summaryrefslogtreecommitdiffstats
path: root/src/video_core/shader/shader_ir.cpp
blob: 1d718ccc6e3aa8d62f50fbe52961c825bcebe3d4 (plain) (blame)
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
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
// Copyright 2018 yuzu Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.

#include <cmath>
#include <unordered_map>

#include "common/assert.h"
#include "common/common_types.h"
#include "common/logging/log.h"
#include "video_core/engines/shader_bytecode.h"
#include "video_core/shader/node_helper.h"
#include "video_core/shader/shader_ir.h"

namespace VideoCommon::Shader {

using Tegra::Shader::Attribute;
using Tegra::Shader::Instruction;
using Tegra::Shader::IpaMode;
using Tegra::Shader::Pred;
using Tegra::Shader::PredCondition;
using Tegra::Shader::PredOperation;
using Tegra::Shader::Register;

ShaderIR::ShaderIR(const ProgramCode& program_code, u32 main_offset, CompilerSettings settings,
                   ConstBufferLocker& locker)
    : program_code{program_code}, main_offset{main_offset}, settings{settings}, locker{locker} {
    Decode();
}

ShaderIR::~ShaderIR() = default;

Node ShaderIR::GetRegister(Register reg) {
    if (reg != Register::ZeroIndex) {
        used_registers.insert(static_cast<u32>(reg));
    }
    return MakeNode<GprNode>(reg);
}

Node ShaderIR::GetImmediate19(Instruction instr) {
    return Immediate(instr.alu.GetImm20_19());
}

Node ShaderIR::GetImmediate32(Instruction instr) {
    return Immediate(instr.alu.GetImm20_32());
}

Node ShaderIR::GetConstBuffer(u64 index_, u64 offset_) {
    const auto index = static_cast<u32>(index_);
    const auto offset = static_cast<u32>(offset_);

    const auto [entry, is_new] = used_cbufs.try_emplace(index);
    entry->second.MarkAsUsed(offset);

    return MakeNode<CbufNode>(index, Immediate(offset));
}

Node ShaderIR::GetConstBufferIndirect(u64 index_, u64 offset_, Node node) {
    const auto index = static_cast<u32>(index_);
    const auto offset = static_cast<u32>(offset_);

    const auto [entry, is_new] = used_cbufs.try_emplace(index);
    entry->second.MarkAsUsedIndirect();

    Node final_offset = [&] {
        // Attempt to inline constant buffer without a variable offset. This is done to allow
        // tracking LDC calls.
        if (const auto gpr = std::get_if<GprNode>(&*node)) {
            if (gpr->GetIndex() == Register::ZeroIndex) {
                return Immediate(offset);
            }
        }
        return Operation(OperationCode::UAdd, NO_PRECISE, std::move(node), Immediate(offset));
    }();
    return MakeNode<CbufNode>(index, std::move(final_offset));
}

Node ShaderIR::GetPredicate(u64 pred_, bool negated) {
    const auto pred = static_cast<Pred>(pred_);
    if (pred != Pred::UnusedIndex && pred != Pred::NeverExecute) {
        used_predicates.insert(pred);
    }

    return MakeNode<PredicateNode>(pred, negated);
}

Node ShaderIR::GetPredicate(bool immediate) {
    return GetPredicate(static_cast<u64>(immediate ? Pred::UnusedIndex : Pred::NeverExecute));
}

Node ShaderIR::GetInputAttribute(Attribute::Index index, u64 element, Node buffer) {
    used_input_attributes.emplace(index);
    return MakeNode<AbufNode>(index, static_cast<u32>(element), std::move(buffer));
}

Node ShaderIR::GetPhysicalInputAttribute(Tegra::Shader::Register physical_address, Node buffer) {
    uses_physical_attributes = true;
    return MakeNode<AbufNode>(GetRegister(physical_address), buffer);
}

Node ShaderIR::GetOutputAttribute(Attribute::Index index, u64 element, Node buffer) {
    if (index == Attribute::Index::LayerViewportPointSize) {
        switch (element) {
        case 0:
            UNIMPLEMENTED();
            break;
        case 1:
            uses_layer = true;
            break;
        case 2:
            uses_viewport_index = true;
            break;
        case 3:
            uses_point_size = true;
            break;
        }
    }
    if (index == Attribute::Index::TessCoordInstanceIDVertexID) {
        switch (element) {
        case 2:
            uses_instance_id = true;
            break;
        case 3:
            uses_vertex_id = true;
            break;
        default:
            break;
        }
    }
    if (index == Attribute::Index::ClipDistances0123 ||
        index == Attribute::Index::ClipDistances4567) {
        const auto clip_index =
            static_cast<u32>((index == Attribute::Index::ClipDistances4567 ? 1 : 0) + element);
        used_clip_distances.at(clip_index) = true;
    }
    used_output_attributes.insert(index);

    return MakeNode<AbufNode>(index, static_cast<u32>(element), std::move(buffer));
}

