suyu/dynarmic
28
1
Fork 1
Code Issues Pull requests Projects Releases Packages Wiki Activity

A64: Implement SQRDMULH's scalar indexed element variant

Browse source
This commit is contained in:
Lioncash 2019-03-20 23:26:55 -04:00 committed by MerryMage
parent 692aba91b6
commit 97dd3d0596
2 changed files with 20 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
src/frontend/A64/decoder/a64.inc
View file

@ -537,7 +537,7 @@ INST(FCVTZU_fix_1, "FCVTZU (vector, fixed-point)", "01111
//INST(SQDMLSL_elt_1, "SQDMLSL, SQDMLSL2 (by element)", "01011111zzLMmmmm0111H0nnnnnddddd") //INST(SQDMLSL_elt_1, "SQDMLSL, SQDMLSL2 (by element)", "01011111zzLMmmmm0111H0nnnnnddddd")
INST(SQDMULL_elt_1, "SQDMULL, SQDMULL2 (by element)", "01011111zzLMmmmm1011H0nnnnnddddd") INST(SQDMULL_elt_1, "SQDMULL, SQDMULL2 (by element)", "01011111zzLMmmmm1011H0nnnnnddddd")
INST(SQDMULH_elt_1, "SQDMULH (by element)", "01011111zzLMmmmm1100H0nnnnnddddd") INST(SQDMULH_elt_1, "SQDMULH (by element)", "01011111zzLMmmmm1100H0nnnnnddddd")
//INST(SQRDMULH_elt_1, "SQRDMULH (by element)", "01011111zzLMmmmm1101H0nnnnnddddd") INST(SQRDMULH_elt_1, "SQRDMULH (by element)", "01011111zzLMmmmm1101H0nnnnnddddd")
//INST(FMLA_elt_1, "FMLA (by element)", "0101111100LMmmmm0001H0nnnnnddddd") //INST(FMLA_elt_1, "FMLA (by element)", "0101111100LMmmmm0001H0nnnnnddddd")
INST(FMLA_elt_2, "FMLA (by element)", "010111111zLMmmmm0001H0nnnnnddddd") INST(FMLA_elt_2, "FMLA (by element)", "010111111zLMmmmm0001H0nnnnnddddd")
//INST(FMLS_elt_1, "FMLS (by element)", "0101111100LMmmmm0101H0nnnnnddddd") //INST(FMLS_elt_1, "FMLS (by element)", "0101111100LMmmmm0101H0nnnnnddddd")

20
src/frontend/A64/translate/impl/simd_scalar_x_indexed_element.cpp
View file

@ -94,6 +94,24 @@ bool TranslatorVisitor::SQDMULH_elt_1(Imm<2> size, Imm<1> L, Imm<1> M, Imm<4> Vm
return true; return true;
} }
bool TranslatorVisitor::SQRDMULH_elt_1(Imm<2> size, Imm<1> L, Imm<1> M, Imm<4> Vmlo, Imm<1> H, Vec Vn, Vec Vd) {
if (size == 0b00 || size == 0b11) {
return UnallocatedEncoding();
}
const size_t esize = 8 << size.ZeroExtend();
const auto [index, Vm] = Combine(size, H, L, M, Vmlo);
const IR::U128 operand1 = ir.ZeroExtendToQuad(ir.VectorGetElement(esize, V(128, Vn), 0));
const IR::UAny operand2 = ir.VectorGetElement(esize, V(128, Vm), index);
const IR::U128 broadcast = ir.VectorBroadcast(esize, operand2);
const IR::UpperAndLower multiply = ir.VectorSignedSaturatedDoublingMultiply(esize, operand1, broadcast);
const IR::U128 result = ir.VectorAdd(esize, multiply.upper, ir.VectorLogicalShiftRight(esize, multiply.lower, static_cast<u8>(esize - 1)));
V(128, Vd, result);
return true;
}
bool TranslatorVisitor::SQDMULL_elt_1(Imm<2> size, Imm<1> L, Imm<1> M, Imm<4> Vmlo, Imm<1> H, Vec Vn, Vec Vd) { bool TranslatorVisitor::SQDMULL_elt_1(Imm<2> size, Imm<1> L, Imm<1> M, Imm<4> Vmlo, Imm<1> H, Vec Vn, Vec Vd) {
if (size == 0b00 || size == 0b11) { if (size == 0b00 || size == 0b11) {
return UnallocatedEncoding(); return UnallocatedEncoding();
@ -107,7 +125,7 @@ bool TranslatorVisitor::SQDMULL_elt_1(Imm<2> size, Imm<1> L, Imm<1> M, Imm<4> Vm
const IR::U128 broadcast = ir.VectorBroadcast(esize, operand2); const IR::U128 broadcast = ir.VectorBroadcast(esize, operand2);
const IR::U128 result = ir.VectorSignedSaturatedDoublingMultiplyLong(esize, operand1, broadcast); const IR::U128 result = ir.VectorSignedSaturatedDoublingMultiplyLong(esize, operand1, broadcast);
V_scalar(esize * 2, Vd, result); V(128, Vd, result);
return true; return true;
} }