From 92c1726647964da9230bc4f4c631a233cd064665 Mon Sep 17 00:00:00 2001 From: BaronKiko Date: Sat, 1 Jun 2019 01:13:57 +0100 Subject: [PATCH] GPU Profiling (#570) * Profiler initial setup * Capture actual timing data * Profiling data dumped to file on close * Support for multiple sessions under the same name * Service profiling * Sort output for easier read * csv output * Split session into 2 seperate values * Refactor name to category * Basic profiling window dummy. Toggle with F1 or set key with config No actual data displayed yet, just a pretty triangle * Simple font rendering * Display some actual timing data * Fix font bearing being ignored * x bearing and advance. Fixed y bearing calc * Different coloured lines to make reading easier * Scrolling * Multiple columns for name * Column titles * display in ms rather than ticks * Bars to display times * Sortable columns * Regex filtering * Better instant timing calculation Fixed minor regex bug * Better filtering Better max value calculation Skip some rendering to reduce profiler weight * Variable update rate * Show/hide inactive button Some other touchups * Add missing project reference * Hide inactive and pause * Fix viewport errors * Update initial window position * Variable name cleanup * Disable timing dump by default * Internal Profile refactor and cleanup * Timing info cleanup * Profile config cleanup * Settings cleanup * Button refactor * Profile refactor * Profile window cleanup * Window manager refactor * Font service cleanup * Fixed bug in profiling method where method was called twice without profiling enabled * Allow update rates of less than 1hz * Stop using window.run because it's apparently not great for performance. Some other performance things, should only draw a new frame when something has changed * Improved time tracking to keep history * Profile window was getting too long so I added regions and split bar rendering out into partial class * Dummy graph view with button to toggle * Realtime graphing initial commit * Display totals on new bar * Simple zooming support with arrow keys * Limit graph zoom and label start and stop * Added support for timing flags * Stop data running away when paused and frame updated * Manual step button * Update at when flag issued (ie every frame) * Removed useless finish profiling call * Enable and disable profiling at compile time. * Better plage for frame swap flag, also kept enough flags to cover larger time spans * No more stopwatches created, uses PerformanceCounter now * public and internal fields to props * Move visible update to update rather than draw as it causes a lockup if called from draw Also added profile window disposal so closing main window closes profiler too * Fixed optimization settings for profiled builds * Appveyer script guess to add profiling builds * Quotes * 1 less quote * Maybe escape space? * Specify config * Different approach * Fix file paths * Fix another path * Better artifact naming * Missing - * test string * Removed for, to test * readd for * moved dashes around so artifacts can begin with letters * quote env vars * martix * Removed configs * Much more efficient capture, ConcurrentDictionary was causing too much overhead * Skip repeating