Merge pull request #2235 from lioncash/min

OpcodeDispatcher: Handle VPMINS{B, D, W}/VPMINU{B, D, W}
This commit is contained in:
Ryan Houdek authored and GitHub committed 2022-12-13 13:31:14 -08:00
commit 40bab6b58e
9 files changed
+271 -7

No files matched your search

@@ -5915,6 +5915,7 @@ void OpDispatchBuilder::InstallHostSpecificOpcodeHandlers() {
{OPD(1, 0b01, 0xD8), 1, &OpDispatchBuilder::AVXVectorALUOp<IR::OP_VUQSUB, 1>},
{OPD(1, 0b01, 0xD9), 1, &OpDispatchBuilder::AVXVectorALUOp<IR::OP_VUQSUB, 2>},
{OPD(1, 0b01, 0xDA), 1, &OpDispatchBuilder::AVXVectorALUOp<IR::OP_VUMIN, 1>},
{OPD(1, 0b01, 0xDB), 1, &OpDispatchBuilder::AVXVectorALUOp<IR::OP_VAND, 16>},
{OPD(1, 0b01, 0xDC), 1, &OpDispatchBuilder::AVXVectorALUOp<IR::OP_VUQADD, 1>},
{OPD(1, 0b01, 0xDD), 1, &OpDispatchBuilder::AVXVectorALUOp<IR::OP_VUQADD, 2>},
@@ -5927,6 +5928,7 @@ void OpDispatchBuilder::InstallHostSpecificOpcodeHandlers() {
{OPD(1, 0b01, 0xE8), 1, &OpDispatchBuilder::AVXVectorALUOp<IR::OP_VSQSUB, 1>},
{OPD(1, 0b01, 0xE9), 1, &OpDispatchBuilder::AVXVectorALUOp<IR::OP_VSQSUB, 2>},
{OPD(1, 0b01, 0xEA), 1, &OpDispatchBuilder::AVXVectorALUOp<IR::OP_VSMIN, 2>},
{OPD(1, 0b01, 0xEB), 1, &OpDispatchBuilder::AVXVectorALUOp<IR::OP_VOR, 16>},
{OPD(1, 0b01, 0xEC), 1, &OpDispatchBuilder::AVXVectorALUOp<IR::OP_VSQADD, 1>},
{OPD(1, 0b01, 0xED), 1, &OpDispatchBuilder::AVXVectorALUOp<IR::OP_VSQADD, 2>},
@@ -5956,7 +5958,10 @@ void OpDispatchBuilder::InstallHostSpecificOpcodeHandlers() {
{OPD(2, 0b01, 0x2A), 1, &OpDispatchBuilder::VMOVVectorNTOp},
{OPD(2, 0b01, 0x37), 1, &OpDispatchBuilder::AVXVectorALUOp<IR::OP_VCMPGT, 8>},
{OPD(2, 0b01, 0x3B), 1, &OpDispatchBuilder::UnimplementedOp},
{OPD(2, 0b01, 0x38), 1, &OpDispatchBuilder::AVXVectorALUOp<IR::OP_VSMIN, 1>},
{OPD(2, 0b01, 0x39), 1, &OpDispatchBuilder::AVXVectorALUOp<IR::OP_VSMIN, 4>},
{OPD(2, 0b01, 0x3A), 1, &OpDispatchBuilder::AVXVectorALUOp<IR::OP_VUMIN, 2>},
{OPD(2, 0b01, 0x3B), 1, &OpDispatchBuilder::AVXVectorALUOp<IR::OP_VUMIN, 4>},
{OPD(2, 0b01, 0x58), 1, &OpDispatchBuilder::VBROADCASTOp<4>},
{OPD(2, 0b01, 0x59), 1, &OpDispatchBuilder::VBROADCASTOp<8>},
@@ -499,6 +499,19 @@ void OpDispatchBuilder::AVXVectorALUOp<IR::OP_VURAVG, 1>(OpcodeArgs);
template
void OpDispatchBuilder::AVXVectorALUOp<IR::OP_VURAVG, 2>(OpcodeArgs);
template
void OpDispatchBuilder::AVXVectorALUOp<IR::OP_VSMIN, 1>(OpcodeArgs);
template
void OpDispatchBuilder::AVXVectorALUOp<IR::OP_VSMIN, 2>(OpcodeArgs);
template
void OpDispatchBuilder::AVXVectorALUOp<IR::OP_VSMIN, 4>(OpcodeArgs);
template
void OpDispatchBuilder::AVXVectorALUOp<IR::OP_VUMIN, 1>(OpcodeArgs);
template
void OpDispatchBuilder::AVXVectorALUOp<IR::OP_VUMIN, 2>(OpcodeArgs);
template
void OpDispatchBuilder::AVXVectorALUOp<IR::OP_VUMIN, 4>(OpcodeArgs);
template<FEXCore::IR::IROps IROp, size_t ElementSize>
void OpDispatchBuilder::VectorALUROp(OpcodeArgs) {
auto Size = GetSrcSize(Op);
@@ -212,7 +212,7 @@ void InitializeVEXTables() {
{OPD(1, 0b01, 0xD8), 1, X86InstInfo{"VPSUBUSB", TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_VEX_1ST_SRC | FLAGS_XMM_FLAGS, 0, nullptr}},
{OPD(1, 0b01, 0xD9), 1, X86InstInfo{"VPSUBUSW", TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_VEX_1ST_SRC | FLAGS_XMM_FLAGS, 0, nullptr}},
{OPD(1, 0b01, 0xDA), 1, X86InstInfo{"VPMINUB", TYPE_INST, FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
