emit_x64_floating_point: Simplify EmitFP{Min,Max}{,Numeric}{32,64}

This commit is contained in:
MerryMage 2018-08-02 20:16:03 +01:00
parent 07e0585994
commit 700088408d

View file

@ -108,75 +108,6 @@ void ZeroIfNaN(BlockOfCode& code, Xbyak::Xmm xmm_value, Xbyak::Xmm xmm_scratch)
code.pand(xmm_value, xmm_scratch); code.pand(xmm_value, xmm_scratch);
} }
template<size_t fsize>
void PreProcessNaNs(BlockOfCode& code, Xbyak::Xmm a, Xbyak::Xmm b, Xbyak::Label& end) {
using FPT = mp::unsigned_integer_of_size<fsize>;
Xbyak::Label nan;
FCODE(ucomis)(a, b);
code.jp(nan, code.T_NEAR);
code.SwitchToFarCode();
code.L(nan);
code.sub(rsp, 8);
ABI_PushCallerSaveRegistersAndAdjustStackExcept(code, HostLocXmmIdx(a.getIdx()));
code.movq(code.ABI_PARAM1, a);
code.movq(code.ABI_PARAM2, b);
code.CallFunction(static_cast<FPT(*)(FPT, FPT)>([](FPT a, FPT b) -> FPT {
return *FP::ProcessNaNs(a, b);
}));
code.movq(a, code.ABI_RETURN);
ABI_PopCallerSaveRegistersAndAdjustStackExcept(code, HostLocXmmIdx(a.getIdx()));
code.add(rsp, 8);
code.jmp(end, code.T_NEAR);
code.SwitchToNearCode();
}
template<size_t fsize, typename NaNHandler>
void PreProcessNaNs(BlockOfCode& code, EmitContext& ctx, Xbyak::Xmm a, Xbyak::Xmm b, Xbyak::Xmm c, Xbyak::Label& end, NaNHandler nan_handler) {
using FPT = mp::unsigned_integer_of_size<fsize>;
Xbyak::Label nan;
FCODE(ucomis)(a, b);
code.jp(nan, code.T_NEAR);
FCODE(ucomis)(c, c);
code.jp(nan, code.T_NEAR);
code.SwitchToFarCode();
code.L(nan);
code.sub(rsp, 8);
ABI_PushCallerSaveRegistersAndAdjustStackExcept(code, HostLocXmmIdx(a.getIdx()));
code.movq(code.ABI_PARAM1, a);
code.movq(code.ABI_PARAM2, b);
code.movq(code.ABI_PARAM3, c);
code.mov(code.ABI_PARAM4, ctx.FPCR());
code.CallFunction(static_cast<FPT(*)(FPT, FPT, FPT, FP::FPCR)>(nan_handler));
code.movq(a, code.ABI_RETURN);
ABI_PopCallerSaveRegistersAndAdjustStackExcept(code, HostLocXmmIdx(a.getIdx()));
code.add(rsp, 8);
code.jmp(end, code.T_NEAR);
code.SwitchToNearCode();
}
template<size_t fsize>
void PostProcessNaNs(BlockOfCode& code, Xbyak::Xmm result, Xbyak::Xmm tmp) {
if constexpr (fsize == 32) {
code.movaps(tmp, result);
code.cmpunordps(tmp, tmp);
code.pslld(tmp, 31);
code.xorps(result, tmp);
} else {
code.movaps(tmp, result);
code.cmpunordpd(tmp, tmp);
code.psllq(tmp, 63);
code.xorps(result, tmp);
}
}
template<size_t fsize> template<size_t fsize>
void ForceToDefaultNaN(BlockOfCode& code, Xbyak::Xmm result) { void ForceToDefaultNaN(BlockOfCode& code, Xbyak::Xmm result) {
if (code.DoesCpuSupport(Xbyak::util::Cpu::tAVX)) { if (code.DoesCpuSupport(Xbyak::util::Cpu::tAVX)) {
@ -207,6 +138,21 @@ Xbyak::Label ProcessNaN(BlockOfCode& code, Xbyak::Xmm a) {
return end; return end;
} }
template<size_t fsize>
void PostProcessNaN(BlockOfCode& code, Xbyak::Xmm result, Xbyak::Xmm tmp) {
if constexpr (fsize == 32) {
code.movaps(tmp, result);
code.cmpunordps(tmp, tmp);
code.pslld(tmp, 31);
code.xorps(result, tmp);
} else {
code.movaps(tmp, result);
code.cmpunordpd(tmp, tmp);
code.psllq(tmp, 63);
code.xorps(result, tmp);
}
}
// This is necessary because x86 and ARM differ in they way they return NaNs from floating point operations // This is necessary because x86 and ARM differ in they way they return NaNs from floating point operations
