Ryujinx/src/Ryujinx.Graphics.OpenGL/Queries/CounterQueue.cs

225 lines
6.5 KiB
C#
Raw Normal View History

using OpenTK.Graphics.OpenGL;
using Ryujinx.Graphics.GAL;
using System;
using System.Collections.Generic;
using System.Threading;
namespace Ryujinx.Graphics.OpenGL.Queries
{
class CounterQueue : IDisposable
{
private const int QueryPoolInitialSize = 100;
public CounterType Type { get; }
public bool Disposed { get; private set; }
private readonly Queue<CounterQueueEvent> _events = new();
private CounterQueueEvent _current;
private ulong _accumulatedCounter;
Add a Multithreading layer for the GAL, multi-thread shader compilation at runtime (#2501) * Initial Implementation About as fast as nvidia GL multithreading, can be improved with faster command queuing. * Struct based command list Speeds up a bit. Still a lot of time lost to resource copy. * Do shader init while the render thread is active. * Introduce circular span pool V1 Ideally should be able to use structs instead of references for storing these spans on commands. Will try that next. * Refactor SpanRef some more Use a struct to represent SpanRef, rather than a reference. * Flush buffers on background thread * Use a span for UpdateRenderScale. Much faster than copying the array. * Calculate command size using reflection * WIP parallel shaders * Some minor optimisation * Only 2 max refs per command now. The command with 3 refs is gone. :relieved: * Don't cast on the GPU side * Remove redundant casts, force sync on window present * Fix Shader Cache * Fix host shader save. * Fixup to work with new renderer stuff * Make command Run static, use array of delegates as lookup Profile says this takes less time than the previous way. * Bring up to date * Add settings toggle. Fix Muiltithreading Off mode. * Fix warning. * Release tracking lock for flushes * Fix Conditional Render fast path with threaded gal * Make handle iteration safe when releasing the lock This is mostly temporary. * Attempt to set backend threading on driver Only really works on nvidia before launching a game. * Fix race condition with BufferModifiedRangeList, exceptions in tracking actions * Update buffer set commands * Some cleanup * Only use stutter workaround when using opengl renderer non-threaded * Add host-conditional reservation of counter events There has always been the possibility that conditional rendering could use a query object just as it is disposed by the counter queue. This change makes it so that when the host decides to use host conditional rendering, the query object is reserved so that it cannot be deleted. Counter events can optionally start reserved, as the threaded implementation can reserve them before the backend creates them, and there would otherwise be a short amount of time where the counter queue could dispose the event before a call to reserve it could be made. * Address Feedback * Make counter flush tracked again. Hopefully does not cause any issues this time. * Wait for FlushTo on the main queue thread. Currently assumes only one thread will want to FlushTo (in this case, the GPU thread) * Add SDL2 headless integration * Add HLE macro commands. Co-authored-by: Mary <mary@mary.zone>
2021-08-26 23:31:29 +01:00
private int _waiterCount;
2024-12-19 21:52:25 -03:00
private readonly Lock _lock = new();
private readonly Queue<BufferedQuery> _queryPool;
private readonly AutoResetEvent _queuedEvent = new(false);
private readonly AutoResetEvent _wakeSignal = new(false);
private readonly AutoResetEvent _eventConsumed = new(false);
private readonly Thread _consumerThread;
Add workflow to automatically check code style issues for PRs (#4670) * Add workflow to perform automated checks for PRs * Downgrade Microsoft.CodeAnalysis to 4.4.0 This is a workaround to fix issues with dotnet-format. See: - https://github.com/dotnet/format/issues/1805 - https://github.com/dotnet/format/issues/1800 * Adjust editorconfig to be more compatible with Ryujinx code-style * Adjust .editorconfig line endings to match .gitattributes * Disable 'prefer switch expression' rule * Remove naming styles These are the default rules, so we don't need to override them. * Silence IDE0060 in .editorconfig * Slightly adjust .editorconfig * Add lost workflow changes * Move .editorconfig comment to the top * .editorconfig: private static readonly fields should be _lowerCamelCase * .editorconfig: Remove alignment for declarations as well * editorconfig: Add rule for local constants * Disable CA1822 for HLE services * Disable CA1822 for ViewModels Bindings won't work with static members, but this issue is silently ignored. * Run dotnet format for the whole solution * Check result code of SDL_GetDisplayBounds * Fix dotnet format style issues * Add missing trailing commas * Update Microsoft.CodeAnalysis.CSharp to 4.6.0 Skipping 4.5.0 since it breaks dotnet format * Restore old default naming rules for dotnet format * Add naming rule exception for CPU tests * checks: Include all files before excluding paths * Fix dotnet format issues * Check dotnet format version * checks: Run dotnet format with severity info again * checks: Disable naming style rules until they won't crash the process anymore * Remove unread private member * checks: Attempt to run analyzers 3 times before giving up * checks: Enable naming style rules again with the new retry logic
2023-07-24 18:35:04 +02:00
internal CounterQueue(CounterType type)
{
Type = type;
QueryTarget glType = GetTarget(Type);
_queryPool = new Queue<BufferedQuery>(QueryPoolInitialSize);
for (int i = 0; i < QueryPoolInitialSize; i++)
{
_queryPool.Enqueue(new BufferedQuery(glType));
}
_current = new CounterQueueEvent(this, glType, 0);
_consumerThread = new Thread(EventConsumer) { Name = "CPU.CounterQueue." + (int)type };
_consumerThread.Start();
}
private void EventConsumer()
{
while (!Disposed)
{
CounterQueueEvent evt = null;
lock (_lock)
{
if (_events.Count > 0)
{
evt = _events.Dequeue();
}
}
if (evt == null)
{
_queuedEvent.WaitOne(); // No more events to go through, wait for more.
