Emulator Issues #13603
open
Enabling Manual Texture Sampling alongside Ubershaders and the D3D11 or D3D12 backend causes Dolphin to lock up
0%
Description
Game Name?
All Wii games, all GameCube games if using the GameCube IPL and Skip Main Menu is not checked
What's the problem? Describe what went wrong.
Enabling Manual Texture Sampling alongside Ubershaders and the D3D11 or D3D12 backend causes Dolphin to lock up when launching any Wii game and any GameCube game, if using the GameCube IPL and Skip Main Menu is not checked.
The Vulkan backend is not affected, and checking Skip Main Menu allows GameCube games to work with Manual Texture Sampling alongside Ubershaders and the D3D11 or D3D12 backend.
What steps will reproduce the problem?
- Use D3D11 or D3D12 backend, select Hybrid or Exclusive Ubershaders, enable Manual Texture Sampling, launch Wii game
- Use D3D11 or D3D12 backend, select Hybrid or Exclusive Ubershaders, enable Manual Texture Sampling, add a GameCube IPL, launch a GameCube game (like Fifa Street 2 which needs Manual Texture Sampling)
Is the issue present in the latest development version? For future reference, please also write down the version number of the latest development version.
Yes (2407-264)
Is the issue present in the latest release? For future reference, please also write down the version number of the latest release.
Yes (2407)
What are your PC specifications? (CPU, GPU, Operating System, more)
Intel i7 11800H, RTX 3060 Laptop GPU, 16 GB RAM, Windows 11
Game Name?
[Put Game Name here]
Game ID? (right click the game in the game list, Properties, Info tab)
[Put Game ID here]
MD5 Hash? (right click the game in the game list, Properties, Verify tab, Verify Integrity button)
[Put MD5 Hash here]
What's the problem? Describe what went wrong.
[Description here]
What steps will reproduce the problem?
[Reproduction steps here]
Is the issue present in the latest development version? For future reference, please also write down the version number of the latest development version.
[Yes/No and version number here]
Is the issue present in the latest release? For future reference, please also write down the version number of the latest release.
[Yes/No and version number here]
If the issue isn't present in the latest release, which is the first broken version? (You can find the first broken version by bisecting. Windows users can use the tool https://forums.dolphin-emu.org/Thread-green-notice-development-thread-unofficial-dolphin-bisection-tool-for-finding-broken-builds and anyone who is building Dolphin on their own can use git bisect.)
[First broken version number here (if applicable)]
If your issue is a graphical issue, please attach screenshots and record a three frame fifolog of the issue if possible. Screenshots showing what it is supposed to look like from either console or older builds of Dolphin will help too. For more information on how to use the fifoplayer, please check here: https://wiki.dolphin-emu.org/index.php?title=FifoPlayer
[Attach any fifologs if possible, write a description of fifologs and screenshots here to assist people unfamiliar with the game.]
What are your PC specifications? (CPU, GPU, Operating System, more)
[PC specs here]
Is there anything else that can help developers narrow down the issue? (e.g. logs, screenshots,
configuration files, savefiles, savestates)
[Anything else here]
Files
Updated by Dentomologist 21 days ago
I believe the root cause here is some sort of exponential explosion in D3D's shader optimization. Compilation will eventually finish, but glacially enough that it looks like a hang. Even trying to close the game will take a while since AsyncShaderCompiler joins all its worker threads when stopping, which requires them all to finish their current work unit. Turn on "Compile Shaders Before Starting" to see this more easily (though the progress bar doesn't render when stopping the game).
Here are a few timing examples to help narrow things down. Before each run I cleared the shader cache but not the uidcache.
Settings are D3D11, Single Core, Exclusive Ubershaders and Manual Texture Sampling unless otherwise noted:
No changed settings: 2m05s
No Manual Texture Sampling: 6s
Manual Texture Sampling enabled but the main implementation stubbed: 6s
WrapCoord stubbed: 30s
D3DCOMPILE_OPTIMIZATION_LEVEL3 changed to D3DCOMPILE_OPTIMIZATION_LEVEL0: 2m05s (no benefit)
D3DCOMPILE_OPTIMIZATION_LEVEL3 changed to D3DCOMPILE_SKIP_OPTIMIZATION: 30s
D3DCOMPILE_OPTIMIZATION_LEVEL3 changed to D3DCOMPILE_SKIP_OPTIMIZATION and WrapCoord stubbed: 10s
Specialized (Default): ~01s
Hybrid Ubershaders: 2m05s
Specialized shaders do compile the Manual Texture Sampling block in question, so I don't know why they're so much faster than the other shader types.
