1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
|
// Copyright 2021 yuzu Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#include "shader_recompiler/frontend/maxwell/translate/impl/half_floating_point_helper.h"
namespace Shader::Maxwell {
namespace {
void HMUL2(TranslatorVisitor& v, u64 insn, Merge merge, bool sat, bool abs_a, bool neg_a,
Swizzle swizzle_a, bool abs_b, bool neg_b, Swizzle swizzle_b, const IR::U32& src_b,
HalfPrecision precision) {
union {
u64 raw;
BitField<0, 8, IR::Reg> dest_reg;
BitField<8, 8, IR::Reg> src_a;
} const hmul2{insn};
auto [lhs_a, rhs_a]{Extract(v.ir, v.X(hmul2.src_a), swizzle_a)};
auto [lhs_b, rhs_b]{Extract(v.ir, src_b, swizzle_b)};
const bool promotion{lhs_a.Type() != lhs_b.Type()};
if (promotion) {
if (lhs_a.Type() == IR::Type::F16) {
lhs_a = v.ir.FPConvert(32, lhs_a);
rhs_a = v.ir.FPConvert(32, rhs_a);
}
if (lhs_b.Type() == IR::Type::F16) {
lhs_b = v.ir.FPConvert(32, lhs_b);
rhs_b = v.ir.FPConvert(32, rhs_b);
}
}
lhs_a = v.ir.FPAbsNeg(lhs_a, abs_a, neg_a);
rhs_a = v.ir.FPAbsNeg(rhs_a, abs_a, neg_a);
lhs_b = v.ir.FPAbsNeg(lhs_b, abs_b, neg_b);
rhs_b = v.ir.FPAbsNeg(rhs_b, abs_b, neg_b);
const IR::FpControl fp_control{
.no_contraction = true,
.rounding = IR::FpRounding::DontCare,
.fmz_mode = HalfPrecision2FmzMode(precision),
};
IR::F16F32F64 lhs{v.ir.FPMul(lhs_a, lhs_b, fp_control)};
IR::F16F32F64 rhs{v.ir.FPMul(rhs_a, rhs_b, fp_control)};
if (precision == HalfPrecision::FMZ && !sat) {
// Do not implement FMZ if SAT is enabled, as it does the logic for us.
// On D3D9 mode, anything * 0 is zero, even NAN and infinity
const IR::F32 zero{v.ir.Imm32(0.0f)};
const IR::U1 lhs_zero_a{v.ir.FPEqual(lhs_a, zero)};
const IR::U1 lhs_zero_b{v.ir.FPEqual(lhs_b, zero)};
const IR::U1 lhs_any_zero{v.ir.LogicalOr(lhs_zero_a, lhs_zero_b)};
lhs = IR::F16F32F64{v.ir.Select(lhs_any_zero, zero, lhs)};
const IR::U1 rhs_zero_a{v.ir.FPEqual(rhs_a, zero)};
const IR::U1 rhs_zero_b{v.ir.FPEqual(rhs_b, zero)};
const IR::U1 rhs_any_zero{v.ir.LogicalOr(rhs_zero_a, rhs_zero_b)};
rhs = IR::F16F32F64{v.ir.Select(rhs_any_zero, zero, rhs)};
}
if (sat) {
lhs = v.ir.FPSaturate(lhs);
rhs = v.ir.FPSaturate(rhs);
}
if (promotion) {
lhs = v.ir.FPConvert(16, lhs);
rhs = v.ir.FPConvert(16, rhs);
}
v.X(hmul2.dest_reg, MergeResult(v.ir, hmul2.dest_reg, lhs, rhs, merge));
}
void HMUL2(TranslatorVisitor& v, u64 insn, bool sat, bool abs_a, bool neg_a, bool abs_b, bool neg_b,
Swizzle swizzle_b, const IR::U32& src_b) {
union {
u64 raw;
BitField<49, 2, Merge> merge;
BitField<47, 2, Swizzle> swizzle_a;
BitField<39, 2, HalfPrecision> precision;
} const hmul2{insn};
HMUL2(v, insn, hmul2.merge, sat, abs_a, neg_a, hmul2.swizzle_a, abs_b, neg_b, swizzle_b, src_b,
hmul2.precision);
}
} // Anonymous namespace
void TranslatorVisitor::HMUL2_reg(u64 insn) {
union {
u64 raw;
BitField<32, 1, u64> sat;
BitField<31, 1, u64> neg_b;
BitField<30, 1, u64> abs_b;
BitField<44, 1, u64> abs_a;
BitField<28, 2, Swizzle> swizzle_b;
} const hmul2{insn};
HMUL2(*this, insn, hmul2.sat != 0, hmul2.abs_a != 0, false, hmul2.abs_b != 0, hmul2.neg_b != 0,
hmul2.swizzle_b, GetReg20(insn));
}
void TranslatorVisitor::HMUL2_cbuf(u64 insn) {
union {
u64 raw;
BitField<52, 1, u64> sat;
BitField<54, 1, u64> abs_b;
BitField<43, 1, u64> neg_a;
BitField<44, 1, u64> abs_a;
} const hmul2{insn};
HMUL2(*this, insn, hmul2.sat != 0, hmul2.abs_a != 0, hmul2.neg_a != 0, hmul2.abs_b != 0, false,
Swizzle::F32, GetCbuf(insn));
}
void TranslatorVisitor::HMUL2_imm(u64 insn) {
union {
u64 raw;
BitField<52, 1, u64> sat;
BitField<56, 1, u64> neg_high;
BitField<30, 9, u64> high;
BitField<29, 1, u64> neg_low;
BitField<20, 9, u64> low;
BitField<43, 1, u64> neg_a;
BitField<44, 1, u64> abs_a;
} const hmul2{insn};
const u32 imm{
static_cast<u32>(hmul2.low << 6) | static_cast<u32>((hmul2.neg_low != 0 ? 1 : 0) << 15) |
static_cast<u32>(hmul2.high << 22) | static_cast<u32>((hmul2.neg_high != 0 ? 1 : 0) << 31)};
HMUL2(*this, insn, hmul2.sat != 0, hmul2.abs_a != 0, hmul2.neg_a != 0, false, false,
Swizzle::H1_H0, ir.Imm32(imm));
}
void TranslatorVisitor::HMUL2_32I(u64 insn) {
union {
u64 raw;
BitField<55, 2, HalfPrecision> precision;
BitField<52, 1, u64> sat;
BitField<53, 2, Swizzle> swizzle_a;
BitField<20, 32, u64> imm32;
} const hmul2{insn};
const u32 imm{static_cast<u32>(hmul2.imm32)};
HMUL2(*this, insn, Merge::H1_H0, hmul2.sat != 0, false, false, hmul2.swizzle_a, false, false,
Swizzle::H1_H0, ir.Imm32(imm), hmul2.precision);
}
} // namespace Shader::Maxwell
|
