ir: Add opcodes for floating-point vector equalities

This commit is contained in:
Lioncash 2018-06-02 16:11:40 -04:00 committed by MerryMage
parent e64978ed89
commit 5ce187a54e
4 changed files with 34 additions and 0 deletions

View file

@ -189,6 +189,26 @@ void EmitX64::EmitFPVectorDiv64(EmitContext& ctx, IR::Inst* inst) {
EmitVectorOperation64(code, ctx, inst, &Xbyak::CodeGenerator::divpd);
}
void EmitX64::EmitFPVectorEqual32(EmitContext& ctx, IR::Inst* inst) {
auto args = ctx.reg_alloc.GetArgumentInfo(inst);
const Xbyak::Xmm a = ctx.reg_alloc.UseScratchXmm(args[0]);
const Xbyak::Xmm b = ctx.reg_alloc.UseXmm(args[1]);
code.cmpeqps(a, b);
ctx.reg_alloc.DefineValue(inst, a);
}
void EmitX64::EmitFPVectorEqual64(EmitContext& ctx, IR::Inst* inst) {
auto args = ctx.reg_alloc.GetArgumentInfo(inst);
const Xbyak::Xmm a = ctx.reg_alloc.UseScratchXmm(args[0]);
const Xbyak::Xmm b = ctx.reg_alloc.UseXmm(args[1]);
code.cmpeqpd(a, b);
ctx.reg_alloc.DefineValue(inst, a);
}
void EmitX64::EmitFPVectorMul32(EmitContext& ctx, IR::Inst* inst) {
EmitVectorOperation32(code, ctx, inst, &Xbyak::CodeGenerator::mulps);
}

View file

@ -1503,6 +1503,17 @@ U128 IREmitter::FPVectorDiv(size_t esize, const U128& a, const U128& b) {
return {};
}
U128 IREmitter::FPVectorEqual(size_t esize, const U128& a, const U128& b) {
switch (esize) {
case 32:
return Inst<U128>(Opcode::FPVectorEqual32, a, b);
case 64:
return Inst<U128>(Opcode::FPVectorEqual64, a, b);
}
UNREACHABLE();
return {};
}
U128 IREmitter::FPVectorMul(size_t esize, const U128& a, const U128& b) {
switch (esize) {
case 32:

View file

@ -274,6 +274,7 @@ public:
U128 FPVectorAdd(size_t esize, const U128& a, const U128& b);
U128 FPVectorDiv(size_t esize, const U128& a, const U128& b);
U128 FPVectorEqual(size_t esize, const U128& a, const U128& b);
U128 FPVectorMul(size_t esize, const U128& a, const U128& b);
U128 FPVectorSub(size_t esize, const U128& a, const U128& b);
U128 FPVectorS32ToSingle(const U128& a);

View file

@ -398,6 +398,8 @@ OPCODE(FPVectorAdd32, T::U128, T::U128, T::U
OPCODE(FPVectorAdd64, T::U128, T::U128, T::U128 )
OPCODE(FPVectorDiv32, T::U128, T::U128, T::U128 )
OPCODE(FPVectorDiv64, T::U128, T::U128, T::U128 )
OPCODE(FPVectorEqual32, T::U128, T::U128, T::U128 )
OPCODE(FPVectorEqual64, T::U128, T::U128, T::U128 )
OPCODE(FPVectorMul32, T::U128, T::U128, T::U128 )
OPCODE(FPVectorMul64, T::U128, T::U128, T::U128 )
OPCODE(FPVectorS32ToSingle, T::U128, T::U128 )