{OPD(1, 0b01, 0xDA), 1, X86InstInfo{"VPMINUB", TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_VEX_1ST_SRC | FLAGS_XMM_FLAGS, 0, nullptr}},
{OPD(1, 0b01, 0xDB), 1, X86InstInfo{"VPAND", TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_VEX_1ST_SRC | FLAGS_XMM_FLAGS, 0, nullptr}},
{OPD(1, 0b01, 0xDC), 1, X86InstInfo{"VPADDUSB", TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_VEX_1ST_SRC | FLAGS_XMM_FLAGS, 0, nullptr}},
{OPD(1, 0b01, 0xDD), 1, X86InstInfo{"VPADDUSW", TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_VEX_1ST_SRC | FLAGS_XMM_FLAGS, 0, nullptr}},
@@ -234,7 +234,7 @@ void InitializeVEXTables() {
{OPD(1, 0b01, 0xE8), 1, X86InstInfo{"VPSUBSB", TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_VEX_1ST_SRC | FLAGS_XMM_FLAGS, 0, nullptr}},
{OPD(1, 0b01, 0xE9), 1, X86InstInfo{"VPSUBSW", TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_VEX_1ST_SRC | FLAGS_XMM_FLAGS, 0, nullptr}},
{OPD(1, 0b01, 0xEA), 1, X86InstInfo{"VPMINSW", TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
{OPD(1, 0b01, 0xEA), 1, X86InstInfo{"VPMINSW", TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_VEX_1ST_SRC | FLAGS_XMM_FLAGS, 0, nullptr}},
{OPD(1, 0b01, 0xEB), 1, X86InstInfo{"VPOR", TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_VEX_1ST_SRC | FLAGS_XMM_FLAGS, 0, nullptr}},
{OPD(1, 0b01, 0xEC), 1, X86InstInfo{"VPADDSB", TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_VEX_1ST_SRC | FLAGS_XMM_FLAGS, 0, nullptr}},
{OPD(1, 0b01, 0xED), 1, X86InstInfo{"VPADDSW", TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_VEX_1ST_SRC | FLAGS_XMM_FLAGS, 0, nullptr}},
@@ -314,10 +314,10 @@ void InitializeVEXTables() {
{OPD(2, 0b01, 0x36), 1, X86InstInfo{"VPERMD", TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
{OPD(2, 0b01, 0x37), 1, X86InstInfo{"VPCMPGTQ", TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_VEX_1ST_SRC | FLAGS_XMM_FLAGS, 0, nullptr}},
{OPD(2, 0b01, 0x38), 1, X86InstInfo{"VPMINSB", TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
{OPD(2, 0b01, 0x39), 1, X86InstInfo{"VPMINSD", TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
{OPD(2, 0b01, 0x3A), 1, X86InstInfo{"VPMINUW", TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
{OPD(2, 0b01, 0x3B), 1, X86InstInfo{"VPMINUD", TYPE_INST, FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
{OPD(2, 0b01, 0x38), 1, X86InstInfo{"VPMINSB", TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_VEX_1ST_SRC | FLAGS_XMM_FLAGS, 0, nullptr}},
{OPD(2, 0b01, 0x39), 1, X86InstInfo{"VPMINSD", TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_VEX_1ST_SRC | FLAGS_XMM_FLAGS, 0, nullptr}},
{OPD(2, 0b01, 0x3A), 1, X86InstInfo{"VPMINUW", TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_VEX_1ST_SRC | FLAGS_XMM_FLAGS, 0, nullptr}},
{OPD(2, 0b01, 0x3B), 1, X86InstInfo{"VPMINUD", TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_VEX_1ST_SRC | FLAGS_XMM_FLAGS, 0, nullptr}},
{OPD(2, 0b01, 0x3C), 1, X86InstInfo{"VPMAXSB", TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
{OPD(2, 0b01, 0x3D), 1, X86InstInfo{"VPMAXSD", TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
{OPD(2, 0b01, 0x3E), 1, X86InstInfo{"VPMAXUW", TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
+41
View File