}
else
{
Add a Multithreading layer for the GAL, multi-thread shader compilation at runtime (#2501) * Initial Implementation About as fast as nvidia GL multithreading, can be improved with faster command queuing. * Struct based command list Speeds up a bit. Still a lot of time lost to resource copy. * Do shader init while the render thread is active. * Introduce circular span pool V1 Ideally should be able to use structs instead of references for storing these spans on commands. Will try that next. * Refactor SpanRef some more Use a struct to represent SpanRef, rather than a reference. * Flush buffers on background thread * Use a span for UpdateRenderScale. Much faster than copying the array. * Calculate command size using reflection * WIP parallel shaders * Some minor optimisation * Only 2 max refs per command now. The command with 3 refs is gone. :relieved: * Don't cast on the GPU side * Remove redundant casts, force sync on window present * Fix Shader Cache * Fix host shader save. * Fixup to work with new renderer stuff * Make command Run static, use array of delegates as lookup Profile says this takes less time than the previous way. * Bring up to date * Add settings toggle. Fix Muiltithreading Off mode. * Fix warning. * Release tracking lock for flushes * Fix Conditional Render fast path with threaded gal * Make handle iteration safe when releasing the lock This is mostly temporary. * Attempt to set backend threading on driver Only really works on nvidia before launching a game. * Fix race condition with BufferModifiedRangeList, exceptions in tracking actions * Update buffer set commands * Some cleanup * Only use stutter workaround when using opengl renderer non-threaded * Add host-conditional reservation of counter events There has always been the possibility that conditional rendering could use a query object just as it is disposed by the counter queue. This change makes it so that when the host decides to use host conditional rendering, the query object is reserved so that it cannot be deleted. Counter events can optionally start reserved, as the threaded implementation can reserve them before the backend creates them, and there would otherwise be a short amount of time where the counter queue could dispose the event before a call to reserve it could be made. * Address Feedback * Make counter flush tracked again. Hopefully does not cause any issues this time. * Wait for FlushTo on the main queue thread. Currently assumes only one thread will want to FlushTo (in this case, the GPU thread) * Add SDL2 headless integration * Add HLE macro commands. Co-authored-by: Mary <mary@mary.zone>
2021-08-26 23:31:29 +01:00
// Spin-wait rather than sleeping if there are any waiters, by passing null instead of the wake signal.
evt.TryConsume(ref _accumulatedCounter, true, _waiterCount == 0 ? _wakeSignal : null);
}
if (_waiterCount > 0)
{
_eventConsumed.Set();
}
}
}
internal BufferedQuery GetQueryObject()
{
// Creating/disposing query objects on a context we're sharing with will cause issues.
// So instead, make a lot of query objects on the main thread and reuse them.
lock (_lock)
{
if (_queryPool.Count > 0)
{
BufferedQuery result = _queryPool.Dequeue();
return result;
}
else
{
return new BufferedQuery(GetTarget(Type));
}
}
}
internal void ReturnQueryObject(BufferedQuery query)
{
lock (_lock)
{
_queryPool.Enqueue(query);
}
}
public CounterQueueEvent QueueReport(EventHandler<ulong> resultHandler, float divisor, ulong lastDrawIndex, bool hostReserved)
{
CounterQueueEvent result;
ulong draws = lastDrawIndex - _current.DrawIndex;
lock (_lock)
{
// A query's result only matters if more than one draw was performed during it.
