ SmallVector<APValue, 4> ResultElements;
+ ResultElements.reserve(NumElts);
+ llvm::RoundingMode RM = getActiveRoundingMode(getEvalInfo(), E);
+ QualType DestEltTy = E->getType()->castAs<VectorType>()->getElementType();
+ unsigned EltBits = Info.Ctx.getTypeSize(DestEltTy);
+ unsigned NumLanes = NumElts * EltBits / 128;
+ unsigned NumElemsPerLane = NumElts / NumLanes;
+ unsigned HalfElemsPerLane = NumElemsPerLane / 2;
+
+ for (unsigned L = 0; L != NumElts; L += NumElemsPerLane) {
+ for (unsigned I = 0; I != HalfElemsPerLane; ++I) {
+ APFloat LHSA = SourceLHS.getVectorElt(L + (2 * I) + 0).getFloat();
+ APFloat LHSB = SourceLHS.getVectorElt(L + (2 * I) + 1).getFloat();
+ switch (E->getBuiltinCallee()) {
+ case clang::X86::BI__builtin_ia32_haddpd:
+ case clang::X86::BI__builtin_ia32_haddps:
+ case clang::X86::BI__builtin_ia32_haddps256:
+ case clang::X86::BI__builtin_ia32_haddpd256:
+ LHSA.add(LHSB, RM);
+ break;
+ case clang::X86::BI__builtin_ia32_hsubpd:
+ case clang::X86::BI__builtin_ia32_hsubps:
+ case clang::X86::BI__builtin_ia32_hsubps256:
+ case clang::X86::BI__builtin_ia32_hsubpd256:
+ LHSA.subtract(LHSB, RM);
+ break;
+ }
+ ResultElements.push_back(APValue(LHSA));
+ }
+ for (unsigned I = 0; I != HalfElemsPerLane; ++I) {
+ APFloat RHSA = SourceRHS.getVectorElt(L + (2 * I) + 0).getFloat();
+ APFloat RHSB = SourceRHS.getVectorElt(L + (2 * I) + 1).getFloat();
+ switch (E->getBuiltinCallee()) {
+ case clang::X86::BI__builtin_ia32_haddpd:
+ case clang::X86::BI__builtin_ia32_haddps:
+ case clang::X86::BI__builtin_ia32_haddps256:
+ case clang::X86::BI__builtin_ia32_haddpd256:
+ RHSA.add(RHSB, RM);
+ break;
+ case clang::X86::BI__builtin_ia32_hsubpd:
+ case clang::X86::BI__builtin_ia32_hsubps:
+ case clang::X86::BI__builtin_ia32_hsubps256:
+ case clang::X86::BI__builtin_ia32_hsubpd256:
+ RHSA.subtract(RHSB, RM);
+ break;
+ }
+ ResultElements.push_back(APValue(RHSA));
+ }
+ }
+ return Success(APValue(ResultElements.data(), ResultElements.size()), E);
+ }
case Builtin::BI__builtin_elementwise_fshl:
case Builtin::BI__builtin_elementwise_fshr: {
APValue SourceHi, SourceLo, SourceShift;
diff --git a/clang/lib/Headers/avx2intrin.h b/clang/lib/Headers/avx2intrin.h
index 4aaca2db8236a..e150aa625305f 100644
--- a/clang/lib/Headers/avx2intrin.h
+++ b/clang/lib/Headers/avx2intrin.h
@@ -834,10 +834,9 @@ _mm256_cmpgt_epi64(__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_hadd_epi16(__m256i __a, __m256i __b)
-{
- return (__m256i)__builtin_ia32_phaddw256((__v16hi)__a, (__v16hi)__b);
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_hadd_epi16(__m256i __a, __m256i __b) {
+ return (__m256i)__builtin_ia32_phaddw256((__v16hi)__a, (__v16hi)__b);
}
/// Horizontally adds the adjacent pairs of 32-bit integers from two 256-bit
@@ -866,10 +865,9 @@ _mm256_hadd_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_hadd_epi32(__m256i __a, __m256i __b)
-{
- return (__m256i)__builtin_ia32_phaddd256((__v8si)__a, (__v8si)__b);
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_hadd_epi32(__m256i __a, __m256i __b) {
+ return (__m256i)__builtin_ia32_phaddd256((__v8si)__a, (__v8si)__b);
}
/// Horizontally adds the adjacent pairs of 16-bit integers from two 256-bit
@@ -901,10 +899,9 @@ _mm256_hadd_epi32(__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_hadds_epi16(__m256i __a, __m256i __b)
-{
- return (__m256i)__builtin_ia32_phaddsw256((__v16hi)__a, (__v16hi)__b);
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_hadds_epi16(__m256i __a, __m256i __b) {
+ return (__m256i)__builtin_ia32_phaddsw256((__v16hi)__a, (__v16hi)__b);
}
/// Horizontally subtracts adjacent pairs of 16-bit integers from two 256-bit
@@ -937,10 +934,9 @@ _mm256_hadds_epi16(__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
diff erences.
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_hsub_epi16(__m256i __a, __m256i __b)
-{
- return (__m256i)__builtin_ia32_phsubw256((__v16hi)__a, (__v16hi)__b);
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_hsub_epi16(__m256i __a, __m256i __b) {
+ return (__m256i)__builtin_ia32_phsubw256((__v16hi)__a, (__v16hi)__b);
}
/// Horizontally subtracts adjacent pairs of 32-bit integers from two 256-bit
@@ -969,10 +965,9 @@ _mm256_hsub_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
diff erences.