@@ -0,0 +1,41 @@
%ifdef CONFIG
{
"HostFeatures": ["AVX"],
"RegData": {
"XMM0": ["0x4142434445464748", "0x7172737475767778", "0x4142434445464748", "0x7172737475767778"],
"XMM1": ["0x8062636465666768", "0x5152535455565758", "0x6162636465666768", "0x5152535455565758"],
"XMM2": ["0x8042434445464748", "0x5152535455565758", "0x0000000000000000", "0x0000000000000000"],
"XMM3": ["0x8042434445464748", "0x5152535455565758", "0x4142434445464748", "0x5152535455565758"],
"XMM4": ["0x8042434445464748", "0x5152535455565758", "0x0000000000000000", "0x0000000000000000"],
"XMM5": ["0x8042434445464748", "0x5152535455565758", "0x4142434445464748", "0x5152535455565758"]
},
"MemoryRegions": {
"0x100000000": "4096"
}
}
%endif
lea rdx, [rel .data]
vmovapd ymm0, [rdx]
vmovapd ymm1, [rdx + 32]
vpminsb xmm2, xmm0, xmm1
vpminsb ymm3, ymm0, ymm1
vpminsb xmm4, xmm0, [rdx + 32]
vpminsb ymm5, ymm0, [rdx + 32]
hlt
align 32
.data:
dq 0x4142434445464748
dq 0x7172737475767778
dq 0x4142434445464748
dq 0x7172737475767778
dq 0x8062636465666768
dq 0x5152535455565758
dq 0x6162636465666768
dq 0x5152535455565758
+41
View File
@@ -0,0 +1,41 @@
%ifdef CONFIG
{
"HostFeatures": ["AVX"],
"RegData": {
"XMM0": ["0x4142434445464748", "0x7172737475767778", "0x4142434445464748", "0x7172737475767778"],
"XMM1": ["0x8000000065666768", "0x5152535455565758", "0x6162636465666768", "0x5152535455565758"],
"XMM2": ["0x8000000045464748", "0x5152535455565758", "0x0000000000000000", "0x0000000000000000"],
"XMM3": ["0x8000000045464748", "0x5152535455565758", "0x4142434445464748", "0x5152535455565758"],
"XMM4": ["0x8000000045464748", "0x5152535455565758", "0x0000000000000000", "0x0000000000000000"],
"XMM5": ["0x8000000045464748", "0x5152535455565758", "0x4142434445464748", "0x5152535455565758"]
},
"MemoryRegions": {
"0x100000000": "4096"
}
}
%endif
lea rdx, [rel .data]
vmovapd ymm0, [rdx]
vmovapd ymm1, [rdx + 32]
vpminsd xmm2, xmm0, xmm1
vpminsd ymm3, ymm0, ymm1
vpminsd xmm4, xmm0, [rdx + 32]
vpminsd ymm5, ymm0, [rdx + 32]
hlt
align 32
.data:
dq 0x4142434445464748
dq 0x7172737475767778
dq 0x4142434445464748
dq 0x7172737475767778
dq 0x8000000065666768
dq 0x5152535455565758
dq 0x6162636465666768
dq 0x5152535455565758
+41
View File
@@ -0,0 +1,41 @@
%ifdef CONFIG
{
"HostFeatures": ["AVX"],
"RegData": {
"XMM0": ["0x4142434445464748", "0x7172737475767778", "0x4142434445464748", "0x7172737475767778"],
"XMM1": ["0x8000636465666768", "0x5152535455565758", "0x6162636465666768", "0x5152535455565758"],
"XMM2": ["0x8000434445464748", "0x5152535455565758", "0x0000000000000000", "0x0000000000000000"],
"XMM3": ["0x8000434445464748", "0x5152535455565758", "0x4142434445464748", "0x5152535455565758"],
"XMM4": ["0x8000434445464748", "0x5152535455565758", "0x0000000000000000", "0x0000000000000000"],
"XMM5": ["0x8000434445464748", "0x5152535455565758", "0x4142434445464748", "0x5152535455565758"]
},
"MemoryRegions": {
"0x100000000": "4096"
}
}
%endif
lea rdx, [rel .data]
vmovapd ymm0, [rdx]
vmovapd ymm1, [rdx + 32]
vpminsw xmm2, xmm0, xmm1
vpminsw ymm3, ymm0, ymm1
vpminsw xmm4, xmm0, [rdx + 32]
vpminsw ymm5, ymm0, [rdx + 32]
hlt
align 32
.data:
dq 0x4142434445464748
dq 0x7172737475767778
dq 0x4142434445464748
dq 0x7172737475767778
dq 0x8000636465666768
dq 0x5152535455565758
dq 0x6162636465666768
dq 0x5152535455565758
+41
View File
