New shader cache implementation (#3194)
* New shader cache implementation
* Remove some debug code
* Take transform feedback varying count into account
* Create shader cache directory if it does not exist + fragment output map related fixes
* Remove debug code
* Only check texture descriptors if the constant buffer is bound
* Also check CPU VA on GetSpanMapped
* Remove more unused code and move cache related code
* XML docs + remove more unused methods
* Better codegen for TransformFeedbackDescriptor.AsSpan
* Support migration from old cache format, remove more unused code
Shader cache rebuild now also rewrites the shared toc and data files
* Fix migration error with BRX shaders
* Add a limit to the async translation queue
Avoid async translation threads not being able to keep up and the queue growing very large
* Re-create specialization state on recompile
This might be required if a new version of the shader translator requires more or less state, or if there is a bug related to the GPU state access
* Make shader cache more error resilient
* Add some missing XML docs and move GpuAccessor docs to the interface/use inheritdoc
* Address early PR feedback
* Fix rebase
* Remove IRenderer.CompileShader and IShader interface, replace with new ShaderSource struct passed to CreateProgram directly
* Handle some missing exceptions
* Make shader cache purge delete both old and new shader caches
* Register textures on new specialization state
* Translate and compile shaders in forward order (eliminates diffs due to different binding numbers)
* Limit in-flight shader compilation to the maximum number of compilation threads
* Replace ParallelDiskCacheLoader state changed event with a callback function
* Better handling for invalid constant buffer 1 data length
* Do not create the old cache directory structure if the old cache does not exist
* Constant buffer use should be per-stage. This change will invalidate existing new caches (file format version was incremented)
* Replace rectangle texture with just coordinate normalization
* Skip incompatible shaders that are missing texture information, instead of crashing
This is required if we, for example, support new texture instruction to the shader translator, and then they allow access to textures that were not accessed before. In this scenario, the old cache entry is no longer usable
* Fix coordinates normalization on cubemap textures
* Check if title ID is null before combining shader cache path
* More robust constant buffer address validation on spec state
* More robust constant buffer address validation on spec state (2)
* Regenerate shader cache with one stream, rather than one per shader.
* Only create shader cache directory during initialization
* Logging improvements
* Proper shader program disposal
* PR feedback, and add a comment on serialized structs
* XML docs for RegisterTexture
Co-authored-by: riperiperi <rhy3756547@hotmail.com>
2022-04-10 15:49:44 +02:00
|
|
|
using Ryujinx.Common.Logging;
|
|
|
|
using Ryujinx.Graphics.Gpu.Image;
|
|
|
|
using Ryujinx.Graphics.Shader;
|
|
|
|
using System;
|
|
|
|
using System.Runtime.InteropServices;
|
|
|
|
|
|
|
|
namespace Ryujinx.Graphics.Gpu.Shader.DiskCache
|
|
|
|
{
|
|
|
|
/// <summary>
|
|
|
|
/// Represents a GPU state and memory accessor.
|
|
|
|
/// </summary>
|
|
|
|
class DiskCacheGpuAccessor : GpuAccessorBase, IGpuAccessor
|
|
|
|
{
|
|
|
|
private readonly ReadOnlyMemory<byte> _data;
|
|
|
|
private readonly ReadOnlyMemory<byte> _cb1Data;
|
|
|
|
private readonly ShaderSpecializationState _oldSpecState;
|
|
|
|
private readonly ShaderSpecializationState _newSpecState;
|
|
|
|
private readonly int _stageIndex;
|
|
|
|
private ResourceCounts _resourceCounts;
|
|
|
|
|
|
|
|
/// <summary>
|
|
|
|
/// Creates a new instance of the cached GPU state accessor for shader translation.