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_hsub_epi32(__m256i __a, __m256i __b)
-{
- return (__m256i)__builtin_ia32_phsubd256((__v8si)__a, (__v8si)__b);
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_hsub_epi32(__m256i __a, __m256i __b) {
+ return (__m256i)__builtin_ia32_phsubd256((__v8si)__a, (__v8si)__b);
}
/// Horizontally subtracts adjacent pairs of 16-bit integers from two 256-bit
@@ -1005,10 +1000,9 @@ _mm256_hsub_epi32(__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
diff erences.
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_hsubs_epi16(__m256i __a, __m256i __b)
-{
- return (__m256i)__builtin_ia32_phsubsw256((__v16hi)__a, (__v16hi)__b);
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_hsubs_epi16(__m256i __a, __m256i __b) {
+ return (__m256i)__builtin_ia32_phsubsw256((__v16hi)__a, (__v16hi)__b);
}
/// Multiplies each unsigned byte from the 256-bit integer vector in \a __a
diff --git a/clang/lib/Headers/avxintrin.h b/clang/lib/Headers/avxintrin.h
index 123fa7933c4f8..696ec31a1ee3c 100644
--- a/clang/lib/Headers/avxintrin.h
+++ b/clang/lib/Headers/avxintrin.h
@@ -694,9 +694,8 @@ _mm256_xor_ps(__m256 __a, __m256 __b)
/// elements of a vector of [4 x double].
/// \returns A 256-bit vector of [4 x double] containing the horizontal sums of
/// both operands.
-static __inline __m256d __DEFAULT_FN_ATTRS
-_mm256_hadd_pd(__m256d __a, __m256d __b)
-{
+static __inline __m256d __DEFAULT_FN_ATTRS_CONSTEXPR
+_mm256_hadd_pd(__m256d __a, __m256d __b) {
return (__m256d)__builtin_ia32_haddpd256((__v4df)__a, (__v4df)__b);
}
@@ -717,9 +716,8 @@ _mm256_hadd_pd(__m256d __a, __m256d __b)
/// index 2, 3, 6, 7 of a vector of [8 x float].
/// \returns A 256-bit vector of [8 x float] containing the horizontal sums of
/// both operands.
-static __inline __m256 __DEFAULT_FN_ATTRS
-_mm256_hadd_ps(__m256 __a, __m256 __b)
-{
+static __inline __m256 __DEFAULT_FN_ATTRS_CONSTEXPR _mm256_hadd_ps(__m256 __a,
+ __m256 __b) {
return (__m256)__builtin_ia32_haddps256((__v8sf)__a, (__v8sf)__b);
}
@@ -740,9 +738,8 @@ _mm256_hadd_ps(__m256 __a, __m256 __b)
/// odd-indexed elements of a vector of [4 x double].
/// \returns A 256-bit vector of [4 x double] containing the horizontal
///
diff erences of both operands.
-static __inline __m256d __DEFAULT_FN_ATTRS
-_mm256_hsub_pd(__m256d __a, __m256d __b)
-{
+static __inline __m256d __DEFAULT_FN_ATTRS_CONSTEXPR
+_mm256_hsub_pd(__m256d __a, __m256d __b) {
return (__m256d)__builtin_ia32_hsubpd256((__v4df)__a, (__v4df)__b);
}
@@ -763,9 +760,8 @@ _mm256_hsub_pd(__m256d __a, __m256d __b)
/// elements with index 2, 3, 6, 7 of a vector of [8 x float].
/// \returns A 256-bit vector of [8 x float] containing the horizontal
///
diff erences of both operands.
-static __inline __m256 __DEFAULT_FN_ATTRS
-_mm256_hsub_ps(__m256 __a, __m256 __b)
-{
+static __inline __m256 __DEFAULT_FN_ATTRS_CONSTEXPR _mm256_hsub_ps(__m256 __a,
+ __m256 __b) {
return (__m256)__builtin_ia32_hsubps256((__v8sf)__a, (__v8sf)__b);
}
diff --git a/clang/lib/Headers/pmmintrin.h b/clang/lib/Headers/pmmintrin.h
index f0c9b2ba38b0c..42bd343e326de 100644
--- a/clang/lib/Headers/pmmintrin.h
+++ b/clang/lib/Headers/pmmintrin.h
@@ -83,9 +83,8 @@ _mm_addsub_ps(__m128 __a, __m128 __b)
/// destination.
/// \returns A 128-bit vector of [4 x float] containing the horizontal sums of
/// both operands.
-static __inline__ __m128 __DEFAULT_FN_ATTRS
-_mm_hadd_ps(__m128 __a, __m128 __b)
-{
+static __inline__ __m128 __DEFAULT_FN_ATTRS_CONSTEXPR _mm_hadd_ps(__m128 __a,
+ __m128 __b) {
return __builtin_ia32_haddps((__v4sf)__a, (__v4sf)__b);
}
@@ -106,9 +105,8 @@ _mm_hadd_ps(__m128 __a, __m128 __b)
/// bits of the destination.
/// \returns A 128-bit vector of [4 x float] containing the horizontal
///
diff erences of both operands.