// //
// ARM behaviour: // ARM behaviour:
@ -372,47 +318,7 @@ void FPTwoOp(BlockOfCode& code, EmitContext& ctx, IR::Inst* inst, Function fn) {
if (ctx.FPSCR_DN()) { if (ctx.FPSCR_DN()) {
ForceToDefaultNaN<fsize>(code, result); ForceToDefaultNaN<fsize>(code, result);
} else if (ctx.AccurateNaN()) { } else if (ctx.AccurateNaN()) {
PostProcessNaNs<fsize>(code, result, ctx.reg_alloc.ScratchXmm()); PostProcessNaN<fsize>(code, result, ctx.reg_alloc.ScratchXmm());
}
code.L(end);
ctx.reg_alloc.DefineValue(inst, result);
}
enum class CallDenormalsAreZero {
Yes,
No,
};
template <size_t fsize, typename PreprocessFunction, typename Function>
void FPThreeOp(BlockOfCode& code, EmitContext& ctx, IR::Inst* inst, [[maybe_unused]] PreprocessFunction preprocess, Function fn, CallDenormalsAreZero call_denormals_are_zero = CallDenormalsAreZero::No) {
auto args = ctx.reg_alloc.GetArgumentInfo(inst);
Xbyak::Label end;
Xbyak::Xmm result = ctx.reg_alloc.UseScratchXmm(args[0]);
Xbyak::Xmm operand = ctx.reg_alloc.UseScratchXmm(args[1]);
Xbyak::Reg64 gpr_scratch = ctx.reg_alloc.ScratchGpr();
if (ctx.FPSCR_FTZ() && call_denormals_are_zero == CallDenormalsAreZero::Yes) {
DenormalsAreZero<fsize>(code, result, gpr_scratch);
DenormalsAreZero<fsize>(code, operand, gpr_scratch);
}
if constexpr(!std::is_same_v<PreprocessFunction, std::nullptr_t>) {
preprocess(result, operand, gpr_scratch, end);
}
if (ctx.AccurateNaN() && !ctx.FPSCR_DN()) {
PreProcessNaNs<fsize>(code, result, operand, end);
}
if constexpr (std::is_member_function_pointer_v<Function>) {
(code.*fn)(result, operand);
} else {
fn(result, operand);
}
if (ctx.FPSCR_DN()) {
ForceToDefaultNaN<fsize>(code, result);
} else if (ctx.AccurateNaN()) {
PostProcessNaNs<fsize>(code, result, operand);
} }
code.L(end); code.L(end);
@ -528,8 +434,8 @@ void EmitX64::EmitFPDiv64(EmitContext& ctx, IR::Inst* inst) {
FPThreeOp<64>(code, ctx, inst, &Xbyak::CodeGenerator::divsd); FPThreeOp<64>(code, ctx, inst, &Xbyak::CodeGenerator::divsd);
} }
template<size_t fsize> template<size_t fsize, bool is_max>
static void EmitFPMax(BlockOfCode& code, EmitContext& ctx, IR::Inst* inst) { static void EmitFPMinMax(BlockOfCode& code, EmitContext& ctx, IR::Inst* inst) {
auto args = ctx.reg_alloc.GetArgumentInfo(inst); auto args = ctx.reg_alloc.GetArgumentInfo(inst);
const Xbyak::Xmm result = ctx.reg_alloc.UseScratchXmm(args[0]); const Xbyak::Xmm result = ctx.reg_alloc.UseScratchXmm(args[0]);
@ -545,14 +451,22 @@ static void EmitFPMax(BlockOfCode& code, EmitContext& ctx, IR::Inst* inst) {
FCODE(ucomis)(result, operand); FCODE(ucomis)(result, operand);
code.jz(equal, code.T_NEAR); code.jz(equal, code.T_NEAR);
if constexpr (is_max) {
FCODE(maxs)(result, operand); FCODE(maxs)(result, operand);
} else {
FCODE(mins)(result, operand);
}
code.L(end); code.L(end);
code.SwitchToFarCode(); code.SwitchToFarCode();
code.L(equal); code.L(equal);
code.jp(nan); code.jp(nan);
if constexpr (is_max) {
code.andps(result, operand); code.andps(result, operand);
} else {
code.orps(result, operand);
}
code.jmp(end); code.jmp(end);
code.L(nan); code.L(nan);
@ -568,196 +482,137 @@ static void EmitFPMax(BlockOfCode& code, EmitContext& ctx, IR::Inst* inst) {
ctx.reg_alloc.DefineValue(inst, result); ctx.reg_alloc.DefineValue(inst, result);