pixels during draw * Stop ram usage getting too high. Compensating for cleanup doing more now * Profile CPU, execute skipped because it's just too much work * Fixed bug with skipping draws. Furthest needed to be reset every loop * Less distracting colour for timing flags * Removed profile method function. It just doesn't play nice with conditional compilation so best to remove it now before it's used a lot * Profile vertex end * Null check for category, group and item * Forgot to reset instant count/time * Increment line when blank * Fix threading conflict Fixed instant count and time. Now accuratly represents the total time and count in the buffer * Fixed bug in time rendering where times were being trimmed to an int. Also added microsecond/millisecond formatting to reduce the number of decimal places needed * Profiled methods used by VertexEndGl * Texture 2d copy profiled * Support for multiple profiling levels * Sometimes it would have to wait a long time for lock to clear so moved it to a tryenter and skip if already locked * Dumb bug regarding clearing of timestamps. Start is already removed so no need to add it to the start * Optimisations in drawing routine: Only calculate bar top and bottom once per bar rather than once per timestamp Pre-calculate the right side of the graph as it was being calculated multiple times per bar Skip rendering timestamps that occupy the same pixel space now uses the raw timestamp to decide. While technically not as accurate it's much easier as the right side of the bar doesn't have to be calculated for a skipped timestamp * Couple alignment changes * Engine 3d call method profiled * Custom equals overload for profile config. The default implpmentation was just too slow * Bump cleanup thread priority. It clears the timer queue so it need to be run frequently * Fixed bug with scrolling caused by recent rendering optimisations. Simply forgot to increment the line index on a skipped line * Stopped blocking memory disposal so much. Also parralised(?) cleanup call * Profiled M2mf, P2mf and shader compilation * Missing file * Removed deprecated config * Removed legacy npaddebug * Using statement no longer needed * Apply suggestions from AcK Co-Authored-By: Ac_K * Apply suggestions from code review Co-Authored-By: Ac_K --- Ryujinx.Graphics/Graphics3d/NvGpuEngine2d.cs | 7 +- Ryujinx.Graphics/Graphics3d/NvGpuEngine3d.cs | 59 ++++++++ .../Graphics3d/NvGpuEngineM2mf.cs | 13 +- .../Graphics3d/NvGpuEngineP2mf.cs | 17 ++- Ryujinx.Profiler/ProfileConfig.cs | 126 ++++++++++++++++++ 5 files changed, 219 insertions(+), 3 deletions(-) diff --git a/Ryujinx.Graphics/Graphics3d/NvGpuEngine2d.cs b/Ryujinx.Graphics/Graphics3d/NvGpuEngine2d.cs index 361b1706..d42dca28 100644 --- a/Ryujinx.Graphics/Graphics3d/NvGpuEngine2d.cs +++ b/Ryujinx.Graphics/Graphics3d/NvGpuEngine2d.cs @@ -1,6 +1,7 @@ using Ryujinx.Graphics.Gal; using Ryujinx.Graphics.Memory; using Ryujinx.Graphics.Texture; +using Ryujinx.Profiler; namespace Ryujinx.Graphics.Graphics3d { @@ -40,6 +41,8 @@ namespace Ryujinx.Graphics.Graphics3d private void TextureCopy(NvGpuVmm vmm) { + Profile.Begin(Profiles.GPU.Engine2d.TextureCopy); + CopyOperation operation = (CopyOperation)ReadRegister(NvGpuEngine2dReg.CopyOperation); int