-static __inline__ __m128 __DEFAULT_FN_ATTRS
-_mm_hsub_ps(__m128 __a, __m128 __b)
-{
+static __inline__ __m128 __DEFAULT_FN_ATTRS_CONSTEXPR _mm_hsub_ps(__m128 __a,
+ __m128 __b) {
return __builtin_ia32_hsubps((__v4sf)__a, (__v4sf)__b);
}
@@ -168,9 +166,8 @@ _mm_moveldup_ps(__m128 __a)
/// A 128-bit vector of [2 x double] containing the right source operand.
/// \returns A 128-bit vector of [2 x double] containing the alternating sums
/// and
diff erences of both operands.
-static __inline__ __m128d __DEFAULT_FN_ATTRS
-_mm_addsub_pd(__m128d __a, __m128d __b)
-{
+static __inline__ __m128d __DEFAULT_FN_ATTRS_CONSTEXPR
+_mm_addsub_pd(__m128d __a, __m128d __b) {
return __builtin_ia32_addsubpd((__v2df)__a, (__v2df)__b);
}
@@ -191,9 +188,8 @@ _mm_addsub_pd(__m128d __a, __m128d __b)
/// destination.
/// \returns A 128-bit vector of [2 x double] containing the horizontal sums of
/// both operands.
-static __inline__ __m128d __DEFAULT_FN_ATTRS
-_mm_hadd_pd(__m128d __a, __m128d __b)
-{
+static __inline__ __m128d __DEFAULT_FN_ATTRS_CONSTEXPR
+_mm_hadd_pd(__m128d __a, __m128d __b) {
return __builtin_ia32_haddpd((__v2df)__a, (__v2df)__b);
}
@@ -214,9 +210,8 @@ _mm_hadd_pd(__m128d __a, __m128d __b)
/// the destination.
/// \returns A 128-bit vector of [2 x double] containing the horizontal
///
diff erences of both operands.
-static __inline__ __m128d __DEFAULT_FN_ATTRS
-_mm_hsub_pd(__m128d __a, __m128d __b)
-{
+static __inline__ __m128d __DEFAULT_FN_ATTRS_CONSTEXPR
+_mm_hsub_pd(__m128d __a, __m128d __b) {
return __builtin_ia32_hsubpd((__v2df)__a, (__v2df)__b);
}
diff --git a/clang/lib/Headers/tmmintrin.h b/clang/lib/Headers/tmmintrin.h
index 3fc9f9834baa9..9d007c8f3b305 100644
--- a/clang/lib/Headers/tmmintrin.h
+++ b/clang/lib/Headers/tmmintrin.h
@@ -202,10 +202,9 @@ _mm_abs_epi32(__m128i __a) {
/// destination.
/// \returns A 128-bit vector of [8 x i16] containing the horizontal sums of
/// both operands.
-static __inline__ __m128i __DEFAULT_FN_ATTRS
-_mm_hadd_epi16(__m128i __a, __m128i __b)
-{
- return (__m128i)__builtin_ia32_phaddw128((__v8hi)__a, (__v8hi)__b);
+static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR
+_mm_hadd_epi16(__m128i __a, __m128i __b) {
+ return (__m128i)__builtin_ia32_phaddw128((__v8hi)__a, (__v8hi)__b);
}
/// Horizontally adds the adjacent pairs of values contained in 2 packed
@@ -225,10 +224,9 @@ _mm_hadd_epi16(__m128i __a, __m128i __b)
/// destination.
/// \returns A 128-bit vector of [4 x i32] containing the horizontal sums of
/// both operands.
-static __inline__ __m128i __DEFAULT_FN_ATTRS
-_mm_hadd_epi32(__m128i __a, __m128i __b)
-{
- return (__m128i)__builtin_ia32_phaddd128((__v4si)__a, (__v4si)__b);
+static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR
+_mm_hadd_epi32(__m128i __a, __m128i __b) {
+ return (__m128i)__builtin_ia32_phaddd128((__v4si)__a, (__v4si)__b);
}
/// Horizontally adds the adjacent pairs of values contained in 2 packed
@@ -248,11 +246,10 @@ _mm_hadd_epi32(__m128i __a, __m128i __b)
/// destination.
/// \returns A 64-bit vector of [4 x i16] containing the horizontal sums of both
/// operands.
-static __inline__ __m64 __DEFAULT_FN_ATTRS
-_mm_hadd_pi16(__m64 __a, __m64 __b)
-{
- return __trunc64(__builtin_ia32_phaddw128(
- (__v8hi)__builtin_shufflevector(__a, __b, 0, 1), (__v8hi){}));
+static __inline__ __m64 __DEFAULT_FN_ATTRS_CONSTEXPR _mm_hadd_pi16(__m64 __a,
+ __m64 __b) {
+ return __trunc64(__builtin_ia32_phaddw128(
+ (__v8hi)__builtin_shufflevector(__a, __b, 0, 1), (__v8hi){}));
}
/// Horizontally adds the adjacent pairs of values contained in 2 packed
@@ -272,11 +269,10 @@ _mm_hadd_pi16(__m64 __a, __m64 __b)
/// destination.
/// \returns A 64-bit vector of [2 x i32] containing the horizontal sums of both
/// operands.