} }
void EmitX64::EmitFPMax32(EmitContext& ctx, IR::Inst* inst) { template<size_t fsize, bool is_max>
EmitFPMax<32>(code, ctx, inst); static void EmitFPMinMaxNumeric(BlockOfCode& code, EmitContext& ctx, IR::Inst* inst) {
} using FPT = mp::unsigned_integer_of_size<fsize>;
constexpr u8 mantissa_msb_bit = static_cast<u8>(FP::FPInfo<FPT>::explicit_mantissa_width - 1);
void EmitX64::EmitFPMax64(EmitContext& ctx, IR::Inst* inst) {
EmitFPMax<64>(code, ctx, inst);
}
void EmitX64::EmitFPMaxNumeric32(EmitContext& ctx, IR::Inst* inst) {
FPThreeOp<32>(code, ctx, inst, [&](Xbyak::Xmm result, Xbyak::Xmm operand, Xbyak::Reg64 scratch, Xbyak::Label& end){
Xbyak::Label normal, normal_or_equal, result_is_result;
code.ucomiss(result, operand);
code.jnp(normal_or_equal);
// If operand == QNaN, result = result.
code.movd(scratch.cvt32(), operand);
code.shl(scratch.cvt32(), 1);
code.cmp(scratch.cvt32(), 0xff800000u);
code.jae(result_is_result);
// If operand == SNaN, let usual NaN code handle it.
code.cmp(scratch.cvt32(), 0xff000000u);
code.ja(normal);
// If result == SNaN, && operand != NaN, result = result.
code.movd(scratch.cvt32(), result);
code.shl(scratch.cvt32(), 1);
code.cmp(scratch.cvt32(), 0xff800000u);
code.jnae(result_is_result);
// If result == QNaN && operand != NaN, result = operand.
code.movaps(result, operand);
code.jmp(end, code.T_NEAR);
code.L(result_is_result);
code.movaps(operand, result);
code.jmp(normal);
code.L(normal_or_equal);
code.jnz(normal);
code.andps(operand, result);
code.L(normal);
}, &Xbyak::CodeGenerator::maxss, CallDenormalsAreZero::Yes);
}
void EmitX64::EmitFPMaxNumeric64(EmitContext& ctx, IR::Inst* inst) {
FPThreeOp<64>(code, ctx, inst, [&](Xbyak::Xmm result, Xbyak::Xmm operand, Xbyak::Reg64 scratch, Xbyak::Label& end){
Xbyak::Label normal, normal_or_equal, result_is_result;
code.ucomisd(result, operand);
code.jnp(normal_or_equal);
// If operand == QNaN, result = result.
code.movq(scratch, operand);
code.shl(scratch, 1);
code.cmp(scratch, code.MConst(qword, 0xfff0'0000'0000'0000u));
code.jae(result_is_result);
// If operand == SNaN, let usual NaN code handle it.
code.cmp(scratch, code.MConst(qword, 0xffe0'0000'0000'0000u));
code.ja(normal);
// If result == SNaN, && operand != NaN, result = result.
code.movq(scratch, result);
code.shl(scratch, 1);
code.cmp(scratch, code.MConst(qword, 0xfff0'0000'0000'0000u));
code.jnae(result_is_result);
// If result == QNaN && operand != NaN, result = operand.
code.movaps(result, operand);
code.jmp(end, code.T_NEAR);
code.L(result_is_result);
code.movaps(operand, result);
code.jmp(normal);
code.L(normal_or_equal);
code.jnz(normal);
code.andps(operand, result);
code.L(normal);
}, &Xbyak::CodeGenerator::maxsd, CallDenormalsAreZero::Yes);
}
template<size_t fsize>
static void EmitFPMin(BlockOfCode& code, EmitContext& ctx, IR::Inst* inst) {
auto args = ctx.reg_alloc.GetArgumentInfo(inst); auto args = ctx.reg_alloc.GetArgumentInfo(inst);
const Xbyak::Xmm result = ctx.reg_alloc.UseScratchXmm(args[0]); const Xbyak::Xmm op1 = ctx.reg_alloc.UseScratchXmm(args[0]);
const Xbyak::Xmm operand = ctx.reg_alloc.UseScratchXmm(args[1]); const Xbyak::Xmm op2 = ctx.reg_alloc.UseScratchXmm(args[1]); // Result stored here!