dstFormat = ReadRegister(NvGpuEngine2dReg.DstFormat); @@ -221,6 +224,8 @@ namespace Ryujinx.Graphics.Graphics3d dstBlitH); vmm.IsRegionModified(dstKey, ImageUtils.GetSize(dstTexture), NvGpuBufferType.Texture); + + Profile.End(Profiles.GPU.Engine2d.TextureCopy); } private static GalMemoryLayout GetLayout(bool linear) @@ -255,4 +260,4 @@ namespace Ryujinx.Graphics.Graphics3d return Registers[(int)reg]; } } -} \ No newline at end of file +} diff --git a/Ryujinx.Graphics/Graphics3d/NvGpuEngine3d.cs b/Ryujinx.Graphics/Graphics3d/NvGpuEngine3d.cs index bbed642b..29b6cc5d 100644 --- a/Ryujinx.Graphics/Graphics3d/NvGpuEngine3d.cs +++ b/Ryujinx.Graphics/Graphics3d/NvGpuEngine3d.cs @@ -5,6 +5,7 @@ using Ryujinx.Graphics.Shader; using Ryujinx.Graphics.Texture; using System; using System.Collections.Generic; +using Ryujinx.Profiler; namespace Ryujinx.Graphics.Graphics3d { @@ -89,7 +90,15 @@ namespace Ryujinx.Graphics.Graphics3d { if (_methods.TryGetValue(methCall.Method, out NvGpuMethod method)) { + ProfileConfig profile = Profiles.GPU.Engine3d.CallMethod; + + profile.SessionItem = method.Method.Name; + + Profile.Begin(profile); + method(vmm, methCall); + + Profile.End(profile); } else { @@ -99,13 +108,19 @@ namespace Ryujinx.Graphics.Graphics3d private void VertexEndGl(NvGpuVmm vmm, GpuMethodCall methCall) { + Profile.Begin(Profiles.GPU.Engine3d.VertexEnd); + LockCaches(); + Profile.Begin(Profiles.GPU.Engine3d.ConfigureState); + GalPipelineState state = new GalPipelineState(); // Framebuffer must be run configured because viewport dimensions may be used in other methods SetFrameBuffer(state); + Profile.End(Profiles.GPU.Engine3d.ConfigureState); + for (int fbIndex = 0; fbIndex < 8; fbIndex++) { SetFrameBuffer(vmm, fbIndex); @@ -135,6 +150,8 @@ namespace Ryujinx.Graphics.Graphics3d DispatchRender(vmm, state); UnlockCaches(); + + Profile.End(Profiles.GPU.Engine3d.VertexEnd); } private void LockCaches() @@ -153,6 +170,8 @@ namespace Ryujinx.Graphics.Graphics3d private void ClearBuffers(NvGpuVmm vmm, GpuMethodCall methCall) { + Profile.Begin(Profiles.GPU.Engine3d.ClearBuffers); + int attachment = (methCall.Argument >> 6) & 0xf; GalClearBufferFlags flags = (GalClearBufferFlags)(methCall.Argument & 0x3f); @@ -178,10 +197,18 @@ namespace Ryujinx.Graphics.Graphics3d _gpu.Renderer.Pipeline.ResetDepthMask(); _gpu.Renderer.Pipeline.ResetColorMask(attachment); + + Profile.End(Profiles.GPU.Engine3d.ClearBuffers); } private void SetFrameBuffer(NvGpuVmm vmm, int fbIndex) { + ProfileConfig profile = Profiles.GPU.Engine3d.SetFrameBuffer; + + profile.SessionItem = fbIndex.ToString(); + + Profile.Begin(profile); + long va = MakeInt64From2xInt32(NvGpuEngine3dReg.FrameBufferNAddress + fbIndex * 0x10); int surfFormat = ReadRegister(NvGpuEngine3dReg.FrameBufferNFormat + fbIndex * 0x10); @@ -190,6 +217,8 @@ namespace Ryujinx.Graphics.Graphics3d { _gpu.Renderer.RenderTarget.UnbindColor(fbIndex); + Profile.End(profile); + return; } @@ -227,6 +256,8 @@ namespace Ryujinx.Graphics.Graphics3d _gpu.ResourceManager.SendColorBuffer(vmm, key, fbIndex, image); _gpu.Renderer.RenderTarget.SetViewport(fbIndex, _viewportX0, _viewportY0, _viewportX1 - _viewportX0, _viewportY1 - _viewportY0); + + Profile.End(profile); } private void SetFrameBuffer(GalPipelineState state) @@ -248,6 +279,8 @@ namespace Ryujinx.Graphics.Graphics3d private void SetZeta(NvGpuVmm vmm) { + Profile.Begin(Profiles.GPU.Engine3d.SetZeta); + long va = MakeInt64From2xInt32(NvGpuEngine3dReg.ZetaAddress); int zetaFormat = ReadRegister(NvGpuEngine3dReg.ZetaFormat); @@ -264,6 +297,8 @@ namespace Ryujinx.Graphics.Graphics3d { _gpu.Renderer.RenderTarget.UnbindZeta(); + Profile.End(Profiles.GPU.Engine3d.SetZeta); + return; } @@ -278,10 +313,14 @@ namespace Ryujinx.Graphics.Graphics3d GalImage image = new GalImage(width, height, 1, 1, 1, gobBlockHeight, 1, layout, format, GalTextureTarget.TwoD); _gpu.ResourceManager.SendZetaBuffer(vmm, key, image); + + Profile.End(Profiles.GPU.Engine3d.SetZeta); } private long[] UploadShaders(NvGpuVmm vmm) { + Profile.Begin(Profiles.GPU.Engine3d.UploadShaders); + long[] keys = new long[5]; long basePosition = MakeInt64From2xInt32(NvGpuEngine3dReg.ShaderAddress); @@ -339,6 +378,8 @@ namespace Ryujinx.Graphics.Graphics3d _gpu.Renderer.Shader.Bind(key); } + Profile.End(Profiles.GPU.Engine3d.UploadShaders); + return keys; } @@ -619,6 +660,8 @@ namespace Ryujinx.Graphics.Graphics3d private void UploadTextures(NvGpuVmm vmm, GalPipelineState state, long[] keys) { + Profile.Begin(Profiles.GPU.Engine3d.UploadTextures); + long baseShPosition = MakeInt64From2xInt32(NvGpuEngine3dReg.ShaderAddress); int textureCbIndex = ReadRegister(NvGpuEngine3dReg.TextureCbIndex); @@ -660,6 +703,8 @@ namespace Ryujinx.Graphics.Graphics3d _gpu.Renderer.Texture.Bind(key, index, image); _gpu.Renderer.Texture.SetSampler(image, sampler); } + + Profile.End(Profiles.GPU.Engine3d.UploadTextures); } private (long, GalImage, GalTextureSampler) UploadTexture(NvGpuVmm vmm, int textureHandle) @@ -671,6 +716,8 @@ namespace Ryujinx.Graphics.Graphics3d return (0, default(GalImage), default(GalTextureSampler)); } + Profile.Begin(Profiles.GPU.Engine3d.UploadTexture); + bool linkedTsc = ReadRegisterBool(NvGpuEngine3dReg.LinkedTsc); int ticIndex = (textureHandle >> 0) & 0xfffff; @@ -702,17 +749,23 @@ namespace Ryujinx.Graphics.Graphics3d if (key == -1) { + Profile.End(Profiles.GPU.Engine3d.UploadTexture); + //FIXME: Shouldn't ignore invalid addresses. return (0, default(GalImage), default(GalTextureSampler)); } _gpu.ResourceManager.SendTexture(vmm, key, image); + Profile.End(Profiles.GPU.Engine3d.UploadTexture); + return (key, image, sampler); } private void UploadConstBuffers(NvGpuVmm vmm, GalPipelineState state, long[] keys) { + Profile.Begin(Profiles.GPU.Engine3d.UploadConstBuffers); + for (int stage = 0; stage < keys.Length; stage++) { foreach (CBufferDescriptor desc in _gpu.Renderer.Shader.GetConstBufferUsage(keys[stage])) @@ -741,10 +794,14 @@ namespace Ryujinx.Graphics.Graphics3d state.ConstBufferKeys[stage][desc.Slot] = key; } } + + Profile.End(Profiles.GPU.Engine3d.UploadConstBuffers); } private void UploadVertexArrays(NvGpuVmm vmm, GalPipelineState state) { + Profile.Begin(Profiles.GPU.Engine3d.UploadVertexArrays); + long ibPosition = MakeInt64From2xInt32(NvGpuEngine3dReg.IndexArrayAddress); long iboKey = vmm.GetPhysicalAddress(ibPosition); @@ -956,6 +1013,8 @@ namespace Ryujinx.Graphics.Graphics3d