-static __inline__ __m64 __DEFAULT_FN_ATTRS
-_mm_hadd_pi32(__m64 __a, __m64 __b)
-{
- return __trunc64(__builtin_ia32_phaddd128(
- (__v4si)__builtin_shufflevector(__a, __b, 0, 1), (__v4si){}));
+static __inline__ __m64 __DEFAULT_FN_ATTRS_CONSTEXPR _mm_hadd_pi32(__m64 __a,
+ __m64 __b) {
+ return __trunc64(__builtin_ia32_phaddd128(
+ (__v4si)__builtin_shufflevector(__a, __b, 0, 1), (__v4si){}));
}
/// Horizontally adds, with saturation, the adjacent pairs of values contained
@@ -299,10 +295,9 @@ _mm_hadd_pi32(__m64 __a, __m64 __b)
/// destination.
/// \returns A 128-bit vector of [8 x i16] containing the horizontal saturated
/// sums of both operands.
-static __inline__ __m128i __DEFAULT_FN_ATTRS
-_mm_hadds_epi16(__m128i __a, __m128i __b)
-{
- return (__m128i)__builtin_ia32_phaddsw128((__v8hi)__a, (__v8hi)__b);
+static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR
+_mm_hadds_epi16(__m128i __a, __m128i __b) {
+ return (__m128i)__builtin_ia32_phaddsw128((__v8hi)__a, (__v8hi)__b);
}
/// Horizontally adds, with saturation, the adjacent pairs of values contained
@@ -325,11 +320,10 @@ _mm_hadds_epi16(__m128i __a, __m128i __b)
/// destination.
/// \returns A 64-bit vector of [4 x i16] containing the horizontal saturated
/// sums of both operands.
-static __inline__ __m64 __DEFAULT_FN_ATTRS
-_mm_hadds_pi16(__m64 __a, __m64 __b)
-{
- return __trunc64(__builtin_ia32_phaddsw128(
- (__v8hi)__builtin_shufflevector(__a, __b, 0, 1), (__v8hi){}));
+static __inline__ __m64 __DEFAULT_FN_ATTRS_CONSTEXPR _mm_hadds_pi16(__m64 __a,
+ __m64 __b) {
+ return __trunc64(__builtin_ia32_phaddsw128(
+ (__v8hi)__builtin_shufflevector(__a, __b, 0, 1), (__v8hi){}));
}
/// Horizontally subtracts the adjacent pairs of values contained in 2
@@ -349,10 +343,9 @@ _mm_hadds_pi16(__m64 __a, __m64 __b)
/// the destination.
/// \returns A 128-bit vector of [8 x i16] containing the horizontal
diff erences
/// of both operands.
-static __inline__ __m128i __DEFAULT_FN_ATTRS
-_mm_hsub_epi16(__m128i __a, __m128i __b)
-{
- return (__m128i)__builtin_ia32_phsubw128((__v8hi)__a, (__v8hi)__b);
+static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR
+_mm_hsub_epi16(__m128i __a, __m128i __b) {
+ return (__m128i)__builtin_ia32_phsubw128((__v8hi)__a, (__v8hi)__b);
}
/// Horizontally subtracts the adjacent pairs of values contained in 2
@@ -372,10 +365,9 @@ _mm_hsub_epi16(__m128i __a, __m128i __b)
/// the destination.
/// \returns A 128-bit vector of [4 x i32] containing the horizontal
diff erences
/// of both operands.
-static __inline__ __m128i __DEFAULT_FN_ATTRS
-_mm_hsub_epi32(__m128i __a, __m128i __b)
-{
- return (__m128i)__builtin_ia32_phsubd128((__v4si)__a, (__v4si)__b);
+static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR
+_mm_hsub_epi32(__m128i __a, __m128i __b) {
+ return (__m128i)__builtin_ia32_phsubd128((__v4si)__a, (__v4si)__b);
}
/// Horizontally subtracts the adjacent pairs of values contained in 2
@@ -395,11 +387,10 @@ _mm_hsub_epi32(__m128i __a, __m128i __b)
/// the destination.
/// \returns A 64-bit vector of [4 x i16] containing the horizontal
diff erences
/// of both operands.
-static __inline__ __m64 __DEFAULT_FN_ATTRS
-_mm_hsub_pi16(__m64 __a, __m64 __b)
-{
- return __trunc64(__builtin_ia32_phsubw128(
- (__v8hi)__builtin_shufflevector(__a, __b, 0, 1), (__v8hi){}));
+static __inline__ __m64 __DEFAULT_FN_ATTRS_CONSTEXPR _mm_hsub_pi16(__m64 __a,
+ __m64 __b) {
+ return __trunc64(__builtin_ia32_phsubw128(
+ (__v8hi)__builtin_shufflevector(__a, __b, 0, 1), (__v8hi){}));
}
/// Horizontally subtracts the adjacent pairs of values contained in 2
@@ -419,11 +410,10 @@ _mm_hsub_pi16(__m64 __a, __m64 __b)
/// the destination.
/// \returns A 64-bit vector of [2 x i32] containing the horizontal
diff erences
/// of both operands.