// Otherwise, dummy it out and return 0 immediately.
Add a Multithreading layer for the GAL, multi-thread shader compilation at runtime (#2501) * Initial Implementation About as fast as nvidia GL multithreading, can be improved with faster command queuing. * Struct based command list Speeds up a bit. Still a lot of time lost to resource copy. * Do shader init while the render thread is active. * Introduce circular span pool V1 Ideally should be able to use structs instead of references for storing these spans on commands. Will try that next. * Refactor SpanRef some more Use a struct to represent SpanRef, rather than a reference. * Flush buffers on background thread * Use a span for UpdateRenderScale. Much faster than copying the array. * Calculate command size using reflection * WIP parallel shaders * Some minor optimisation * Only 2 max refs per command now. The command with 3 refs is gone. :relieved: * Don't cast on the GPU side * Remove redundant casts, force sync on window present * Fix Shader Cache * Fix host shader save. * Fixup to work with new renderer stuff * Make command Run static, use array of delegates as lookup Profile says this takes less time than the previous way. * Bring up to date * Add settings toggle. Fix Muiltithreading Off mode. * Fix warning. * Release tracking lock for flushes * Fix Conditional Render fast path with threaded gal * Make handle iteration safe when releasing the lock This is mostly temporary. * Attempt to set backend threading on driver Only really works on nvidia before launching a game. * Fix race condition with BufferModifiedRangeList, exceptions in tracking actions * Update buffer set commands * Some cleanup * Only use stutter workaround when using opengl renderer non-threaded * Add host-conditional reservation of counter events There has always been the possibility that conditional rendering could use a query object just as it is disposed by the counter queue. This change makes it so that when the host decides to use host conditional rendering, the query object is reserved so that it cannot be deleted. Counter events can optionally start reserved, as the threaded implementation can reserve them before the backend creates them, and there would otherwise be a short amount of time where the counter queue could dispose the event before a call to reserve it could be made. * Address Feedback * Make counter flush tracked again. Hopefully does not cause any issues this time. * Wait for FlushTo on the main queue thread. Currently assumes only one thread will want to FlushTo (in this case, the GPU thread) * Add SDL2 headless integration * Add HLE macro commands. Co-authored-by: Mary <mary@mary.zone>
2021-08-26 23:31:29 +01:00
if (hostReserved)
{
// This counter event is guaranteed to be available for host conditional rendering.
_current.ReserveForHostAccess();
}
_current.Complete(draws > 0, divisor);
_events.Enqueue(_current);
_current.OnResult += resultHandler;
result = _current;
_current = new CounterQueueEvent(this, GetTarget(Type), lastDrawIndex);
}
_queuedEvent.Set();
return result;
}
public void QueueReset()
{
lock (_lock)
{
_current.Clear();
}
}
private static QueryTarget GetTarget(CounterType type)
{
return type switch
{
CounterType.SamplesPassed => QueryTarget.SamplesPassed,
CounterType.PrimitivesGenerated => QueryTarget.PrimitivesGenerated,
CounterType.TransformFeedbackPrimitivesWritten => QueryTarget.TransformFeedbackPrimitivesWritten,
_ => QueryTarget.SamplesPassed,
};
}
public void Flush(bool blocking)
{
if (!blocking)
{
// Just wake the consumer thread - it will update the queries.
_wakeSignal.Set();
return;
}
lock (_lock)
{
// Tell the queue to process all events.
while (_events.Count > 0)
{
CounterQueueEvent flush = _events.Peek();
if (!flush.TryConsume(ref _accumulatedCounter, true))
{
return; // If not blocking, then return when we encounter an event that is not ready yet.
