citra/src/video_core/shader/shader_ir.cpp
2021-01-23 13:16:37 -05:00

510 lines
18 KiB
C++

// Copyright 2018 yuzu Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#include <algorithm>
#include <array>
#include <cmath>
#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.h"
#include "video_core/shader/node_helper.h"
#include "video_core/shader/registry.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_,
Registry& registry_)
: program_code{program_code_}, main_offset{main_offset_}, settings{settings_}, registry{
registry_} {
Decode();
PostDecode();
}
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::GetCustomVariable(u32 id) {
return MakeNode<CustomVarNode>(id);
}
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_);
used_cbufs.try_emplace(index).first->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_);
used_cbufs.try_emplace(index).first->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) {
MarkAttributeUsage(index, element);
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) {
MarkAttributeUsage(index, element);
used_output_attributes.insert(index);
return MakeNode<AbufNode>(index, static_cast<u32>(element), std::move(buffer));
}
Node ShaderIR::GetInternalFlag(InternalFlag flag, bool negated) const {
Node node = MakeNode<InternalFlagNode>(flag);
if (negated) {
return Operation(OperationCode::LogicalNegate, std::move(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));
return value;
case Register::Size::Word:
// Default - do nothing
return value;
default:
UNREACHABLE_MSG("Unimplemented conversion size: {}", 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) {
if (condition == PredCondition::T) {
return GetPredicate(true);
} else if (condition == PredCondition::F) {
return GetPredicate(false);
}
static constexpr std::array comparison_table{
OperationCode(0),
OperationCode::LogicalFOrdLessThan, // LT
OperationCode::LogicalFOrdEqual, // EQ
OperationCode::LogicalFOrdLessEqual, // LE
OperationCode::LogicalFOrdGreaterThan, // GT
OperationCode::LogicalFOrdNotEqual, // NE
OperationCode::LogicalFOrdGreaterEqual, // GE
OperationCode::LogicalFOrdered, // NUM
OperationCode::LogicalFUnordered, // NAN
OperationCode::LogicalFUnordLessThan, // LTU
OperationCode::LogicalFUnordEqual, // EQU
OperationCode::LogicalFUnordLessEqual, // LEU
OperationCode::LogicalFUnordGreaterThan, // GTU
OperationCode::LogicalFUnordNotEqual, // NEU
OperationCode::LogicalFUnordGreaterEqual, // GEU
};
const std::size_t index = static_cast<std::size_t>(condition);
ASSERT_MSG(index < std::size(comparison_table), "Invalid condition={}", index);
return Operation(comparison_table[index], op_a, op_b);
}
Node ShaderIR::GetPredicateComparisonInteger(PredCondition condition, bool is_signed, Node op_a,
Node op_b) {
static constexpr std::array comparison_table{
std::pair{PredCondition::LT, OperationCode::LogicalILessThan},
std::pair{PredCondition::EQ, OperationCode::LogicalIEqual},
std::pair{PredCondition::LE, OperationCode::LogicalILessEqual},
std::pair{PredCondition::GT, OperationCode::LogicalIGreaterThan},
std::pair{PredCondition::NE, OperationCode::LogicalINotEqual},
std::pair{PredCondition::GE, OperationCode::LogicalIGreaterEqual},
};
const auto comparison =
std::find_if(comparison_table.cbegin(), comparison_table.cend(),
[condition](const auto entry) { return condition == entry.first; });
UNIMPLEMENTED_IF_MSG(comparison == comparison_table.cend(),
"Unknown predicate comparison operation");
return SignedOperation(comparison->second, is_signed, NO_PRECISE, std::move(op_a),
std::move(op_b));
}
Node ShaderIR::GetPredicateComparisonHalf(Tegra::Shader::PredCondition condition, Node op_a,
Node op_b) {
static constexpr std::array comparison_table{
std::pair{PredCondition::LT, OperationCode::Logical2HLessThan},
std::pair{PredCondition::EQ, OperationCode::Logical2HEqual},
std::pair{PredCondition::LE, OperationCode::Logical2HLessEqual},
std::pair{PredCondition::GT, OperationCode::Logical2HGreaterThan},
std::pair{PredCondition::NE, OperationCode::Logical2HNotEqual},
std::pair{PredCondition::GE, OperationCode::Logical2HGreaterEqual},