const Xbyak::Reg64 gpr_scratch = ctx.reg_alloc.ScratchGpr(); Xbyak::Reg tmp = ctx.reg_alloc.ScratchGpr();
tmp.setBit(fsize);
const auto move_to_tmp = [&](const Xbyak::Xmm& xmm) {
if constexpr (fsize == 32) {
code.movd(tmp.cvt32(), xmm);
} else {
code.movq(tmp.cvt64(), xmm);
}
};
if (ctx.FPSCR_FTZ()) { if (ctx.FPSCR_FTZ()) {
DenormalsAreZero<fsize>(code, result, gpr_scratch); DenormalsAreZero<fsize>(code, op1, tmp.cvt64());
DenormalsAreZero<fsize>(code, operand, gpr_scratch); DenormalsAreZero<fsize>(code, op2, tmp.cvt64());
} }
Xbyak::Label equal, end, nan; Xbyak::Label end, z, nan, op2_is_nan, snan, maybe_both_nan, normal;
FCODE(ucomis)(result, operand); FCODE(ucomis)(op1, op2);
code.jz(equal, code.T_NEAR); code.jz(z, code.T_NEAR);
FCODE(mins)(result, operand); code.L(normal);
if constexpr (is_max) {
FCODE(maxs)(op2, op1);
} else {
FCODE(mins)(op2, op1);
}
code.L(end); code.L(end);
code.SwitchToFarCode(); code.SwitchToFarCode();
code.L(equal); code.L(z);
code.jp(nan); code.jp(nan);
code.orps(result, operand); if constexpr (is_max) {
code.andps(op2, op1);
} else {
code.orps(op2, op1);
}
code.jmp(end); code.jmp(end);
// NaN requirements:
// op1 op2 result
// SNaN anything op1
// !SNaN SNaN op2
// QNaN !NaN op2
// !NaN QNaN op1
// QNaN QNaN op1
code.L(nan); code.L(nan);
FCODE(ucomis)(op1, op1);
code.jnp(op2_is_nan);
// op1 is NaN
move_to_tmp(op1);
code.bt(tmp, mantissa_msb_bit);
code.jc(maybe_both_nan);
if (ctx.FPSCR_DN()) { if (ctx.FPSCR_DN()) {
code.movaps(result, code.MConst(xword, fsize == 32 ? f32_nan : f64_nan)); code.L(snan);
code.movaps(op2, code.MConst(xword, FP::FPInfo<FPT>::DefaultNaN()));
code.jmp(end); code.jmp(end);
} else { } else {
EmitProcessNaNs<fsize>(code, result, result, operand, gpr_scratch, end); code.movaps(op2, op1);
code.L(snan);
code.orps(op2, code.MConst(xword, FP::FPInfo<FPT>::mantissa_msb));
code.jmp(end);
} }
code.L(maybe_both_nan);
FCODE(ucomis)(op2, op2);
code.jnp(end, code.T_NEAR);
if (ctx.FPSCR_DN()) {
code.jmp(snan);
} else {
move_to_tmp(op2);
code.bt(tmp.cvt64(), mantissa_msb_bit);
code.jnc(snan);
code.movaps(op2, op1);
code.jmp(end);
}
// op2 is NaN
code.L(op2_is_nan);
move_to_tmp(op2);
code.bt(tmp, mantissa_msb_bit);
code.jnc(snan);
code.movaps(op2, op1);
code.jmp(end);
code.SwitchToNearCode(); code.SwitchToNearCode();
ctx.reg_alloc.DefineValue(inst, result); ctx.reg_alloc.DefineValue(inst, op2);
}
void EmitX64::EmitFPMax32(EmitContext& ctx, IR::Inst* inst) {
EmitFPMinMax<32, true>(code, ctx, inst);
}
void EmitX64::EmitFPMax64(EmitContext& ctx, IR::Inst* inst) {
EmitFPMinMax<64, true>(code, ctx, inst);
}
void EmitX64::EmitFPMaxNumeric32(EmitContext& ctx, IR::Inst* inst) {
EmitFPMinMaxNumeric<32, true>(code, ctx, inst);
}
void EmitX64::EmitFPMaxNumeric64(EmitContext& ctx, IR::Inst* inst) {
EmitFPMinMaxNumeric<64, true>(code, ctx, inst);
} }
void EmitX64::EmitFPMin32(EmitContext& ctx, IR::Inst* inst) { void EmitX64::EmitFPMin32(EmitContext& ctx, IR::Inst* inst) {