-static __inline__ __m64 __DEFAULT_FN_ATTRS
-_mm_hsub_pi32(__m64 __a, __m64 __b)
-{
- return __trunc64(__builtin_ia32_phsubd128(
- (__v4si)__builtin_shufflevector(__a, __b, 0, 1), (__v4si){}));
+static __inline__ __m64 __DEFAULT_FN_ATTRS_CONSTEXPR _mm_hsub_pi32(__m64 __a,
+ __m64 __b) {
+ return __trunc64(__builtin_ia32_phsubd128(
+ (__v4si)__builtin_shufflevector(__a, __b, 0, 1), (__v4si){}));
}
/// Horizontally subtracts, with saturation, the adjacent pairs of values
@@ -446,10 +436,9 @@ _mm_hsub_pi32(__m64 __a, __m64 __b)
/// the destination.
/// \returns A 128-bit vector of [8 x i16] containing the horizontal saturated
///
diff erences of both operands.
-static __inline__ __m128i __DEFAULT_FN_ATTRS
-_mm_hsubs_epi16(__m128i __a, __m128i __b)
-{
- return (__m128i)__builtin_ia32_phsubsw128((__v8hi)__a, (__v8hi)__b);
+static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR
+_mm_hsubs_epi16(__m128i __a, __m128i __b) {
+ return (__m128i)__builtin_ia32_phsubsw128((__v8hi)__a, (__v8hi)__b);
}
/// Horizontally subtracts, with saturation, the adjacent pairs of values
@@ -472,11 +461,10 @@ _mm_hsubs_epi16(__m128i __a, __m128i __b)
/// the destination.
/// \returns A 64-bit vector of [4 x i16] containing the horizontal saturated
///
diff erences of both operands.
-static __inline__ __m64 __DEFAULT_FN_ATTRS
-_mm_hsubs_pi16(__m64 __a, __m64 __b)
-{
- return __trunc64(__builtin_ia32_phsubsw128(
- (__v8hi)__builtin_shufflevector(__a, __b, 0, 1), (__v8hi){}));
+static __inline__ __m64 __DEFAULT_FN_ATTRS_CONSTEXPR _mm_hsubs_pi16(__m64 __a,
+ __m64 __b) {
+ return __trunc64(__builtin_ia32_phsubsw128(
+ (__v8hi)__builtin_shufflevector(__a, __b, 0, 1), (__v8hi){}));
}
/// Multiplies corresponding pairs of packed 8-bit unsigned integer
@@ -556,10 +544,9 @@ _mm_maddubs_pi16(__m64 __a, __m64 __b) {
/// A 128-bit vector of [8 x i16] containing one of the source operands.
/// \returns A 128-bit vector of [8 x i16] containing the rounded and scaled
/// products of both operands.
-static __inline__ __m128i __DEFAULT_FN_ATTRS
-_mm_mulhrs_epi16(__m128i __a, __m128i __b)
-{
- return (__m128i)__builtin_ia32_pmulhrsw128((__v8hi)__a, (__v8hi)__b);
+static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_mulhrs_epi16(__m128i __a,
+ __m128i __b) {
+ return (__m128i)__builtin_ia32_pmulhrsw128((__v8hi)__a, (__v8hi)__b);
}
/// Multiplies packed 16-bit signed integer values, truncates the 32-bit
diff --git a/clang/test/CodeGen/X86/avx-builtins.c b/clang/test/CodeGen/X86/avx-builtins.c
index 8f3d4590d0b3e..bcffd861fd7f7 100644
--- a/clang/test/CodeGen/X86/avx-builtins.c
+++ b/clang/test/CodeGen/X86/avx-builtins.c
@@ -1100,6 +1100,7 @@ __m256d test_mm256_hadd_pd(__m256d A, __m256d B) {
// CHECK: call {{.*}}<4 x double> @llvm.x86.avx.hadd.pd.256(<4 x double> %{{.*}}, <4 x double> %{{.*}})
return _mm256_hadd_pd(A, B);
}
+TEST_CONSTEXPR(match_m256d(_mm256_hadd_pd((__m256d){+1.0, +2.0, +3.0, +4.0}, (__m256d){+5.0, +6.0, +7.0, +8.0}), +3.0, +11.0, +7.0, +15.0));
__m256 test_mm256_hadd_ps(__m256 A, __m256 B) {
// CHECK-LABEL: test_mm256_hadd_ps
@@ -1107,17 +1108,27 @@ __m256 test_mm256_hadd_ps(__m256 A, __m256 B) {
return _mm256_hadd_ps(A, B);
}
+TEST_CONSTEXPR(match_m256(_mm256_hadd_ps(
+ (__m256){+1.0f, +2.0f, +3.0f, +4.0f, +5.0f, +6.0f, +7.0f, +8.0f},
+ (__m256){+9.0f, +10.0f, +11.0f, +12.0f, +13.0f, +14.0f, +15.0f, +16.0f}),