Node ShaderIR::GetInternalFlag(InternalFlag flag, bool negated) const {
    const Node node = MakeNode<InternalFlagNode>(flag);
    if (negated) {
        return Operation(OperationCode::LogicalNegate, node);
    }
    return node;
}

Node ShaderIR::GetLocalMemory(Node address) {
    return MakeNode<LmemNode>(std::move(address));
}

Node ShaderIR::GetSharedMemory(Node address) {
    return MakeNode<SmemNode>(std::move(address));
}

Node ShaderIR::GetTemporary(u32 id) {
    return GetRegister(Register::ZeroIndex + 1 + id);
}

Node ShaderIR::GetOperandAbsNegFloat(Node value, bool absolute, bool negate) {
    if (absolute) {
        value = Operation(OperationCode::FAbsolute, NO_PRECISE, std::move(value));
    }
    if (negate) {
        value = Operation(OperationCode::FNegate, NO_PRECISE, std::move(value));
    }
    return value;
}

Node ShaderIR::GetSaturatedFloat(Node value, bool saturate) {
    if (!saturate) {
        return value;
    }

    Node positive_zero = Immediate(std::copysignf(0, 1));
    Node positive_one = Immediate(1.0f);
    return Operation(OperationCode::FClamp, NO_PRECISE, std::move(value), std::move(positive_zero),
                     std::move(positive_one));
}

Node ShaderIR::ConvertIntegerSize(Node value, Register::Size size, bool is_signed) {
    switch (size) {
    case Register::Size::Byte:
        value = SignedOperation(OperationCode::ILogicalShiftLeft, is_signed, NO_PRECISE,
                                std::move(value), Immediate(24));
        value = SignedOperation(OperationCode::IArithmeticShiftRight, is_signed, NO_PRECISE,
                                std::move(value), Immediate(24));
        return value;
    case Register::Size::Short:
        value = SignedOperation(OperationCode::ILogicalShiftLeft, is_signed, NO_PRECISE,
                                std::move(value), Immediate(16));
        value = SignedOperation(OperationCode::IArithmeticShiftRight, is_signed, NO_PRECISE,
                                std::move(value), Immediate(16));
    case Register::Size::Word:
        // Default - do nothing
        return value;
    default:
        UNREACHABLE_MSG("Unimplemented conversion size: {}", static_cast<u32>(size));
        return value;
    }
}

Node ShaderIR::GetOperandAbsNegInteger(Node value, bool absolute, bool negate, bool is_signed) {
    if (!is_signed) {
        // Absolute or negate on an unsigned is pointless
        return value;
    }
    if (absolute) {
        value = Operation(OperationCode::IAbsolute, NO_PRECISE, std::move(value));
    }
    if (negate) {
        value = Operation(OperationCode::INegate, NO_PRECISE, std::move(value));
    }
    return value;
}

Node ShaderIR::UnpackHalfImmediate(Instruction instr, bool has_negation) {
    Node value = Immediate(instr.half_imm.PackImmediates());
    if (!has_negation) {
        return value;
    }

    Node first_negate = GetPredicate(instr.half_imm.first_negate != 0);
    Node second_negate = GetPredicate(instr.half_imm.second_negate != 0);

    return Operation(OperationCode::HNegate, NO_PRECISE, std::move(value), std::move(first_negate),
                     std::move(second_negate));
}

Node ShaderIR::UnpackHalfFloat(Node value, Tegra::Shader::HalfType type) {
    return Operation(OperationCode::HUnpack, type, std::move(value));
}

Node ShaderIR::HalfMerge(Node dest, Node src, Tegra::Shader::HalfMerge merge) {
    switch (merge) {
    case Tegra::Shader::HalfMerge::H0_H1:
        return src;
    case Tegra::Shader::HalfMerge::F32:
        return Operation(OperationCode::HMergeF32, std::move(src));
    case Tegra::Shader::HalfMerge::Mrg_H0:
        return Operation(OperationCode::HMergeH0, std::move(dest), std::move(src));
    case Tegra::Shader::HalfMerge::Mrg_H1:
        return Operation(OperationCode::HMergeH1, std::move(dest), std::move(src));
    }
    UNREACHABLE();
    return src;
}

Node ShaderIR::GetOperandAbsNegHalf(Node value, bool absolute, bool negate) {
    if (absolute) {
        value = Operation(OperationCode::HAbsolute, NO_PRECISE, std::move(value));
    }
    if (negate) {
        value = Operation(OperationCode::HNegate, NO_PRECISE, std::move(value), GetPredicate(true),
                          GetPredicate(true));
    }
    return value;
}

