diff options
| author | gdkchan <gab.dark.100@gmail.com> | 2022-04-10 10:49:44 -0300 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2022-04-10 10:49:44 -0300 |
| commit | 43ebd7a9bbba0c1290a9e98b9224f0752627c400 (patch) | |
| tree | 02e1006242113be10aa0c6312de51120eeff91a7 /Ryujinx.Graphics.Gpu/Shader/GpuAccessor.cs | |
| parent | 26a881176eb6513a98889648e0d5b7fe647cd0e3 (diff) | |
New shader cache implementation (#3194)1.1.101
* 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>
Diffstat (limited to 'Ryujinx.Graphics.Gpu/Shader/GpuAccessor.cs')
| -rw-r--r-- | Ryujinx.Graphics.Gpu/Shader/GpuAccessor.cs | 249 |
1 files changed, 106 insertions, 143 deletions
diff --git a/Ryujinx.Graphics.Gpu/Shader/GpuAccessor.cs b/Ryujinx.Graphics.Gpu/Shader/GpuAccessor.cs index a5c7575f..192467b7 100644 --- a/Ryujinx.Graphics.Gpu/Shader/GpuAccessor.cs +++ b/Ryujinx.Graphics.Gpu/Shader/GpuAccessor.cs @@ -1,5 +1,5 @@ using Ryujinx.Common.Logging; -using Ryujinx.Graphics.GAL; +using Ryujinx.Graphics.Gpu.Image; using Ryujinx.Graphics.Shader; using System; using System.Runtime.InteropServices; @@ -9,19 +9,12 @@ namespace Ryujinx.Graphics.Gpu.Shader /// <summary> /// Represents a GPU state and memory accessor. /// </summary> - class GpuAccessor : TextureDescriptorCapableGpuAccessor, IGpuAccessor + class GpuAccessor : GpuAccessorBase, IGpuAccessor { private readonly GpuChannel _channel; private readonly GpuAccessorState _state; private readonly int _stageIndex; private readonly bool _compute; - private readonly int _localSizeX; - private readonly int _localSizeY; - private readonly int _localSizeZ; - private readonly int _localMemorySize; - private readonly int _sharedMemorySize; - - public int Cb1DataSize { get; private set; } /// <summary> /// Creates a new instance of the GPU state accessor for graphics shader translation. @@ -43,43 +36,16 @@ namespace Ryujinx.Graphics.Gpu.Shader /// <param name="context">GPU context</param> /// <param name="channel">GPU channel</param> /// <param name="state">Current GPU state</param> - /// <param name="localSizeX">Local group size X of the compute shader</param> - /// <param name="localSizeY">Local group size Y of the compute shader</param> - /// <param name="localSizeZ">Local group size Z of the compute shader</param> - /// <param name="localMemorySize">Local memory size of the compute shader</param> - /// <param name="sharedMemorySize">Shared memory size of the compute shader</param> - public GpuAccessor( - GpuContext context, - GpuChannel channel, - GpuAccessorState state, - int localSizeX, - int localSizeY, - int localSizeZ, - int localMemorySize, - int sharedMemorySize) : base(context) + public GpuAccessor(GpuContext context, GpuChannel channel, GpuAccessorState state) : base(context) { _channel = channel; _state = state; _compute = true; - _localSizeX = localSizeX; - _localSizeY = localSizeY; - _localSizeZ = localSizeZ; - _localMemorySize = localMemorySize; - _sharedMemorySize = sharedMemorySize; } - /// <summary> - /// Reads data from the constant buffer 1. - /// </summary> - /// <param name="offset">Offset in bytes to read from</param> - /// <returns>Value at the given offset</returns> + /// <inheritdoc/> public uint ConstantBuffer1Read(int offset) { - if (Cb1DataSize < offset + 4) - { - Cb1DataSize = offset + 4; - } - ulong baseAddress = _compute ? _channel.BufferManager.GetComputeUniformBufferAddress(1) : _channel.BufferManager.GetGraphicsUniformBufferAddress(_stageIndex, 1); @@ -87,111 +53,115 @@ namespace Ryujinx.Graphics.Gpu.Shader return _channel.MemoryManager.Physical.Read<uint>(baseAddress + (ulong)offset); } - /// <summary> - /// Prints a log message. - /// </summary> - /// <param name="message">Message to print</param> + /// <inheritdoc/> public void Log(string message) { Logger.Warning?.Print(LogClass.Gpu, $"Shader translator: {message}"); } - /// <summary> - /// Gets a span of the specified memory location, containing shader code. - /// </summary> - /// <param name="address">GPU virtual address of the data</param> - /// <param name="minimumSize">Minimum size that the returned span may have</param> - /// <returns>Span of the memory location</returns> - public override ReadOnlySpan<ulong> GetCode(ulong address, int minimumSize) + /// <inheritdoc/> + public ReadOnlySpan<ulong> GetCode(ulong address, int minimumSize) { int size = Math.Max(minimumSize, 0x1000 - (int)(address & 0xfff)); return MemoryMarshal.Cast<byte, ulong>(_channel.MemoryManager.GetSpan(address, size)); } - /// <summary> - /// Queries Local Size X for compute shaders. - /// </summary> - /// <returns>Local Size X</returns> - public int QueryComputeLocalSizeX() => _localSizeX; + /// <inheritdoc/> + public int QueryBindingConstantBuffer(int index) + { + return _state.ResourceCounts.UniformBuffersCount++; + } - /// <summary> - /// Queries Local Size Y for compute shaders. - /// </summary> - /// <returns>Local Size Y</returns> - public int QueryComputeLocalSizeY() => _localSizeY; + /// <inheritdoc/> + public int QueryBindingStorageBuffer(int index) + { + return _state.ResourceCounts.StorageBuffersCount++; + } - /// <summary> - /// Queries Local Size Z for compute shaders. - /// </summary> - /// <returns>Local Size Z</returns> - public int QueryComputeLocalSizeZ() => _localSizeZ; + /// <inheritdoc/> + public int QueryBindingTexture(int index) + { + return _state.ResourceCounts.TexturesCount++; + } - /// <summary> - /// Queries Local Memory size in bytes for compute shaders. - /// </summary> - /// <returns>Local Memory size in bytes</returns> - public int QueryComputeLocalMemorySize() => _localMemorySize; + /// <inheritdoc/> + public int QueryBindingImage(int index) + { + return _state.ResourceCounts.ImagesCount++; + } - /// <summary> - /// Queries Shared Memory size in bytes for compute shaders. - /// </summary> - /// <returns>Shared Memory size in bytes</returns> - public int QueryComputeSharedMemorySize() => _sharedMemorySize; + /// <inheritdoc/> + public int QueryComputeLocalSizeX() => _state.ComputeState.LocalSizeX; - /// <summary> - /// Queries Constant Buffer usage information. - /// </summary> - /// <returns>A mask where each bit set indicates a bound constant buffer</returns> + /// <inheritdoc/> + public int QueryComputeLocalSizeY() => _state.ComputeState.LocalSizeY; + + /// <inheritdoc/> + public int QueryComputeLocalSizeZ() => _state.ComputeState.LocalSizeZ; + + /// <inheritdoc/> + public int QueryComputeLocalMemorySize() => _state.ComputeState.LocalMemorySize; + + /// <inheritdoc/> + public int QueryComputeSharedMemorySize() => _state.ComputeState.SharedMemorySize; + + /// <inheritdoc/> public uint QueryConstantBufferUse() { - return _compute + uint useMask = _compute ? _channel.BufferManager.GetComputeUniformBufferUseMask() : _channel.BufferManager.GetGraphicsUniformBufferUseMask(_stageIndex); + + _state.SpecializationState?.RecordConstantBufferUse(_stageIndex, useMask); + return useMask; } - /// <summary> - /// Queries current primitive topology for geometry shaders. - /// </summary> - /// <returns>Current primitive topology</returns> + /// <inheritdoc/> public InputTopology QueryPrimitiveTopology() { - return _state.Topology switch - { - PrimitiveTopology.Points => InputTopology.Points, - PrimitiveTopology.Lines or - PrimitiveTopology.LineLoop or - PrimitiveTopology.LineStrip => InputTopology.Lines, - PrimitiveTopology.LinesAdjacency or - PrimitiveTopology.LineStripAdjacency => InputTopology.LinesAdjacency, - PrimitiveTopology.Triangles or - PrimitiveTopology.TriangleStrip or - PrimitiveTopology.TriangleFan => InputTopology.Triangles, - PrimitiveTopology.TrianglesAdjacency or - PrimitiveTopology.TriangleStripAdjacency => InputTopology.TrianglesAdjacency, - PrimitiveTopology.Patches => _state.TessellationMode.UnpackPatchType() == TessPatchType.Isolines - ? InputTopology.Lines - : InputTopology.Triangles, - _ => InputTopology.Points - }; + _state.SpecializationState?.RecordPrimitiveTopology(); + return ConvertToInputTopology(_state.GraphicsState.Topology, _state.GraphicsState.TessellationMode); } - /// <summary> - /// Queries the tessellation evaluation shader primitive winding order. - /// </summary> - /// <returns>True if the primitive winding order is clockwise, false if counter-clockwise</returns> - public bool QueryTessCw() => _state.TessellationMode.UnpackCw(); + /// <inheritdoc/> + public bool QueryTessCw() + { + return _state.GraphicsState.TessellationMode.UnpackCw(); + } - /// <summary> - /// Queries the tessellation evaluation shader abstract patch type. - /// </summary> - /// <returns>Abstract patch type</returns> - public TessPatchType QueryTessPatchType() => _state.TessellationMode.UnpackPatchType(); + /// <inheritdoc/> + public