|
|
|
|
/// </summary>
|
|
|
|
/// <param name="context">GPU context</param>
|
|
|
|
/// <param name="data">The data of the shader</param>
|
|
|
|
/// <param name="cb1Data">The constant buffer 1 data of the shader</param>
|
|
|
|
/// <param name="oldSpecState">Shader specialization state of the cached shader</param>
|
|
|
|
/// <param name="newSpecState">Shader specialization state of the recompiled shader</param>
|
|
|
|
/// <param name="stageIndex">Shader stage index</param>
|
|
|
|
public DiskCacheGpuAccessor(
|
|
|
|
GpuContext context,
|
|
|
|
ReadOnlyMemory<byte> data,
|
|
|
|
ReadOnlyMemory<byte> cb1Data,
|
|
|
|
ShaderSpecializationState oldSpecState,
|
|
|
|
ShaderSpecializationState newSpecState,
|
|
|
|
ResourceCounts counts,
|
|
|
|
int stageIndex) : base(context)
|
|
|
|
{
|
|
|
|
_data = data;
|
|
|
|
_cb1Data = cb1Data;
|
|
|
|
_oldSpecState = oldSpecState;
|
|
|
|
_newSpecState = newSpecState;
|
|
|
|
_stageIndex = stageIndex;
|
|
|
|
_resourceCounts = counts;
|
|
|
|
}
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public uint ConstantBuffer1Read(int offset)
|
|
|
|
{
|
|
|
|
if (offset + sizeof(uint) > _cb1Data.Length)
|
|
|
|
{
|
|
|
|
throw new DiskCacheLoadException(DiskCacheLoadResult.InvalidCb1DataLength);
|
|
|
|
}
|
|
|
|
|
|
|
|
return MemoryMarshal.Cast<byte, uint>(_cb1Data.Span.Slice(offset))[0];
|
|
|
|
}
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public void Log(string message)
|
|
|
|
{
|
|
|
|
Logger.Warning?.Print(LogClass.Gpu, $"Shader translator: {message}");
|
|
|
|
}
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public ReadOnlySpan<ulong> GetCode(ulong address, int minimumSize)
|
|
|
|
{
|
|
|
|
return MemoryMarshal.Cast<byte, ulong>(_data.Span.Slice((int)address));
|
|
|
|
}
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public int QueryBindingConstantBuffer(int index)
|
|
|
|
{
|
|
|
|
return _resourceCounts.UniformBuffersCount++;
|
|
|
|
}
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public int QueryBindingStorageBuffer(int index)
|
|
|
|
{
|
|
|
|
return _resourceCounts.StorageBuffersCount++;
|
|
|
|
}
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public int QueryBindingTexture(int index)
|
|
|
|
{
|
|
|
|
return _resourceCounts.TexturesCount++;
|
|
|
|
}
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public int QueryBindingImage(int index)
|
|
|
|
{
|
|
|
|
return _resourceCounts.ImagesCount++;
|
|
|
|
}
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public int QueryComputeLocalSizeX() => _oldSpecState.ComputeState.LocalSizeX;
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public int QueryComputeLocalSizeY() => _oldSpecState.ComputeState.LocalSizeY;
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public int QueryComputeLocalSizeZ() => _oldSpecState.ComputeState.LocalSizeZ;
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public int QueryComputeLocalMemorySize() => _oldSpecState.ComputeState.LocalMemorySize;
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public int QueryComputeSharedMemorySize() => _oldSpecState.ComputeState.SharedMemorySize;
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public uint QueryConstantBufferUse()
|
|
|
|
{
|
|
|
|
_newSpecState.RecordConstantBufferUse(_stageIndex, _oldSpecState.ConstantBufferUse[_stageIndex]);
|
|
|
|
return _oldSpecState.ConstantBufferUse[_stageIndex];
|
|
|
|
}
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public InputTopology QueryPrimitiveTopology()
|
|
|
|
{
|
|
|
|
_newSpecState.RecordPrimitiveTopology();
|
|
|
|
return ConvertToInputTopology(_oldSpecState.GraphicsState.Topology, _oldSpecState.GraphicsState.TessellationMode);
|
|
|
|
}
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public bool QueryTessCw()
|
|
|
|
{
|
|
|
|
return _oldSpecState.GraphicsState.TessellationMode.UnpackCw();
|
|
|
|
}
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public TessPatchType QueryTessPatchType()
|
|
|
|
{
|
|
|
|
return _oldSpecState.GraphicsState.TessellationMode.UnpackPatchType();
|
|
|
|
}
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public TessSpacing QueryTessSpacing()
|
|
|
|
{
|
|
|
|
return _oldSpecState.GraphicsState.TessellationMode.UnpackSpacing();
|
|
|
|
}
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public TextureFormat QueryTextureFormat(int handle, int cbufSlot)
|
|
|
|
{
|
|
|
|
_newSpecState.RecordTextureFormat(_stageIndex, handle, cbufSlot);
|
|
|
|
(uint format, bool formatSrgb) = _oldSpecState.GetFormat(_stageIndex, handle, cbufSlot);
|
|
|
|
return ConvertToTextureFormat(format, formatSrgb);
|
|
|
|
}
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public SamplerType QuerySamplerType(int handle, int cbufSlot)
|
|
|
|
{
|
|
|
|
_newSpecState.RecordTextureSamplerType(_stageIndex, handle, cbufSlot);
|
|
|
|
return _oldSpecState.GetTextureTarget(_stageIndex, handle, cbufSlot).ConvertSamplerType();
|
|
|
|
}