state.VertexBindings[index].Divisor = vertexDivisor; state.VertexBindings[index].Attribs = attribs[index].ToArray(); } + + Profile.End(Profiles.GPU.Engine3d.UploadVertexArrays); } private void DispatchRender(NvGpuVmm vmm, GalPipelineState state) diff --git a/Ryujinx.Graphics/Graphics3d/NvGpuEngineM2mf.cs b/Ryujinx.Graphics/Graphics3d/NvGpuEngineM2mf.cs index 45b0bbd7..172c919b 100644 --- a/Ryujinx.Graphics/Graphics3d/NvGpuEngineM2mf.cs +++ b/Ryujinx.Graphics/Graphics3d/NvGpuEngineM2mf.cs @@ -1,6 +1,7 @@ using Ryujinx.Graphics.Memory; using Ryujinx.Graphics.Texture; using System.Collections.Generic; +using Ryujinx.Profiler; namespace Ryujinx.Graphics.Graphics3d { @@ -37,7 +38,13 @@ namespace Ryujinx.Graphics.Graphics3d { if (_methods.TryGetValue(methCall.Method, out NvGpuMethod method)) { + ProfileConfig profile = Profiles.GPU.EngineM2mf.CallMethod; + + profile.SessionItem = method.Method.Name; + + Profile.Begin(profile); method(vmm, methCall); + Profile.End(profile); } else { @@ -47,6 +54,8 @@ namespace Ryujinx.Graphics.Graphics3d private void Execute(NvGpuVmm vmm, GpuMethodCall methCall) { + Profile.Begin(Profiles.GPU.EngineM2mf.Execute); + //TODO: Some registers and copy modes are still not implemented. int control = methCall.Argument; @@ -176,6 +185,8 @@ namespace Ryujinx.Graphics.Graphics3d { vmm.Memory.CopyBytes(srcPa, dstPa, xCount); } + + Profile.End(Profiles.GPU.EngineM2mf.Execute); } private long MakeInt64From2xInt32(NvGpuEngineM2mfReg reg) @@ -200,4 +211,4 @@ namespace Ryujinx.Graphics.Graphics3d Registers[(int)reg] = value; } } -} \ No newline at end of file +} diff --git a/Ryujinx.Graphics/Graphics3d/NvGpuEngineP2mf.cs b/Ryujinx.Graphics/Graphics3d/NvGpuEngineP2mf.cs index d24f2303..83ad0e70 100644 --- a/Ryujinx.Graphics/Graphics3d/NvGpuEngineP2mf.cs +++ b/Ryujinx.Graphics/Graphics3d/NvGpuEngineP2mf.cs @@ -1,6 +1,7 @@ using Ryujinx.Graphics.Memory; using Ryujinx.Graphics.Texture; using System.Collections.Generic; +using Ryujinx.Profiler; namespace Ryujinx.Graphics.Graphics3d { @@ -54,7 +55,13 @@ namespace Ryujinx.Graphics.Graphics3d { if (_methods.TryGetValue(methCall.Method, out NvGpuMethod method)) { + ProfileConfig profile = Profiles.GPU.EngineP2mf.PushData; + + profile.SessionItem = method.Method.Name; + + Profile.Begin(profile); method(vmm, methCall); + Profile.End(profile); } else { @@ -64,6 +71,8 @@ namespace Ryujinx.Graphics.Graphics3d private void Execute(NvGpuVmm vmm, GpuMethodCall methCall) { + Profile.Begin(Profiles.GPU.EngineP2mf.Execute); + //TODO: Some registers and copy modes are still not implemented. int control = methCall.Argument; @@ -97,6 +106,8 @@ namespace Ryujinx.Graphics.Graphics3d _copySize = lineLengthIn * lineCount; _buffer = new byte[_copySize]; + + Profile.End(Profiles.GPU.EngineP2mf.Execute); } private void PushData(NvGpuVmm vmm, GpuMethodCall methCall) @@ -106,6 +117,8 @@ namespace Ryujinx.Graphics.Graphics3d return; } + Profile.Begin(Profiles.GPU.EngineP2mf.PushData); + for (int shift = 0; shift < 32 && _copyOffset < _copySize; shift += 8, _copyOffset++) { _buffer[_copyOffset] = (byte)(methCall.Argument >> shift); @@ -137,6 +150,8 @@ namespace Ryujinx.Graphics.Graphics3d _buffer = null; } + + Profile.End(Profiles.GPU.EngineP2mf.PushData); } private