ShuriZma
/
xenia
mirror of https://github.com/xenia-project/xenia.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

[x64] Add AX512 optimization for `OPCODE_SELECT`(F64)

This commit is contained in:
Wunkolo 2022-10-23 22:21:25 -07:00 committed by Rick Gibbed
parent ba75a016b4
commit e55cb737c1
1 changed files with 20 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
src/xenia/cpu/backend/x64/x64_sequences.cc
View File

@ -698,6 +698,26 @@ struct SELECT_F64
: Sequence<SELECT_F64, I<OPCODE_SELECT, F64Op, I8Op, F64Op, F64Op>> {
static void Emit(X64Emitter& e, const EmitArgType& i) {
// dest = src1 != 0 ? src2 : src3
if (e.IsFeatureEnabled(kX64EmitAVX512Ortho)) {
e.movzx(e.rax, i.src1);
e.vmovq(e.xmm0, e.rax);
e.vptestmq(e.k1, e.xmm0, e.xmm0);
const Xmm src2 = i.src2.is_constant ? e.xmm1 : i.src2;
if (i.src2.is_constant) {
e.LoadConstantXmm(src2, i.src2.constant());
}
const Xmm src3 = i.src3.is_constant ? e.xmm2 : i.src3;
if (i.src3.is_constant) {
e.LoadConstantXmm(src3, i.src3.constant());
}
e.vpblendmq(i.dest.reg() | e.k1, src3, src2);
return;
}
e.movzx(e.eax, i.src1);
e.vmovd(e.xmm1, e.eax);
e.vpxor(e.xmm0, e.xmm0);