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public bool QueryTextureCoordNormalized(int handle, int cbufSlot)
|
|
|
|
{
|
|
|
|
_newSpecState.RecordTextureCoordNormalized(_stageIndex, handle, cbufSlot);
|
|
|
|
return _oldSpecState.GetCoordNormalized(_stageIndex, handle, cbufSlot);
|
|
|
|
}
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public bool QueryTransformFeedbackEnabled()
|
|
|
|
{
|
|
|
|
return _oldSpecState.TransformFeedbackDescriptors != null;
|
|
|
|
}
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public ReadOnlySpan<byte> QueryTransformFeedbackVaryingLocations(int bufferIndex)
|
|
|
|
{
|
|
|
|
return _oldSpecState.TransformFeedbackDescriptors[bufferIndex].AsSpan();
|
|
|
|
}
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public int QueryTransformFeedbackStride(int bufferIndex)
|
|
|
|
{
|
|
|
|
return _oldSpecState.TransformFeedbackDescriptors[bufferIndex].Stride;
|
|
|
|
}
|
|
|
|
|
|
|
|
/// <inheritdoc/>
|
|
|
|
public bool QueryEarlyZForce()
|
|
|
|
{
|
|
|
|
_newSpecState.RecordEarlyZForce();
|
|
|
|
return _oldSpecState.GraphicsState.EarlyZForce;
|
|
|
|
}
|
|
|
|
|
2022-05-12 15:47:13 +02:00
|
|
|
/// <inheritdoc/>
|
|
|
|
public bool QueryViewportTransformDisable()
|
|
|
|
{
|
|
|
|
return _oldSpecState.GraphicsState.ViewportTransformDisable;
|
|
|
|
}
|
|
|
|
|
New shader cache implementation (#3194)
* New shader cache implementation
* Remove some debug code
* Take transform feedback varying count into account
* Create shader cache directory if it does not exist + fragment output map related fixes
* Remove debug code
* Only check texture descriptors if the constant buffer is bound
* Also check CPU VA on GetSpanMapped
* Remove more unused code and move cache related code
* XML docs + remove more unused methods
* Better codegen for TransformFeedbackDescriptor.AsSpan
* Support migration from old cache format, remove more unused code
Shader cache rebuild now also rewrites the shared toc and data files
* Fix migration error with BRX shaders
* Add a limit to the async translation queue
Avoid async translation threads not being able to keep up and the queue growing very large
* Re-create specialization state on recompile
This might be required if a new version of the shader translator requires more or less state, or if there is a bug related to the GPU state access
* Make shader cache more error resilient
* Add some missing XML docs and move GpuAccessor docs to the interface/use inheritdoc
* Address early PR feedback
* Fix rebase
* Remove IRenderer.CompileShader and IShader interface, replace with new ShaderSource struct passed to CreateProgram directly
* Handle some missing exceptions
* Make shader cache purge delete both old and new shader caches
* Register textures on new specialization state
* Translate and compile shaders in forward order (eliminates diffs due to different binding numbers)
* Limit in-flight shader compilation to the maximum number of compilation threads
* Replace ParallelDiskCacheLoader state changed event with a callback function
* Better handling for invalid constant buffer 1 data length
* Do not create the old cache directory structure if the old cache does not exist
* Constant buffer use should be per-stage. This change will invalidate existing new caches (file format version was incremented)
* Replace rectangle texture with just coordinate normalization
* Skip incompatible shaders that are missing texture information, instead of crashing
This is required if we, for example, support new texture instruction to the shader translator, and then they allow access to textures that were not accessed before. In this scenario, the old cache entry is no longer usable
* Fix coordinates normalization on cubemap textures
* Check if title ID is null before combining shader cache path
* More robust constant buffer address validation on spec state
* More robust constant buffer address validation on spec state (2)
* Regenerate shader cache with one stream, rather than one per shader.
* Only create shader cache directory during initialization
* Logging improvements
* Proper shader program disposal
* PR feedback, and add a comment on serialized structs
* XML docs for RegisterTexture
Co-authored-by: riperiperi <rhy3756547@hotmail.com>
2022-04-10 15:49:44 +02:00
|
|
|
/// <inheritdoc/>
|
|
|
|
public void RegisterTexture(int handle, int cbufSlot)
|
|
|
|
{
|
|
|
|
if (!_oldSpecState.TextureRegistered(_stageIndex, handle, cbufSlot))
|
|
|
|
{
|
|
|
|
throw new DiskCacheLoadException(DiskCacheLoadResult.MissingTextureDescriptor);
|
|
|
|
}
|
|
|
|
|
|
|
|
(uint format, bool formatSrgb) = _oldSpecState.GetFormat(_stageIndex, handle, cbufSlot);
|
|
|
|
TextureTarget target = _oldSpecState.GetTextureTarget(_stageIndex, handle, cbufSlot);
|
|
|
|
bool coordNormalized = _oldSpecState.GetCoordNormalized(_stageIndex, handle, cbufSlot);
|
|
|
|
_newSpecState.RegisterTexture(_stageIndex, handle, cbufSlot, format, formatSrgb, target, coordNormalized);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|