shadPS4/src/shader_recompiler/recompiler.cpp
Lander Gallastegi f9bbde9c79
video_core: Implement DMA. (#2819)
* Import memory

* 64K pages and fix memory mapping

* Queue coverage

* Buffer syncing, faulted readback adn BDA in Buffer

* Base DMA implementation

* Preparations for implementing SPV DMA access

* Base impl (pending 16K pages and getbuffersize)

* 16K pages and stack overflow fix

* clang-format

* clang-format but for real this time

* Try to fix macOS build

* Correct decltype

* Add testing log

* Fix stride and patch phi node blocks

* No need to check if it is a deleted buffer

* Clang format once more

* Offset in bytes

* Removed host buffers (may do it in another PR)

Also some random barrier fixes

* Add IR dumping from my read-const branch

* clang-format

* Correct size insteed of end

* Fix incorrect assert

* Possible fix for NieR deadlock

* Copy to avoid deadlock

* Use 2 mutexes insteed of copy

* Attempt to range sync error

* Revert "Attempt to range sync error"

This reverts commit dd287b48682b50f215680bb0956e39c2809bf3fe.

* Fix size truncated when syncing range

And memory barrier

* Some fixes (and async testing (doesn't work))

* Use compute to parse fault buffer

* Process faults on submit

* Only sync in the first time we see a readconst

Thsi is partialy wrong. We need to save the state into the submission context itself, not the rasterizer since we can yield and process another sumission (if im not understanding wrong).

* Use spec const and 32 bit atomic

* 32 bit counter

* Fix store_index

* Better sync (WIP, breaks PR now)

* Fixes for better sync

* Better sync

* Remove memory coveragte logic

* Point sirit to upstream

* Less waiting and barriers

* Correctly checkout moltenvk

* Bring back applying pending operations in wait

* Sync the whole buffer insteed of only the range

* Implement recursive shared/scoped locks

* Iterators

* Faster syncing with ranges

* Some alignment fixes

* fixed clang format

* Fix clang-format again

* Port page_manager from readbacks-poc

* clang-format

* Defer memory protect

* Remove RENDERER_TRACE

* Experiment: only sync on first readconst

* Added profiling (will be removed)

* Don't sync entire buffers

* Added logging for testing

* Updated temporary workaround to use 4k pages

* clang.-format

* Cleanup part 1

* Make ReadConst a SPIR-V function

---------

Co-authored-by: georgemoralis <giorgosmrls@gmail.com>
2025-05-22 21:00:15 +03:00

93 lines
3.8 KiB
C++

// SPDX-FileCopyrightText: Copyright 2024 shadPS4 Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#include "shader_recompiler/frontend/control_flow_graph.h"
#include "shader_recompiler/frontend/decode.h"
#include "shader_recompiler/frontend/structured_control_flow.h"
#include "shader_recompiler/ir/passes/ir_passes.h"
#include "shader_recompiler/ir/post_order.h"
#include "shader_recompiler/recompiler.h"
namespace Shader {
IR::BlockList GenerateBlocks(const IR::AbstractSyntaxList& syntax_list) {
size_t num_syntax_blocks{};
for (const auto& node : syntax_list) {
if (node.type == IR::AbstractSyntaxNode::Type::Block) {
++num_syntax_blocks;
}
}
IR::BlockList blocks;
blocks.reserve(num_syntax_blocks);
u32 order_index{};
for (const auto& node : syntax_list) {
if (node.type == IR::AbstractSyntaxNode::Type::Block) {
blocks.push_back(node.data.block);
}
}
return blocks;
}
IR::Program TranslateProgram(std::span<const u32> code, Pools& pools, Info& info,
RuntimeInfo& runtime_info, const Profile& profile) {
// Ensure first instruction is expected.
constexpr u32 token_mov_vcchi = 0xBEEB03FF;
if (code[0] != token_mov_vcchi) {
LOG_WARNING(Render_Recompiler, "First instruction is not s_mov_b32 vcc_hi, #imm");
}
Gcn::GcnCodeSlice slice(code.data(), code.data() + code.size());
Gcn::GcnDecodeContext decoder;
// Decode and save instructions
IR::Program program{info};
program.ins_list.reserve(code.size());
while (!slice.atEnd()) {
program.ins_list.emplace_back(decoder.decodeInstruction(slice));
}
// Clear any previous pooled data.
pools.ReleaseContents();
// Create control flow graph
Common::ObjectPool<Gcn::Block> gcn_block_pool{64};
Gcn::CFG cfg{gcn_block_pool, program.ins_list};
// Structurize control flow graph and create program.
program.syntax_list = Shader::Gcn::BuildASL(pools.inst_pool, pools.block_pool, cfg,
program.info, runtime_info, profile);
program.blocks = GenerateBlocks(program.syntax_list);
program.post_order_blocks = Shader::IR::PostOrder(program.syntax_list.front());
// Run optimization passes
if (!profile.support_float64) {
Shader::Optimization::LowerFp64ToFp32(program);
}
Shader::Optimization::SsaRewritePass(program.post_order_blocks);
Shader::Optimization::ConstantPropagationPass(program.post_order_blocks);
Shader::Optimization::IdentityRemovalPass(program.blocks);
if (info.l_stage == LogicalStage::TessellationControl) {
Shader::Optimization::TessellationPreprocess(program, runtime_info);
Shader::Optimization::HullShaderTransform(program, runtime_info);
} else if (info.l_stage == LogicalStage::TessellationEval) {
Shader::Optimization::TessellationPreprocess(program, runtime_info);
Shader::Optimization::DomainShaderTransform(program, runtime_info);
}
Shader::Optimization::RingAccessElimination(program, runtime_info);
Shader::Optimization::ReadLaneEliminationPass(program);
Shader::Optimization::FlattenExtendedUserdataPass(program);
Shader::Optimization::ResourceTrackingPass(program);
Shader::Optimization::LowerBufferFormatToRaw(program);
Shader::Optimization::SharedMemoryToStoragePass(program, runtime_info, profile);
Shader::Optimization::SharedMemoryBarrierPass(program, runtime_info, profile);
Shader::Optimization::IdentityRemovalPass(program.blocks);
Shader::Optimization::DeadCodeEliminationPass(program);
Shader::Optimization::ConstantPropagationPass(program.post_order_blocks);
Shader::Optimization::CollectShaderInfoPass(program);
Shader::IR::DumpProgram(program, info);
return program;
}
} // namespace Shader