+ +3.0f, +7.0f, +19.0f, +23.0f, +11.0f, +15.0f, +27.0f, +31.0f));
+
__m256d test_mm256_hsub_pd(__m256d A, __m256d B) {
// CHECK-LABEL: test_mm256_hsub_pd
// CHECK: call {{.*}}<4 x double> @llvm.x86.avx.hsub.pd.256(<4 x double> %{{.*}}, <4 x double> %{{.*}})
return _mm256_hsub_pd(A, B);
}
+TEST_CONSTEXPR(match_m256d(_mm256_hsub_pd((__m256d){+1.0, +2.0, +4.0, +3.0}, (__m256d){+10.0, +6.0, +16.0, +8.0}), -1.0,+4.0,+1.0,+8.0));
__m256 test_mm256_hsub_ps(__m256 A, __m256 B) {
// CHECK-LABEL: test_mm256_hsub_ps
// CHECK: call {{.*}}<8 x float> @llvm.x86.avx.hsub.ps.256(<8 x float> %{{.*}}, <8 x float> %{{.*}})
return _mm256_hsub_ps(A, B);
}
+TEST_CONSTEXPR(match_m256(_mm256_hsub_ps(
+ (__m256){1.0f, 2.0f, 4.0f, 3.0f, 5.0f, 7.0f, 7.0f, 5.0f},
+ (__m256){6.0f, 9.0f, 11.0f, 8.0f, 13.0f, 17.0f, 15.0f, 11.0f}),
+ -1.0f, 1.0f, -3.0f, 3.0f, -2.0f, 2.0f, -4.0f, 4.0f));
__m256i test_mm256_insert_epi8(__m256i x, char b) {
// CHECK-LABEL: test_mm256_insert_epi8
diff --git a/clang/test/CodeGen/X86/avx2-builtins.c b/clang/test/CodeGen/X86/avx2-builtins.c
index 55f18f947b96f..dc64f96f3001d 100644
--- a/clang/test/CodeGen/X86/avx2-builtins.c
+++ b/clang/test/CodeGen/X86/avx2-builtins.c
@@ -485,36 +485,60 @@ __m256i test_mm256_hadd_epi16(__m256i a, __m256i b) {
// CHECK: call <16 x i16> @llvm.x86.avx2.phadd.w(<16 x i16> %{{.*}}, <16 x i16> %{{.*}})
return _mm256_hadd_epi16(a, b);
}
+TEST_CONSTEXPR(match_v16hi(_mm256_hadd_epi16(
+ (__m256i)(__v16hi){1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16},
+ (__m256i)(__v16hi){17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32}),
+ 3,7,11,15,35,39,43,47,19,23,27,31,51,55,59,63));
__m256i test_mm256_hadd_epi32(__m256i a, __m256i b) {
// CHECK-LABEL: test_mm256_hadd_epi32
// CHECK: call <8 x i32> @llvm.x86.avx2.phadd.d(<8 x i32> %{{.*}}, <8 x i32> %{{.*}})
return _mm256_hadd_epi32(a, b);
}
+TEST_CONSTEXPR(match_v8si(_mm256_hadd_epi32(
+ (__m256i)(__v8si){10, 20, 30, 40, 50, 60, 70, 80},
+ (__m256i)(__v8si){5, 15, 25, 35, 45, 55, 65, 75}),
+ 30,70,20,60,110,150,100,140));
__m256i test_mm256_hadds_epi16(__m256i a, __m256i b) {
// CHECK-LABEL: test_mm256_hadds_epi16
// CHECK:call <16 x i16> @llvm.x86.avx2.phadd.sw(<16 x i16> %{{.*}}, <16 x i16> %{{.*}})
return _mm256_hadds_epi16(a, b);
}
+TEST_CONSTEXPR(match_v16hi( _mm256_hadds_epi16(
+ (__m256i)(__v16hi){32767, 32767, 1,2,3,4,5,6,7,8,9,10,11,12,13,14},
+ (__m256i)(__v16hi){19,20,21,22,23,24,25,26,27,28,29,30,31,32, 32767, 5}),
+ 32767, 3,7,11, 39,43,47,51,15,19,23,27, 55,59,63, 32767));
__m256i test_mm256_hsub_epi16(__m256i a, __m256i b) {
// CHECK-LABEL: test_mm256_hsub_epi16
// CHECK: call <16 x i16> @llvm.x86.avx2.phsub.w(<16 x i16> %{{.*}}, <16 x i16> %{{.*}})
return _mm256_hsub_epi16(a, b);
}
+TEST_CONSTEXPR(match_v16hi(_mm256_hsub_epi16(
+ (__m256i)(__v16hi){2,1,1,2,5,3,3,5,7,4,4,7,9,5,5,9},
+ (__m256i)(__v16hi){10,5,5,10,12,6,6,12,21,14,14,21,24,16,16,24}),
+ 1,-1,2,-2,5,-5,6,-6,3,-3,4,-4, 7,-7,8,-8));
__m256i test_mm256_hsub_epi32(__m256i a, __m256i b) {
// CHECK-LABEL: test_mm256_hsub_epi32
// CHECK: call <8 x i32> @llvm.x86.avx2.phsub.d(<8 x i32> %{{.*}}, <8 x i32> %{{.*}})
return _mm256_hsub_epi32(a, b);
}
+TEST_CONSTEXPR(match_v8si(_mm256_hsub_epi32(
+ (__m256i)(__v8si){10, 20, 30,50,60,90,100,140},
+ (__m256i)(__v8si){200,150,260,200,420,350,800,720}),
