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
|
// 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 HSETP2(TranslatorVisitor& v, u64 insn, const IR::U32& src_b, bool neg_b, bool abs_b,
Swizzle swizzle_b, FPCompareOp compare_op, bool h_and) {
union {
u64 insn;
BitField<8, 8, IR::Reg> src_a_reg;
BitField<3, 3, IR::Pred> dest_pred_a;
BitField<0, 3, IR::Pred> dest_pred_b;
BitField<39, 3, IR::Pred> pred;
BitField<42, 1, u64> neg_pred;
BitField<43, 1, u64> neg_a;
BitField<45, 2, BooleanOp> bop;
BitField<44, 1, u64> abs_a;
BitField<6, 1, u64> ftz;
BitField<47, 2, Swizzle> swizzle_a;
} const hsetp2{insn};
auto [lhs_a, rhs_a]{Extract(v.ir, v.X(hsetp2.src_a_reg), hsetp2.swizzle_a)};
auto [lhs_b, rhs_b]{Extract(v.ir, src_b, swizzle_b)};
if (lhs_a.Type() != lhs_b.Type()) {
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, hsetp2.abs_a != 0, hsetp2.neg_a != 0);
rhs_a = v.ir.FPAbsNeg(rhs_a, hsetp2.abs_a != 0, hsetp2.neg_a != 0);
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 control{
.no_contraction = false,
.rounding = IR::FpRounding::DontCare,
.fmz_mode = (hsetp2.ftz != 0 ? IR::FmzMode::FTZ : IR::FmzMode::None),
};
IR::U1 pred{v.ir.GetPred(hsetp2.pred)};
if (hsetp2.neg_pred != 0) {
pred = v.ir.LogicalNot(pred);
}
const IR::U1 cmp_result_lhs{FloatingPointCompare(v.ir, lhs_a, lhs_b, compare_op, control)};
const IR::U1 cmp_result_rhs{FloatingPointCompare(v.ir, rhs_a, rhs_b, compare_op, control)};
const IR::U1 bop_result_lhs{PredicateCombine(v.ir, cmp_result_lhs, pred, hsetp2.bop)};
const IR::U1 bop_result_rhs{PredicateCombine(v.ir, cmp_result_rhs, pred, hsetp2.bop)};
if (h_and) {
auto result = v.ir.LogicalAnd(bop_result_lhs, bop_result_rhs);
v.ir.SetPred(hsetp2.dest_pred_a, result);
v.ir.SetPred(hsetp2.dest_pred_b, v.ir.LogicalNot(result));
} else {
v.ir.SetPred(hsetp2.dest_pred_a, bop_result_lhs);
v.ir.SetPred(hsetp2.dest_pred_b, bop_result_rhs);
}
}
} // Anonymous namespace
void TranslatorVisitor::HSETP2_reg(u64 insn) {
union {
u64 insn;
BitField<30, 1, u64> abs_b;
BitField<49, 1, u64> h_and;
BitField<31, 1, u64> neg_b;
BitField<35, 4, FPCompareOp> compare_op;
BitField<28, 2, Swizzle> swizzle_b;
} const hsetp2{insn};
HSETP2(*this, insn, GetReg20(insn), hsetp2.neg_b != 0, hsetp2.abs_b != 0, hsetp2.swizzle_b,
hsetp2.compare_op, hsetp2.h_and != 0);
}
void TranslatorVisitor::HSETP2_cbuf(u64 insn) {
union {
u64 insn;
BitField<53, 1, u64> h_and;
BitField<54, 1, u64> abs_b;
BitField<56, 1, u64> neg_b;
BitField<49, 4, FPCompareOp> compare_op;
} const hsetp2{insn};
HSETP2(*this, insn, GetCbuf(insn), hsetp2.neg_b != 0, hsetp2.abs_b != 0, Swizzle::F32,
hsetp2.compare_op, hsetp2.h_and != 0);
}
void TranslatorVisitor::HSETP2_imm(u64 insn) {
union {
u64 insn;
BitField<53, 1, u64> h_and;
BitField<54, 1, u64> ftz;
BitField<49, 4, FPCompareOp> compare_op;
BitField<56, 1, u64> neg_high;
BitField<30, 9, u64> high;
BitField<29, 1, u64> neg_low;
BitField<20, 9, u64> low;
} const hsetp2{insn};
const u32 imm{static_cast<u32>(hsetp2.low << 6) |
static_cast<u32>((hsetp2.neg_low != 0 ? 1 : 0) << 15) |
static_cast<u32>(hsetp2.high << 22) |
static_cast<u32>((hsetp2.neg_high != 0 ? 1 : 0) << 31)};
HSETP2(*this, insn, ir.Imm32(imm), false, false, Swizzle::H1_H0, hsetp2.compare_op,
hsetp2.h_and != 0);
}
} // namespace Shader::Maxwell
|