FEX-Emu/FEX
1
0
Fork 0
mirror of https://github.com/FEX-Emu/FEX.git synced 2024-12-14 01:18:46 +00:00
Code Issues Actions 6 Packages Projects Releases Wiki Activity

AVX128: Implement support for imm vpermil{ps,pd}

Browse Source
This commit is contained in:
Ryan Houdek 2024-06-19 04:10:34 -07:00 committed by Alyssa Rosenzweig
parent 96a51650b1
commit d6bf276b5a
2 changed files with 41 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
FEXCore/Source/Interface/Core/OpcodeDispatcher.h
View File

@ -1157,6 +1157,9 @@ public:
template<size_t ElementSize>
void AVX128_VSHUF(OpcodeArgs);
template<size_t ElementSize>
void AVX128_VPERMILImm(OpcodeArgs);
// End of AVX 128-bit implementation
void InvalidOp(OpcodeArgs);

40
FEXCore/Source/Interface/Core/OpcodeDispatcher/AVX_128.cpp
View File

@ -336,8 +336,8 @@ void OpDispatchBuilder::InstallAVX128Handlers() {
{OPD(3, 0b01, 0x00), 1, &OpDispatchBuilder::AVX128_VPERMQ},
{OPD(3, 0b01, 0x01), 1, &OpDispatchBuilder::AVX128_VPERMQ},
// TODO: {OPD(3, 0b01, 0x02), 1, &OpDispatchBuilder::VPBLENDDOp},
// TODO: {OPD(3, 0b01, 0x04), 1, &OpDispatchBuilder::VPERMILImmOp<4>},
// TODO: {OPD(3, 0b01, 0x05), 1, &OpDispatchBuilder::VPERMILImmOp<8>},
{OPD(3, 0b01, 0x04), 1, &OpDispatchBuilder::AVX128_VPERMILImm<4>},
{OPD(3, 0b01, 0x05), 1, &OpDispatchBuilder::AVX128_VPERMILImm<8>},
// TODO: {OPD(3, 0b01, 0x06), 1, &OpDispatchBuilder::VPERM2Op},
{OPD(3, 0b01, 0x08), 1, &OpDispatchBuilder::AVX128_VectorRound<4>},
{OPD(3, 0b01, 0x09), 1, &OpDispatchBuilder::AVX128_VectorRound<8>},
@ -1808,4 +1808,40 @@ void OpDispatchBuilder::AVX128_VSHUF(OpcodeArgs) {
AVX128_StoreResult_WithOpSize(Op, Op->Dest, Result);
}
template<size_t ElementSize>
void OpDispatchBuilder::AVX128_VPERMILImm(OpcodeArgs) {
const auto SrcSize = GetSrcSize(Op);
const auto Is128Bit = SrcSize == Core::CPUState::XMM_SSE_REG_SIZE;
const auto Selector = Op->Src[1].Literal() & 0xFF;
auto Src = AVX128_LoadSource_WithOpSize(Op, Op->Src[0], Op->Flags, !Is128Bit);
RefPair Result = AVX128_Zext(LoadZeroVector(OpSize::i128Bit));
///< TODO: This could be optimized.
if constexpr (ElementSize == OpSize::i64Bit) {
Result.Low = _VInsElement(OpSize::i128Bit, ElementSize, 0, Selector & 0b0001, Result.Low, Src.Low);
Result.Low = _VInsElement(OpSize::i128Bit, ElementSize, 1, (Selector & 0b0010) >> 1, Result.Low, Src.Low);
if (!Is128Bit) {
Result.High = _VInsElement(OpSize::i128Bit, ElementSize, 0, ((Selector & 0b0100) >> 2), Result.High, Src.High);
Result.High = _VInsElement(OpSize::i128Bit, ElementSize, 1, ((Selector & 0b1000) >> 3), Result.High, Src.High);
}
} else {
Result.Low = _VInsElement(OpSize::i128Bit, ElementSize, 0, Selector & 0b00000011, Result.Low, Src.Low);
Result.Low = _VInsElement(OpSize::i128Bit, ElementSize, 1, (Selector & 0b00001100) >> 2, Result.Low, Src.Low);
Result.Low = _VInsElement(OpSize::i128Bit, ElementSize, 2, (Selector & 0b00110000) >> 4, Result.Low, Src.Low);
Result.Low = _VInsElement(OpSize::i128Bit, ElementSize, 3, (Selector & 0b11000000) >> 6, Result.Low, Src.Low);
if (!Is128Bit) {
Result.High = _VInsElement(OpSize::i128Bit, ElementSize, 0, (Selector & 0b00000011), Result.High, Src.High);
Result.High = _VInsElement(OpSize::i128Bit, ElementSize, 1, ((Selector & 0b00001100) >> 2), Result.High, Src.High);
Result.High = _VInsElement(OpSize::i128Bit, ElementSize, 2, ((Selector & 0b00110000) >> 4), Result.High, Src.High);
Result.High = _VInsElement(OpSize::i128Bit, ElementSize, 3, ((Selector & 0b11000000) >> 6), Result.High, Src.High);
}
}
AVX128_StoreResult_WithOpSize(Op, Op->Dest, Result);
}
} // namespace FEXCore::IR