long MakeInt64From2xInt32(NvGpuEngineP2mfReg reg) @@ -161,4 +176,4 @@ namespace Ryujinx.Graphics.Graphics3d Registers[(int)reg] = value; } } -} \ No newline at end of file +} diff --git a/Ryujinx.Profiler/ProfileConfig.cs b/Ryujinx.Profiler/ProfileConfig.cs index 6a2b2bc0..d70da602 100644 --- a/Ryujinx.Profiler/ProfileConfig.cs +++ b/Ryujinx.Profiler/ProfileConfig.cs @@ -105,6 +105,132 @@ namespace Ryujinx.Profiler }; } + public static class GPU + { + public static class Engine2d + { + public static ProfileConfig TextureCopy = new ProfileConfig() + { + Category = "GPU.Engine2D", + SessionGroup = "TextureCopy" + }; + } + + public static class Engine3d + { + public static ProfileConfig CallMethod = new ProfileConfig() + { + Category = "GPU.Engine3D", + SessionGroup = "CallMethod", + }; + + public static ProfileConfig VertexEnd = new ProfileConfig() + { + Category = "GPU.Engine3D", + SessionGroup = "VertexEnd" + }; + + public static ProfileConfig ClearBuffers = new ProfileConfig() + { + Category = "GPU.Engine3D", + SessionGroup = "ClearBuffers" + }; + + public static ProfileConfig SetFrameBuffer = new ProfileConfig() + { + Category = "GPU.Engine3D", + SessionGroup = "SetFrameBuffer", + }; + + public static ProfileConfig SetZeta = new ProfileConfig() + { + Category = "GPU.Engine3D", + SessionGroup = "SetZeta" + }; + + public static ProfileConfig UploadShaders = new ProfileConfig() + { + Category = "GPU.Engine3D", + SessionGroup = "UploadShaders" + }; + + public static ProfileConfig UploadTextures = new ProfileConfig() + { + Category = "GPU.Engine3D", + SessionGroup = "UploadTextures" + }; + + public static ProfileConfig UploadTexture = new ProfileConfig() + { + Category = "GPU.Engine3D", + SessionGroup = "UploadTexture" + }; + + public static ProfileConfig UploadConstBuffers = new ProfileConfig() + { + Category = "GPU.Engine3D", + SessionGroup = "UploadConstBuffers" + }; + + public static ProfileConfig UploadVertexArrays = new ProfileConfig() + { + Category = "GPU.Engine3D", + SessionGroup = "UploadVertexArrays" + }; + + public static ProfileConfig ConfigureState = new ProfileConfig() + { + Category = "GPU.Engine3D", + SessionGroup = "ConfigureState" + }; + } + + public static class EngineM2mf + { + public static ProfileConfig CallMethod = new ProfileConfig() + { + Category = "GPU.EngineM2mf", + SessionGroup = "CallMethod", + }; + + public static ProfileConfig Execute = new ProfileConfig() + { + Category = "GPU.EngineM2mf", + SessionGroup = "Execute", + }; + } + + public static class EngineP2mf + { + public static ProfileConfig CallMethod = new ProfileConfig() + { + Category = "GPU.EngineP2mf", + SessionGroup = "CallMethod", + }; + + public static ProfileConfig Execute = new ProfileConfig() + { + Category = "GPU.EngineP2mf", + SessionGroup = "Execute", + }; + + public static ProfileConfig PushData = new ProfileConfig() + { + Category = "GPU.EngineP2mf", + SessionGroup = "PushData", + }; + } + + public static class Shader + { + public static ProfileConfig Decompile = new ProfileConfig() + { + Category = "GPU.Shader", + SessionGroup = "Decompile", + }; + } + } + public static ProfileConfig ServiceCall = new ProfileConfig() { Category = "ServiceCall",