+ -10,-20,50,60, -30,-40, 70,80));
__m256i test_mm256_hsubs_epi16(__m256i a, __m256i b) {
// CHECK-LABEL: test_mm256_hsubs_epi16
// CHECK:call <16 x i16> @llvm.x86.avx2.phsub.sw(<16 x i16> %{{.*}}, <16 x i16> %{{.*}})
return _mm256_hsubs_epi16(a, b);
}
+TEST_CONSTEXPR(match_v16hi(_mm256_hsubs_epi16(
+ (__m256i)(__v16hi){32726, -100, 3, 2, 6, 4, 8, 5,15,10 ,21, 14, 27, 18, 100, 90},
+ (__m256i)(__v16hi){40, 20, 100, 70, 200,150, 100,40, 1000,900,300,150, 500,300, 1, 1}),
+ 32767, 1, 2, 3, 20, 30, 50, 60, 5, 7, 9, 10, 100, 150, 200, 0));
__m128i test_mm_i32gather_epi32(int const *b, __m128i c) {
// CHECK-LABEL: test_mm_i32gather_epi32
diff --git a/clang/test/CodeGen/X86/mmx-builtins.c b/clang/test/CodeGen/X86/mmx-builtins.c
index 2b45b920cf2bc..a1e05a1ba205f 100644
--- a/clang/test/CodeGen/X86/mmx-builtins.c
+++ b/clang/test/CodeGen/X86/mmx-builtins.c
@@ -312,36 +312,42 @@ __m64 test_mm_hadd_pi16(__m64 a, __m64 b) {
// CHECK: call <8 x i16> @llvm.x86.ssse3.phadd.w.128(
return _mm_hadd_pi16(a, b);
}
+TEST_CONSTEXPR(match_v4hi(_mm_hadd_pi16((__m64)(__v4hi){1,2,3,4},(__m64)(__v4hi){5,6,7,8}),3,7,11,15));
__m64 test_mm_hadd_pi32(__m64 a, __m64 b) {
// CHECK-LABEL: test_mm_hadd_pi32
// CHECK: call <4 x i32> @llvm.x86.ssse3.phadd.d.128(
return _mm_hadd_pi32(a, b);
}
+TEST_CONSTEXPR(match_v2si(_mm_hadd_pi32((__m64)(__v2si){1,2},(__m64)(__v2si){3,4}),3,7));
__m64 test_mm_hadds_pi16(__m64 a, __m64 b) {
// CHECK-LABEL: test_mm_hadds_pi16
// CHECK: call <8 x i16> @llvm.x86.ssse3.phadd.sw.128(
return _mm_hadds_pi16(a, b);
}
+TEST_CONSTEXPR(match_v4hi(_mm_hadds_pi16((__m64)(__v4hi){32767, 32767, 1,3},(__m64)(__v4hi){-1,3, 40, 60}),32767, 4, 2,100));
__m64 test_mm_hsub_pi16(__m64 a, __m64 b) {
// CHECK-LABEL: test_mm_hsub_pi16
// CHECK: call <8 x i16> @llvm.x86.ssse3.phsub.w.128(
return _mm_hsub_pi16(a, b);
}
+TEST_CONSTEXPR(match_v4hi(_mm_hsub_pi16((__m64)(__v4hi){1,2,4,3},(__m64)(__v4hi){10,5,0,-10}),-1,1,5,10));
__m64 test_mm_hsub_pi32(__m64 a, __m64 b) {
// CHECK-LABEL: test_mm_hsub_pi32
// CHECK: call <4 x i32> @llvm.x86.ssse3.phsub.d.128(
return _mm_hsub_pi32(a, b);
}
+TEST_CONSTEXPR(match_v2si(_mm_hsub_pi32((__m64)(__v2si){1,2},(__m64)(__v2si){4,3}),-1,1));
__m64 test_mm_hsubs_pi16(__m64 a, __m64 b) {
// CHECK-LABEL: test_mm_hsubs_pi16
// CHECK: call <8 x i16> @llvm.x86.ssse3.phsub.sw.128(
return _mm_hsubs_pi16(a, b);
}
+TEST_CONSTEXPR(match_v4hi(_mm_hsubs_pi16((__m64)(__v4hi){32767, 32767, 5, -32767},(__m64)(__v4hi){4,5,10,5}),0,32767,-1,5));
__m64 test_mm_insert_pi16(__m64 a, int d) {
// CHECK-LABEL: test_mm_insert_pi16
diff --git a/clang/test/CodeGen/X86/sse3-builtins.c b/clang/test/CodeGen/X86/sse3-builtins.c
index c53afc56e7246..a82dd4080670b 100644
--- a/clang/test/CodeGen/X86/sse3-builtins.c
+++ b/clang/test/CodeGen/X86/sse3-builtins.c
@@ -31,24 +31,28 @@ __m128d test_mm_hadd_pd(__m128d A, __m128d B) {
// CHECK: call {{.*}}<2 x double> @llvm.x86.sse3.hadd.pd(<2 x double> %{{.*}}, <2 x double> %{{.*}})
return _mm_hadd_pd(A, B);
}
+TEST_CONSTEXPR(match_m128d(_mm_hadd_pd((__m128d){+1.0, +2.0}, (__m128d){+3.0, +4.0}), +3.0, +7.0));
__m128 test_mm_hadd_ps(__m128 A, __m128 B) {
// CHECK-LABEL: test_mm_hadd_ps
// CHECK: call {{.*}}<4 x float> @llvm.x86.sse3.hadd.ps(<4 x float> %{{.*}}, <4 x float> %{{.*}})
return _mm_hadd_ps(A, B);
}
