Mirrors/shadPS4
1
0
Fork 0
mirror of https://github.com/shadps4-emu/shadPS4.git synced 2025-07-12 20:55:56 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 2

shader_recompiler: Improve divergence handling and readlane elimintation (#2667)

Browse source 
* control_flow_graph: Improve divergence handling

* recompiler: Simplify optimization passes

Removes a redudant constant propagation and cleans up the passes a little

* ir_passes: Add new readlane elimination pass

The algorithm has grown complex enough where it deserves its own pass. The old implementation could only handle a single phi level properly,
however this one should be able to eliminate vast majority of lane cases remaining. It first performs a traversal of the phi tree to ensure
that all phi sources can be rewritten into an expected value and then performs elimintation by recursively duplicating the phi nodes at each step,
in order to preserve control flow.

* clang format

* control_flow_graph: Remove debug code
This commit is contained in:
TheTurtle 2025-03-23 00:35:42 +02:00 • committed by GitHub
parent a80c4a7f48
commit 1f9ac53c28
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
9 changed files with 211 additions and 139 deletions
Download patch file Download diff file Expand all files Collapse all files

17
src/shader_recompiler/recompiler.cpp
View file

@ -1,9 +1,6 @@
// SPDX-FileCopyrightText: Copyright 2024 shadPS4 Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#include "common/config.h"
#include "common/io_file.h"
#include "common/path_util.h"
#include "shader_recompiler/frontend/control_flow_graph.h"
#include "shader_recompiler/frontend/decode.h"
#include "shader_recompiler/frontend/structured_control_flow.h"
@ -63,26 +60,18 @@ IR::Program TranslateProgram(std::span<const u32> code, Pools& pools, Info& info
program.post_order_blocks = Shader::IR::PostOrder(program.syntax_list.front());
// Run optimization passes
const auto stage = program.info.stage;
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) {
// Tess passes require previous const prop passes for now (for simplicity). TODO allow
// fine grained folding or opportunistic folding we set an operand to an immediate
Shader::Optimization::ConstantPropagationPass(program.post_order_blocks);
Shader::Optimization::TessellationPreprocess(program, runtime_info);
Shader::Optimization::ConstantPropagationPass(program.post_order_blocks);
Shader::Optimization::HullShaderTransform(program, runtime_info);
} else if (info.l_stage == LogicalStage::TessellationEval) {
Shader::Optimization::ConstantPropagationPass(program.post_order_blocks);
Shader::Optimization::TessellationPreprocess(program, runtime_info);
Shader::Optimization::ConstantPropagationPass(program.post_order_blocks);
Shader::Optimization::DomainShaderTransform(program, runtime_info);
}
Shader::Optimization::ConstantPropagationPass(program.post_order_blocks);
Shader::Optimization::RingAccessElimination(program, runtime_info, stage);
Shader::Optimization::ConstantPropagationPass(program.post_order_blocks);
Shader::Optimization::RingAccessElimination(program, runtime_info);
Shader::Optimization::ReadLaneEliminationPass(program);
Shader::Optimization::FlattenExtendedUserdataPass(program);
Shader::Optimization::ResourceTrackingPass(program);
Shader::Optimization::LowerBufferFormatToRaw(program);