Node ShaderIR::GetSaturatedHalfFloat(Node value, bool saturate) {
    if (!saturate) {
        return value;
    }

    Node positive_zero = Immediate(std::copysignf(0, 1));
    Node positive_one = Immediate(1.0f);
    return Operation(OperationCode::HClamp, NO_PRECISE, std::move(value), std::move(positive_zero),
                     std::move(positive_one));
}

Node ShaderIR::GetPredicateComparisonFloat(PredCondition condition, Node op_a, Node op_b) {
    const std::unordered_map<PredCondition, OperationCode> PredicateComparisonTable = {
        {PredCondition::LessThan, OperationCode::LogicalFLessThan},
        {PredCondition::Equal, OperationCode::LogicalFEqual},
        {PredCondition::LessEqual, OperationCode::LogicalFLessEqual},
        {PredCondition::GreaterThan, OperationCode::LogicalFGreaterThan},
        {PredCondition::NotEqual, OperationCode::LogicalFNotEqual},
        {PredCondition::GreaterEqual, OperationCode::LogicalFGreaterEqual},
        {PredCondition::LessThanWithNan, OperationCode::LogicalFLessThan},
        {PredCondition::NotEqualWithNan, OperationCode::LogicalFNotEqual},
        {PredCondition::LessEqualWithNan, OperationCode::LogicalFLessEqual},
        {PredCondition::GreaterThanWithNan, OperationCode::LogicalFGreaterThan},
        {PredCondition::GreaterEqualWithNan, OperationCode::LogicalFGreaterEqual}};

    const auto comparison{PredicateComparisonTable.find(condition)};
    UNIMPLEMENTED_IF_MSG(comparison == PredicateComparisonTable.end(),
                         "Unknown predicate comparison operation");

    Node predicate = Operation(comparison->second, NO_PRECISE, op_a, op_b);

    if (condition == PredCondition::LessThanWithNan ||
        condition == PredCondition::NotEqualWithNan ||
        condition == PredCondition::LessEqualWithNan ||
        condition == PredCondition::GreaterThanWithNan ||
        condition == PredCondition::GreaterEqualWithNan) {
        predicate = Operation(OperationCode::LogicalOr, predicate,
                              Operation(OperationCode::LogicalFIsNan, op_a));
        predicate = Operation(OperationCode::LogicalOr, predicate,
                              Operation(OperationCode::LogicalFIsNan, op_b));
    }

    return predicate;
}

Node ShaderIR::GetPredicateComparisonInteger(PredCondition condition, bool is_signed, Node op_a,
                                             Node op_b) {
    const std::unordered_map<PredCondition, OperationCode> PredicateComparisonTable = {
        {PredCondition::LessThan, OperationCode::LogicalILessThan},
        {PredCondition::Equal, OperationCode::LogicalIEqual},
        {PredCondition::LessEqual, OperationCode::LogicalILessEqual},
        {PredCondition::GreaterThan, OperationCode::LogicalIGreaterThan},
        {PredCondition::NotEqual, OperationCode::LogicalINotEqual},
        {PredCondition::GreaterEqual, OperationCode::LogicalIGreaterEqual},
        {PredCondition::LessThanWithNan, OperationCode::LogicalILessThan},
        {PredCondition::NotEqualWithNan, OperationCode::LogicalINotEqual},
        {PredCondition::LessEqualWithNan, OperationCode::LogicalILessEqual},
        {PredCondition::GreaterThanWithNan, OperationCode::LogicalIGreaterThan},
        {PredCondition::GreaterEqualWithNan, OperationCode::LogicalIGreaterEqual}};

    const auto comparison{PredicateComparisonTable.find(condition)};
    UNIMPLEMENTED_IF_MSG(comparison == PredicateComparisonTable.end(),
                         "Unknown predicate comparison operation");

    Node predicate = SignedOperation(comparison->second, is_signed, NO_PRECISE, std::move(op_a),
                                     std::move(op_b));

    UNIMPLEMENTED_IF_MSG(condition == PredCondition::LessThanWithNan ||
                             condition == PredCondition::NotEqualWithNan ||
                             condition == PredCondition::LessEqualWithNan ||
                             condition == PredCondition::GreaterThanWithNan ||
                             condition == PredCondition::GreaterEqualWithNan,
                         "NaN comparisons for integers are not implemented");
    return predicate;
}