+TEST_CONSTEXPR(match_m128(_mm_hadd_ps((__m128){+1.0f, +2.0f, +3.0f, +4.0f}, (__m128){+5.0f,+6.0f,+7.0f,+8.0f}), +3.0f, +7.0f, +11.0f, +15.0f));
__m128d test_mm_hsub_pd(__m128d A, __m128d B) {
// CHECK-LABEL: test_mm_hsub_pd
// CHECK: call {{.*}}<2 x double> @llvm.x86.sse3.hsub.pd(<2 x double> %{{.*}}, <2 x double> %{{.*}})
return _mm_hsub_pd(A, B);
}
+TEST_CONSTEXPR(match_m128d(_mm_hsub_pd((__m128d){+1.0, +2.0}, (__m128d){+4.0, +3.0}), -1.0, +1.0));
__m128 test_mm_hsub_ps(__m128 A, __m128 B) {
// CHECK-LABEL: test_mm_hsub_ps
// CHECK: call {{.*}}<4 x float> @llvm.x86.sse3.hsub.ps(<4 x float> %{{.*}}, <4 x float> %{{.*}})
return _mm_hsub_ps(A, B);
}
+TEST_CONSTEXPR(match_m128(_mm_hsub_ps((__m128){+1.0f, +2.0f, +4.0f, +3.0f}, (__m128){+5.0f,+7.0f,+10.0f,+8.0f}), -1.0f, +1.0f, -2.0f, +2.0f));
__m128i test_mm_lddqu_si128(__m128i const* P) {
// CHECK-LABEL: test_mm_lddqu_si128
diff --git a/clang/test/CodeGen/X86/ssse3-builtins.c b/clang/test/CodeGen/X86/ssse3-builtins.c
index 588576879903b..e6235998334ee 100644
--- a/clang/test/CodeGen/X86/ssse3-builtins.c
+++ b/clang/test/CodeGen/X86/ssse3-builtins.c
@@ -60,36 +60,43 @@ __m128i test_mm_hadd_epi16(__m128i a, __m128i b) {
// CHECK: call <8 x i16> @llvm.x86.ssse3.phadd.w.128(<8 x i16> %{{.*}}, <8 x i16> %{{.*}})
return _mm_hadd_epi16(a, b);
}
+TEST_CONSTEXPR(match_v8hi(_mm_hadd_epi16((__m128i)(__v8hi){1,2,3,4,5,6,7,8}, (__m128i)(__v8hi){17,18,19,20,21,22,23,24}), 3,7,11,15,35,39,43,47));
__m128i test_mm_hadd_epi32(__m128i a, __m128i b) {
// CHECK-LABEL: test_mm_hadd_epi32
// CHECK: call <4 x i32> @llvm.x86.ssse3.phadd.d.128(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
return _mm_hadd_epi32(a, b);
}
+TEST_CONSTEXPR(match_v4si(_mm_hadd_epi32((__m128i)(__v4si){1,2,3,4}, (__m128i)(__v4si){5,6,7,8}), 3,7,11,15));
__m128i test_mm_hadds_epi16(__m128i a, __m128i b) {
// CHECK-LABEL: test_mm_hadds_epi16
// CHECK: call <8 x i16> @llvm.x86.ssse3.phadd.sw.128(<8 x i16> %{{.*}}, <8 x i16> %{{.*}})
return _mm_hadds_epi16(a, b);
}
+TEST_CONSTEXPR(match_v8hi(_mm_hadds_epi16((__m128i)(__v8hi){30000,30000,-1,2,-3,3,1,4}, (__m128i)(__v8hi){2,6,1,9,-4,16,7,8}), 32767, 1,0,5,8,10,12,15));
+
__m128i test_mm_hsub_epi16(__m128i a, __m128i b) {
// CHECK-LABEL: test_mm_hsub_epi16
// CHECK: call <8 x i16> @llvm.x86.ssse3.phsub.w.128(<8 x i16> %{{.*}}, <8 x i16> %{{.*}})
return _mm_hsub_epi16(a, b);
}
+TEST_CONSTEXPR(match_v8hi(_mm_hsub_epi16((__m128i)(__v8hi){20,15,16,12,9,6,4,2}, (__m128i)(__v8hi){3,2,1,1,4,5,0,2}), 5,4,3,2,1,0,-1,-2));
__m128i test_mm_hsub_epi32(__m128i a, __m128i b) {
// CHECK-LABEL: test_mm_hsub_epi32
// CHECK: call <4 x i32> @llvm.x86.ssse3.phsub.d.128(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
return _mm_hsub_epi32(a, b);
}
+TEST_CONSTEXPR(match_v4si(_mm_hsub_epi32((__m128i)(__v4si){4,3,1,1}, (__m128i)(__v4si){7,5,10,5}), 1,0,2,5));
__m128i test_mm_hsubs_epi16(__m128i a, __m128i b) {
// CHECK-LABEL: test_mm_hsubs_epi16
// CHECK: call <8 x i16> @llvm.x86.ssse3.phsub.sw.128(<8 x i16> %{{.*}}, <8 x i16> %{{.*}})
return _mm_hsubs_epi16(a, b);
}
+TEST_CONSTEXPR(match_v8hi(_mm_hsubs_epi16((__m128i)(__v8hi){32767, -15,16,12,9,6,4,2},(__m128i)(__v8hi){3,2,1,1,4,5,0,2}), 32767,4,3,2,1,0,-1,-2));
__m128i test_mm_maddubs_epi16(__m128i a, __m128i b) {
// CHECK-LABEL: test_mm_maddubs_epi16
More information about the cfe-commits
mailing list