Suyu/dynarmic
1
0
Fork 0
mirror of https://git.suyu.dev/suyu/dynarmic.git synced 2026-01-05 22:18:16 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions

backend/x64: Add further Unsafe_InaccurateNaN locations

Browse source
This commit is contained in:
MerryMage 2021-01-02 17:28:22 +00:00
parent f9ccf91b94
commit c15917b350
2 changed files with 27 additions and 7 deletions
Download patch file Download diff file Expand all files Collapse all files

14
src/backend/x64/emit_x64_vector_floating_point.cpp
View file

@ -290,7 +290,7 @@ void EmitTwoOpVectorOperation(BlockOfCode& code, EmitContext& ctx, IR::Inst* ins
auto args = ctx.reg_alloc.GetArgumentInfo(inst);
const bool fpcr_controlled = args[fpcr_controlled_arg_index].GetImmediateU1();
if (ctx.FPCR(fpcr_controlled).DN()) {
if (ctx.FPCR(fpcr_controlled).DN() || ctx.HasOptimization(OptimizationFlag::Unsafe_InaccurateNaN)) {
Xbyak::Xmm result;
if constexpr (std::is_member_function_pointer_v<Function>) {
@ -306,7 +306,9 @@ void EmitTwoOpVectorOperation(BlockOfCode& code, EmitContext& ctx, IR::Inst* ins
});
}
ForceToDefaultNaN<fsize>(code, ctx.FPCR(fpcr_controlled), result);
if (!ctx.HasOptimization(OptimizationFlag::Unsafe_InaccurateNaN)) {
ForceToDefaultNaN<fsize>(code, ctx.FPCR(fpcr_controlled), result);
}
ctx.reg_alloc.DefineValue(inst, result);
return;
@ -342,7 +344,7 @@ void EmitThreeOpVectorOperation(BlockOfCode& code, EmitContext& ctx, IR::Inst* i
auto args = ctx.reg_alloc.GetArgumentInfo(inst);
const bool fpcr_controlled = args[2].GetImmediateU1();
if (ctx.FPCR(fpcr_controlled).DN()) {
if (ctx.FPCR(fpcr_controlled).DN() || ctx.HasOptimization(OptimizationFlag::Unsafe_InaccurateNaN)) {
const Xbyak::Xmm xmm_a = ctx.reg_alloc.UseScratchXmm(args[0]);
const Xbyak::Xmm xmm_b = ctx.reg_alloc.UseXmm(args[1]);
@ -356,7 +358,9 @@ void EmitThreeOpVectorOperation(BlockOfCode& code, EmitContext& ctx, IR::Inst* i
});
}
ForceToDefaultNaN<fsize>(code, ctx.FPCR(fpcr_controlled), xmm_a);
if (!ctx.HasOptimization(OptimizationFlag::Unsafe_InaccurateNaN)) {
ForceToDefaultNaN<fsize>(code, ctx.FPCR(fpcr_controlled), xmm_a);
}
ctx.reg_alloc.DefineValue(inst, xmm_a);
return;
@ -988,7 +992,7 @@ void EmitFPVectorMulAdd(BlockOfCode& code, EmitContext& ctx, IR::Inst* inst) {
auto args = ctx.reg_alloc.GetArgumentInfo(inst);
const bool fpcr_controlled = args[3].GetImmediateU1();
if (code.HasFMA() && code.HasAVX()) {
if (code.HasFMA() && code.HasAVX() && ctx.HasOptimization(OptimizationFlag::Unsafe_UnfuseFMA)) {
const Xbyak::Xmm result = ctx.reg_alloc.UseScratchXmm(args[0]);
const Xbyak::Xmm xmm_b = ctx.reg_alloc.UseXmm(args[1]);
const Xbyak::Xmm xmm_c = ctx.reg_alloc.UseXmm(args[2]);