@@ -0,0 +1,41 @@
%ifdef CONFIG
{
"HostFeatures": ["AVX"],
"RegData": {
"XMM0": ["0x4142434445464748", "0x7172737475767778", "0x4142434445464748", "0x7172737475767778"],
"XMM1": ["0x6162636465666768", "0x5152535455565758", "0x6162636465666768", "0x5152535455565758"],
"XMM2": ["0x4142434445464748", "0x5152535455565758", "0x0000000000000000", "0x0000000000000000"],
"XMM3": ["0x4142434445464748", "0x5152535455565758", "0x4142434445464748", "0x5152535455565758"],
"XMM4": ["0x4142434445464748", "0x5152535455565758", "0x0000000000000000", "0x0000000000000000"],
"XMM5": ["0x4142434445464748", "0x5152535455565758", "0x4142434445464748", "0x5152535455565758"]
},
"MemoryRegions": {
"0x100000000": "4096"
}
}
%endif
lea rdx, [rel .data]
vmovapd ymm0, [rdx]
vmovapd ymm1, [rdx + 32]
vpminub xmm2, xmm0, xmm1
vpminub ymm3, ymm0, ymm1
vpminub xmm4, xmm0, [rdx + 32]
vpminub ymm5, ymm0, [rdx + 32]
hlt
align 32
.data:
dq 0x4142434445464748
dq 0x7172737475767778
dq 0x4142434445464748
dq 0x7172737475767778
dq 0x6162636465666768
dq 0x5152535455565758
dq 0x6162636465666768
dq 0x5152535455565758
+41
View File
@@ -0,0 +1,41 @@
%ifdef CONFIG
{
"HostFeatures": ["AVX"],
"RegData": {
"XMM0": ["0x4142434445464748", "0x7172737475767778", "0x4142434445464748", "0x7172737475767778"],
"XMM1": ["0x6162636465666768", "0x5152535455565758", "0x6162636465666768", "0x5152535455565758"],
"XMM2": ["0x4142434445464748", "0x5152535455565758", "0x0000000000000000", "0x0000000000000000"],
"XMM3": ["0x4142434445464748", "0x5152535455565758", "0x4142434445464748", "0x5152535455565758"],
"XMM4": ["0x4142434445464748", "0x5152535455565758", "0x0000000000000000", "0x0000000000000000"],
"XMM5": ["0x4142434445464748", "0x5152535455565758", "0x4142434445464748", "0x5152535455565758"]
},
"MemoryRegions": {
"0x100000000": "4096"
}
}
%endif
lea rdx, [rel .data]
vmovapd ymm0, [rdx]
vmovapd ymm1, [rdx + 32]
vpminud xmm2, xmm0, xmm1
vpminud ymm3, ymm0, ymm1
vpminud xmm4, xmm0, [rdx + 32]
vpminud ymm5, ymm0, [rdx + 32]
hlt
align 32
.data:
dq 0x4142434445464748
dq 0x7172737475767778
dq 0x4142434445464748
dq 0x7172737475767778
dq 0x6162636465666768
dq 0x5152535455565758
dq 0x6162636465666768
dq 0x5152535455565758
+41
View File
@@ -0,0 +1,41 @@
%ifdef CONFIG
{
"HostFeatures": ["AVX"],
"RegData": {
"XMM0": ["0x4142434445464748", "0x7172737475767778", "0x4142434445464748", "0x7172737475767778"],
"XMM1": ["0x6162636465666768", "0x5152535455565758", "0x6162636465666768", "0x5152535455565758"],
"XMM2": ["0x4142434445464748", "0x5152535455565758", "0x0000000000000000", "0x0000000000000000"],
"XMM3": ["0x4142434445464748", "0x5152535455565758", "0x4142434445464748", "0x5152535455565758"],
"XMM4": ["0x4142434445464748", "0x5152535455565758", "0x0000000000000000", "0x0000000000000000"],
"XMM5": ["0x4142434445464748", "0x5152535455565758", "0x4142434445464748", "0x5152535455565758"]
},
"MemoryRegions": {
"0x100000000": "4096"
}
}
%endif
lea rdx, [rel .data]
vmovapd ymm0, [rdx]
vmovapd ymm1, [rdx + 32]
vpminuw xmm2, xmm0, xmm1
vpminuw ymm3, ymm0, ymm1
vpminuw xmm4, xmm0, [rdx + 32]
vpminuw ymm5, ymm0, [rdx + 32]
hlt
align 32
.data:
dq 0x4142434445464748
dq 0x7172737475767778
dq 0x4142434445464748
dq 0x7172737475767778
dq 0x6162636465666768
dq 0x5152535455565758
dq 0x6162636465666768
dq 0x5152535455565758