std::pair{PredCondition::LTU, OperationCode::Logical2HLessThanWithNan},
std::pair{PredCondition::LEU, OperationCode::Logical2HLessEqualWithNan},
std::pair{PredCondition::GTU, OperationCode::Logical2HGreaterThanWithNan},
std::pair{PredCondition::NEU, OperationCode::Logical2HNotEqualWithNan},
std::pair{PredCondition::GEU, OperationCode::Logical2HGreaterEqualWithNan},
};
const auto comparison =
std::find_if(comparison_table.cbegin(), comparison_table.cend(),
[condition](const auto entry) { return condition == entry.first; });
UNIMPLEMENTED_IF_MSG(comparison == comparison_table.cend(),
"Unknown predicate comparison operation");
return Operation(comparison->second, NO_PRECISE, std::move(op_a), std::move(op_b));
}
OperationCode ShaderIR::GetPredicateCombiner(PredOperation operation) {
static constexpr std::array operation_table{
OperationCode::LogicalAnd,
OperationCode::LogicalOr,
OperationCode::LogicalXor,
};
const auto index = static_cast<std::size_t>(operation);
if (index >= operation_table.size()) {
UNIMPLEMENTED_MSG("Unknown predicate operation.");
return {};
}
return operation_table[index];
}
Node ShaderIR::GetConditionCode(ConditionCode cc) const {
switch (cc) {
case ConditionCode::NEU:
return GetInternalFlag(InternalFlag::Zero, true);
case ConditionCode::FCSM_TR:
UNIMPLEMENTED_MSG("EXIT.FCSM_TR is not implemented");
return MakeNode<PredicateNode>(Pred::NeverExecute, false);
default:
UNIMPLEMENTED_MSG("Unimplemented condition code: {}", 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::LogicalFOrdEqual, 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;
}
switch (value->index()) {
case 0: // Operation Node
SearchOperands(bb, value);
break;
case 2: // General Purpose Node
if (const auto* gpr = std::get_if<GprNode>(value.get())) {
LOG_DEBUG(HW_GPU, "GprNode: index={}", gpr->GetIndex());
Node zerop = Operation(OperationCode::LogicalIEqual, std::move(value),
Immediate(gpr->GetIndex()));
SetInternalFlag(bb, InternalFlag::Zero, std::move(zerop));
}
break;
default:
Node zerop = Operation(OperationCode::LogicalIEqual, std::move(value), Immediate(0));
SetInternalFlag(bb, InternalFlag::Zero, std::move(zerop));
LOG_WARNING(HW_GPU, "Node Type: {}", value->index());
break;
}
}
void ShaderIR::SearchOperands(NodeBlock& nb, Node var) {
const auto* op = std::get_if<OperationNode>(var.get());
if (op == nullptr) {
return;
}
if (op->GetOperandsCount() == 0) {
return;
}
for (auto& operand : op->GetOperands()) {
switch (operand->index()) {
case 0: // Operation Node
return SearchOperands(nb, operand);
case 2: // General Purpose Node
if (const auto* gpr = std::get_if<GprNode>(operand.get())) {
LOG_DEBUG(HW_GPU, "Child GprNode: index={}", gpr->GetIndex());
Node zerop = Operation(OperationCode::LogicalIEqual, std::move(operand),
Immediate(gpr->GetIndex()));
SetInternalFlag(nb, InternalFlag::Zero, std::move(zerop));
}
break;
default:
LOG_WARNING(HW_GPU, "Child Node Type: {}", operand->index());
break;
}
}
}
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));
}
void ShaderIR::MarkAttributeUsage(Attribute::Index index, u64 element) {
switch (index) {
case 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;
}
break;
case Attribute::Index::TessCoordInstanceIDVertexID:
switch (element) {
case 2:
uses_instance_id = true;
break;
case 3:
uses_vertex_id = true;
break;
}
break;
case Attribute::Index::ClipDistances0123:
case Attribute::Index::ClipDistances4567: {
const u64 clip_index = (index == Attribute::Index::ClipDistances4567 ? 4 : 0) + element;
used_clip_distances.at(clip_index) = true;
break;
}
case Attribute::Index::FrontColor:
case Attribute::Index::FrontSecondaryColor:
case Attribute::Index::BackColor:
case Attribute::Index::BackSecondaryColor:
uses_legacy_varyings = true;
break;
default:
if (index >= Attribute::Index::TexCoord_0 && index <= Attribute::Index::TexCoord_7) {
uses_legacy_varyings = true;
}
break;
}
}
std::size_t ShaderIR::DeclareAmend(Node new_amend) {
const auto id = amend_code.size();
amend_code.push_back(std::move(new_amend));
return id;
}
u32 ShaderIR::NewCustomVariable() {
return num_custom_variables++;
}
} // namespace VideoCommon::Shader