Unfortunately Visual Studio's profile won't show me anything deeper than the call to D3DCompile (which takes 90-98% of the total time) despite having the symbols, but I was able to use Parallel Stacks to get a bunch of call stacks resembling this:
CCompiler::ResetNestedSimulation(unsigned int,unsigned int,class CArgument * const *,class CTrackVarMapChanges *,class CSaveSimulationState *)
CCompiler::EmitSwitchInst(class CSwitchInfo *)
CCompiler::EmitSwitch(class CNode const *,class CNode const *,class CNode const *,class CArgument *)
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitFunction()
CCompiler::EmitCallExpr()
CCompiler::EmitExpression()
CCompiler::EmitExpression()
CCompiler::EmitExpression()
CCompiler::EmitExpression()
CCompiler::EmitAndMarkExpression(struct ArSourceLocation const *,class CNode const *)
CCompiler::EmitVariableInit(class ArSymbolVariable *)
CCompiler::EmitSymbol()
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitIf_EmitStatement(bool,bool,bool,class CArgument *,class CArgument * * const,class CNode const *,struct ArSourceLocation const *,struct ArSourceLocation const *,class CNode const *,struct SEmitIfSwitchData *,bool *,struct ArSourceLocation const * *,bool *,bool *,bool *)
CCompiler::EmitIf()
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitIf_EmitStatement(bool,bool,bool,class CArgument *,class CArgument * * const,class CNode const *,struct ArSourceLocation const *,struct ArSourceLocation const *,class CNode const *,struct SEmitIfSwitchData *,bool *,struct ArSourceLocation const * *,bool *,bool *,bool *)
CCompiler::EmitIf()
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitFunction()
CCompiler::EmitCallExpr()
CCompiler::EmitExpression()
CCompiler::EmitExpression()
CCompiler::EmitExpression()
CCompiler::EmitAndMarkExpression(struct ArSourceLocation const *,class CNode const *)
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitSwitchInst(class CSwitchInfo *)
CCompiler::EmitSwitch(class CNode const *,class CNode const *,class CNode const *,class CArgument *)
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitFunction()
CCompiler::EmitCallExpr()
CCompiler::EmitExpression()
CCompiler::EmitExpression()
CCompiler::EmitExpression()
CCompiler::EmitExpression()
CCompiler::EmitExpression()
CCompiler::EmitAndMarkExpression(struct ArSourceLocation const *,class CNode const *)
CCompiler::EmitVariableInit(class ArSymbolVariable *)
CCompiler::EmitSymbol()
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitIf_EmitStatement(bool,bool,bool,class CArgument *,class CArgument * * const,class CNode const *,struct ArSourceLocation const *,struct ArSourceLocation const *,class CNode const *,struct SEmitIfSwitchData *,bool *,struct ArSourceLocation const * *,bool *,bool *,bool *)
CCompiler::EmitIf()
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitIf_EmitStatement(bool,bool,bool,class CArgument *,class CArgument * * const,class CNode const *,struct ArSourceLocation const *,struct ArSourceLocation const *,class CNode const *,struct SEmitIfSwitchData *,bool *,struct ArSourceLocation const * *,bool *,bool *,bool *)
CCompiler::EmitIf()
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitForLoop(class CNode const *,class CNode const *,class CNode const *,class CNode const *,class CNode const *,class CArgument * *,bool,bool,bool)
CCompiler::EmitFor(class CNode const *,class CNode const *,class CNode const *,class CNode const *,class CNode const *)
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitFunction()
CCompiler::EmitCallExpr()
CCompiler::EmitExpression()
CCompiler::EmitAndMarkExpression(struct ArSourceLocation const *,class CNode const *)
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitStatement()
CCompiler::EmitFunction()
CCompiler::EmitAndGenerateCode()
CCompiler::CompileInternal()
D3DCompileShader()
D3DCompileImpl()
D3DCompileSaveReplay()
D3DCompile2()
D3DCompile()
D3DCommon::Shader::CompileShader(D3D_FEATURE_LEVEL, ShaderStage, std::basic_string_view<char,std::char_traits>)
DX11::Gfx::CreateShaderFromSource(ShaderStage, std::basic_string_view<char,std::char_traits>, std::basic_string_view<char,std::char_traits>)
VideoCommon::ShaderCache::CompilePixelUberShader(const ShaderUidUberShader::pixel_ubershader_uid_data &)
VideoCommon::ShaderCache::QueuePixelUberShaderCompile'::2'::PixelUberShaderWorkItem::Compile()
VideoCommon::AsyncShaderCompiler::WorkerThreadRun()
VideoCommon::AsyncShaderCompiler::WorkerThreadEntryPoint(void )
std::invoke<void (__cdecl VideoCommon::AsyncShaderCompiler::)(void *),VideoCommon::AsyncShaderCompiler *,void >(void(VideoCommon::AsyncShaderCompiler::)(void ) &&, VideoCommon::AsyncShaderCompiler * &&, void * &&)
std::thread::_Invoke<std::tuple<void (__cdecl VideoCommon::AsyncShaderCompiler::)(void *),VideoCommon::AsyncShaderCompiler *,void *>,0,1,2>(void *)
00007ffa4d213010()
00007ffb4c71e8d7()
00007ffb4d9bbf6c()