Node ShaderIR::GetPredicateComparisonHalf(Tegra::Shader::PredCondition condition, Node op_a,
                                          Node op_b) {
    const std::unordered_map<PredCondition, OperationCode> PredicateComparisonTable = {
        {PredCondition::LessThan, OperationCode::Logical2HLessThan},
        {PredCondition::Equal, OperationCode::Logical2HEqual},
        {PredCondition::LessEqual, OperationCode::Logical2HLessEqual},
        {PredCondition::GreaterThan, OperationCode::Logical2HGreaterThan},
        {PredCondition::NotEqual, OperationCode::Logical2HNotEqual},
        {PredCondition::GreaterEqual, OperationCode::Logical2HGreaterEqual},
        {PredCondition::LessThanWithNan, OperationCode::Logical2HLessThanWithNan},
        {PredCondition::NotEqualWithNan, OperationCode::Logical2HNotEqualWithNan},
        {PredCondition::LessEqualWithNan, OperationCode::Logical2HLessEqualWithNan},
        {PredCondition::GreaterThanWithNan, OperationCode::Logical2HGreaterThanWithNan},
        {PredCondition::GreaterEqualWithNan, OperationCode::Logical2HGreaterEqualWithNan}};

    const auto comparison{PredicateComparisonTable.find(condition)};
    UNIMPLEMENTED_IF_MSG(comparison == PredicateComparisonTable.end(),
                         "Unknown predicate comparison operation");

    return Operation(comparison->second, NO_PRECISE, std::move(op_a), std::move(op_b));
}

OperationCode ShaderIR::GetPredicateCombiner(PredOperation operation) {
    const std::unordered_map<PredOperation, OperationCode> PredicateOperationTable = {
        {PredOperation::And, OperationCode::LogicalAnd},
        {PredOperation::Or, OperationCode::LogicalOr},
        {PredOperation::Xor, OperationCode::LogicalXor},
    };

    const auto op = PredicateOperationTable.find(operation);
    UNIMPLEMENTED_IF_MSG(op == PredicateOperationTable.end(), "Unknown predicate operation");
    return op->second;
}

Node ShaderIR::GetConditionCode(Tegra::Shader::ConditionCode cc) const {
    switch (cc) {
    case Tegra::Shader::ConditionCode::NEU:
        return GetInternalFlag(InternalFlag::Zero, true);
    default:
        UNIMPLEMENTED_MSG("Unimplemented condition code: {}", static_cast<u32>(cc));
        return MakeNode<PredicateNode>(Pred::NeverExecute, false);
    }
}

void ShaderIR::SetRegister(NodeBlock& bb, Register dest, Node src) {
    bb.push_back(Operation(OperationCode::Assign, GetRegister(dest), std::move(src)));
}

void ShaderIR::SetPredicate(NodeBlock& bb, u64 dest, Node src) {
    bb.push_back(Operation(OperationCode::LogicalAssign, GetPredicate(dest), std::move(src)));
}

void ShaderIR::SetInternalFlag(NodeBlock& bb, InternalFlag flag, Node value) {
    bb.push_back(Operation(OperationCode::LogicalAssign, GetInternalFlag(flag), std::move(value)));
}

void ShaderIR::SetLocalMemory(NodeBlock& bb, Node address, Node value) {
    bb.push_back(
        Operation(OperationCode::Assign, GetLocalMemory(std::move(address)), std::move(value)));
}

void ShaderIR::SetSharedMemory(NodeBlock& bb, Node address, Node value) {
    bb.push_back(
        Operation(OperationCode::Assign, GetSharedMemory(std::move(address)), std::move(value)));
}

void ShaderIR::SetTemporary(NodeBlock& bb, u32 id, Node value) {
    SetRegister(bb, Register::ZeroIndex + 1 + id, std::move(value));
}

void ShaderIR::SetInternalFlagsFromFloat(NodeBlock& bb, Node value, bool sets_cc) {
    if (!sets_cc) {
        return;
    }
    Node zerop = Operation(OperationCode::LogicalFEqual, std::move(value), Immediate(0.0f));
    SetInternalFlag(bb, InternalFlag::Zero, std::move(zerop));
    LOG_WARNING(HW_GPU, "Condition codes implementation is incomplete");
}

void ShaderIR::SetInternalFlagsFromInteger(NodeBlock& bb, Node value, bool sets_cc) {
    if (!sets_cc) {
        return;
    }
    Node zerop = Operation(OperationCode::LogicalIEqual, std::move(value), Immediate(0));
    SetInternalFlag(bb, InternalFlag::Zero, std::move(zerop));
    LOG_WARNING(HW_GPU, "Condition codes implementation is incomplete");
}

Node ShaderIR::BitfieldExtract(Node value, u32 offset, u32 bits) {
    return Operation(OperationCode::UBitfieldExtract, NO_PRECISE, std::move(value),
                     Immediate(offset), Immediate(bits));
}

Node ShaderIR::BitfieldInsert(Node base, Node insert, u32 offset, u32 bits) {
    return Operation(OperationCode::UBitfieldInsert, NO_PRECISE, base, insert, Immediate(offset),
                     Immediate(bits));
}

} // namespace VideoCommon::Shader