TessPatchType QueryTessPatchType() + { + return _state.GraphicsState.TessellationMode.UnpackPatchType(); + } - /// <summary> - /// Queries the tessellation evaluation shader spacing between tessellated vertices of the patch. - /// </summary> - /// <returns>Spacing between tessellated vertices of the patch</returns> - public TessSpacing QueryTessSpacing() => _state.TessellationMode.UnpackSpacing(); + /// <inheritdoc/> + public TessSpacing QueryTessSpacing() + { + return _state.GraphicsState.TessellationMode.UnpackSpacing(); + } + + //// <inheritdoc/> + public TextureFormat QueryTextureFormat(int handle, int cbufSlot) + { + _state.SpecializationState?.RecordTextureFormat(_stageIndex, handle, cbufSlot); + var descriptor = GetTextureDescriptor(handle, cbufSlot); + return ConvertToTextureFormat(descriptor.UnpackFormat(), descriptor.UnpackSrgb()); + } + + /// <inheritdoc/> + public SamplerType QuerySamplerType(int handle, int cbufSlot) + { + _state.SpecializationState?.RecordTextureSamplerType(_stageIndex, handle, cbufSlot); + return GetTextureDescriptor(handle, cbufSlot).UnpackTextureTarget().ConvertSamplerType(); + } + + /// <inheritdoc/> + public bool QueryTextureCoordNormalized(int handle, int cbufSlot) + { + _state.SpecializationState?.RecordTextureCoordNormalized(_stageIndex, handle, cbufSlot); + return GetTextureDescriptor(handle, cbufSlot).UnpackTextureCoordNormalized(); + } /// <summary> /// Gets the texture descriptor for a given texture on the pool. @@ -199,65 +169,58 @@ namespace Ryujinx.Graphics.Gpu.Shader /// <param name="handle">Index of the texture (this is the word offset of the handle in the constant buffer)</param> /// <param name="cbufSlot">Constant buffer slot for the texture handle</param> /// <returns>Texture descriptor</returns> - public override Image.ITextureDescriptor GetTextureDescriptor(int handle, int cbufSlot) + private Image.TextureDescriptor GetTextureDescriptor(int handle, int cbufSlot) { if (_compute) { return _channel.TextureManager.GetComputeTextureDescriptor( - _state.TexturePoolGpuVa, - _state.TextureBufferIndex, - _state.TexturePoolMaximumId, + _state.PoolState.TexturePoolGpuVa, + _state.PoolState.TextureBufferIndex, + _state.PoolState.TexturePoolMaximumId, handle, cbufSlot); } else { return _channel.TextureManager.GetGraphicsTextureDescriptor( - _state.TexturePoolGpuVa, - _state.TextureBufferIndex, - _state.TexturePoolMaximumId, + _state.PoolState.TexturePoolGpuVa, + _state.PoolState.TextureBufferIndex, + _state.PoolState.TexturePoolMaximumId, _stageIndex, handle, cbufSlot); } } - /// <summary> - /// Queries transform feedback enable state. - /// </summary> - /// <returns>True if the shader uses transform feedback, false otherwise</returns> + /// <inheritdoc/> public bool QueryTransformFeedbackEnabled() { return _state.TransformFeedbackDescriptors != null; } - /// <summary> - /// Queries the varying locations that should be written to the transform feedback buffer. - /// </summary> - /// <param name="bufferIndex">Index of the transform feedback buffer</param> - /// <returns>Varying locations for the specified buffer</returns> + /// <inheritdoc/> public ReadOnlySpan<byte> QueryTransformFeedbackVaryingLocations(int bufferIndex) { - return _state.TransformFeedbackDescriptors[bufferIndex].VaryingLocations; + return _state.TransformFeedbackDescriptors[bufferIndex].AsSpan(); } - /// <summary> - /// Queries the stride (in bytes) of the per vertex data written into the transform feedback buffer. - /// </summary> - /// <param name="bufferIndex">Index of the transform feedback buffer</param> - /// <returns>Stride for the specified buffer</returns> + /// <inheritdoc/> public int QueryTransformFeedbackStride(int bufferIndex) { return _state.TransformFeedbackDescriptors[bufferIndex].Stride; } - /// <summary> - /// Queries if host state forces early depth testing. - /// </summary> - /// <returns>True if early depth testing is forced</returns> + /// <inheritdoc/> public bool QueryEarlyZForce() { - return _state.EarlyZForce; + _state.SpecializationState?.RecordEarlyZForce(); + return _state.GraphicsState.EarlyZForce; + } + + /// <inheritdoc/> + public void RegisterTexture(int handle, int cbufSlot) + { + _state.SpecializationState?.RegisterTexture(_stageIndex, handle, cbufSlot, GetTextureDescriptor(handle, cbufSlot)); } } } |