EmitFPMin<32>(code, ctx, inst); EmitFPMinMax<32, false>(code, ctx, inst);
} }
void EmitX64::EmitFPMin64(EmitContext& ctx, IR::Inst* inst) { void EmitX64::EmitFPMin64(EmitContext& ctx, IR::Inst* inst) {
EmitFPMin<64>(code, ctx, inst); EmitFPMinMax<64, false>(code, ctx, inst);
} }
void EmitX64::EmitFPMinNumeric32(EmitContext& ctx, IR::Inst* inst) { void EmitX64::EmitFPMinNumeric32(EmitContext& ctx, IR::Inst* inst) {
FPThreeOp<32>(code, ctx, inst, [&](Xbyak::Xmm result, Xbyak::Xmm operand, Xbyak::Reg64 scratch, Xbyak::Label& end){ EmitFPMinMaxNumeric<32, false>(code, ctx, inst);
Xbyak::Label normal, normal_or_equal, result_is_result;
code.ucomiss(result, operand);
code.jnp(normal_or_equal);
// If operand == QNaN, result = result.
code.movd(scratch.cvt32(), operand);
code.shl(scratch.cvt32(), 1);
code.cmp(scratch.cvt32(), 0xff800000u);
code.jae(result_is_result);
// If operand == SNaN, let usual NaN code handle it.
code.cmp(scratch.cvt32(), 0xff000000u);
code.ja(normal);
// If result == SNaN, && operand != NaN, result = result.
code.movd(scratch.cvt32(), result);
code.shl(scratch.cvt32(), 1);
code.cmp(scratch.cvt32(), 0xff800000u);
code.jnae(result_is_result);
// If result == QNaN && operand != NaN, result = operand.
code.movaps(result, operand);
code.jmp(end, code.T_NEAR);
code.L(result_is_result);
code.movaps(operand, result);
code.jmp(normal);
code.L(normal_or_equal);
code.jnz(normal);
code.orps(operand, result);
code.L(normal);
}, &Xbyak::CodeGenerator::minss, CallDenormalsAreZero::Yes);
} }
void EmitX64::EmitFPMinNumeric64(EmitContext& ctx, IR::Inst* inst) { void EmitX64::EmitFPMinNumeric64(EmitContext& ctx, IR::Inst* inst) {
FPThreeOp<64>(code, ctx, inst, [&](Xbyak::Xmm result, Xbyak::Xmm operand, Xbyak::Reg64 scratch, Xbyak::Label& end){ EmitFPMinMaxNumeric<64, false>(code, ctx, inst);
Xbyak::Label normal, normal_or_equal, result_is_result;
code.ucomisd(result, operand);
code.jnp(normal_or_equal);
// If operand == QNaN, result = result.
code.movq(scratch, operand);
code.shl(scratch, 1);
code.cmp(scratch, code.MConst(qword, 0xfff0'0000'0000'0000u));
code.jae(result_is_result);
// If operand == SNaN, let usual NaN code handle it.
code.cmp(scratch, code.MConst(qword, 0xffe0'0000'0000'0000u));
code.ja(normal);
// If result == SNaN, && operand != NaN, result = result.
code.movq(scratch, result);
code.shl(scratch, 1);
code.cmp(scratch, code.MConst(qword, 0xfff0'0000'0000'0000u));
code.jnae(result_is_result);
// If result == QNaN && operand != NaN, result = operand.
code.movaps(result, operand);
code.jmp(end, code.T_NEAR);
code.L(result_is_result);
code.movaps(operand, result);
code.jmp(normal);
code.L(normal_or_equal);
code.jnz(normal);
code.orps(operand, result);
code.L(normal);
}, &Xbyak::CodeGenerator::minsd, CallDenormalsAreZero::Yes);
} }
void EmitX64::EmitFPMul32(EmitContext& ctx, IR::Inst* inst) { void EmitX64::EmitFPMul32(EmitContext& ctx, IR::Inst* inst) {