}
_events.Dequeue();
}
}
}
public void FlushTo(CounterQueueEvent evt)
{
Add a Multithreading layer for the GAL, multi-thread shader compilation at runtime (#2501) * Initial Implementation About as fast as nvidia GL multithreading, can be improved with faster command queuing. * Struct based command list Speeds up a bit. Still a lot of time lost to resource copy. * Do shader init while the render thread is active. * Introduce circular span pool V1 Ideally should be able to use structs instead of references for storing these spans on commands. Will try that next. * Refactor SpanRef some more Use a struct to represent SpanRef, rather than a reference. * Flush buffers on background thread * Use a span for UpdateRenderScale. Much faster than copying the array. * Calculate command size using reflection * WIP parallel shaders * Some minor optimisation * Only 2 max refs per command now. The command with 3 refs is gone. :relieved: * Don't cast on the GPU side * Remove redundant casts, force sync on window present * Fix Shader Cache * Fix host shader save. * Fixup to work with new renderer stuff * Make command Run static, use array of delegates as lookup Profile says this takes less time than the previous way. * Bring up to date * Add settings toggle. Fix Muiltithreading Off mode. * Fix warning. * Release tracking lock for flushes * Fix Conditional Render fast path with threaded gal * Make handle iteration safe when releasing the lock This is mostly temporary. * Attempt to set backend threading on driver Only really works on nvidia before launching a game. * Fix race condition with BufferModifiedRangeList, exceptions in tracking actions * Update buffer set commands * Some cleanup * Only use stutter workaround when using opengl renderer non-threaded * Add host-conditional reservation of counter events There has always been the possibility that conditional rendering could use a query object just as it is disposed by the counter queue. This change makes it so that when the host decides to use host conditional rendering, the query object is reserved so that it cannot be deleted. Counter events can optionally start reserved, as the threaded implementation can reserve them before the backend creates them, and there would otherwise be a short amount of time where the counter queue could dispose the event before a call to reserve it could be made. * Address Feedback * Make counter flush tracked again. Hopefully does not cause any issues this time. * Wait for FlushTo on the main queue thread. Currently assumes only one thread will want to FlushTo (in this case, the GPU thread) * Add SDL2 headless integration * Add HLE macro commands. Co-authored-by: Mary <mary@mary.zone>
2021-08-26 23:31:29 +01:00
// Flush the counter queue on the main thread.
Add a Multithreading layer for the GAL, multi-thread shader compilation at runtime (#2501) * Initial Implementation About as fast as nvidia GL multithreading, can be improved with faster command queuing. * Struct based command list Speeds up a bit. Still a lot of time lost to resource copy. * Do shader init while the render thread is active. * Introduce circular span pool V1 Ideally should be able to use structs instead of references for storing these spans on commands. Will try that next. * Refactor SpanRef some more Use a struct to represent SpanRef, rather than a reference. * Flush buffers on background thread * Use a span for UpdateRenderScale. Much faster than copying the array. * Calculate command size using reflection * WIP parallel shaders * Some minor optimisation * Only 2 max refs per command now. The command with 3 refs is gone. :relieved: * Don't cast on the GPU side * Remove redundant casts, force sync on window present * Fix Shader Cache * Fix host shader save. * Fixup to work with new renderer stuff * Make command Run static, use array of delegates as lookup Profile says this takes less time than the previous way. * Bring up to date * Add settings toggle. Fix Muiltithreading Off mode. * Fix warning. * Release tracking lock for flushes * Fix Conditional Render fast path with threaded gal * Make handle iteration safe when releasing the lock This is mostly temporary. * Attempt to set backend threading on driver Only really works on nvidia before launching a game. * Fix race condition with BufferModifiedRangeList, exceptions in tracking actions * Update buffer set commands * Some cleanup * Only use stutter workaround when using opengl renderer non-threaded * Add host-conditional reservation of counter events There has always been the possibility that conditional rendering could use a query object just as it is disposed by the counter queue. This change makes it so that when the host decides to use host conditional rendering, the query object is reserved so that it cannot be deleted. Counter events can optionally start reserved, as the threaded implementation can reserve them before the backend creates them, and there would otherwise be a short amount of time where the counter queue could dispose the event before a call to reserve it could be made. * Address Feedback * Make counter flush tracked again. Hopefully does not cause any issues this time. * Wait for FlushTo on the main queue thread. Currently assumes only one thread will want to FlushTo (in this case, the GPU thread) * Add SDL2 headless integration * Add HLE macro commands. Co-authored-by: Mary <mary@mary.zone>
2021-08-26 23:31:29 +01:00
Interlocked.Increment(ref _waiterCount);
Add a Multithreading layer for the GAL, multi-thread shader compilation at runtime (#2501) * Initial Implementation About as fast as nvidia GL multithreading, can be improved with faster command queuing. * Struct based command list Speeds up a bit. Still a lot of time lost to resource copy. * Do shader init while the render thread is active. * Introduce circular span pool V1 Ideally should be able to use structs instead of references for storing these spans on commands. Will try that next. * Refactor SpanRef some more Use a struct to represent SpanRef, rather than a reference. * Flush buffers on background thread * Use a span for UpdateRenderScale. Much faster than copying the array. * Calculate command size using reflection * WIP parallel shaders * Some minor optimisation * Only 2 max refs per command now. The command with 3 refs is gone. :relieved: * Don't cast on the GPU side * Remove redundant casts, force sync on window present * Fix Shader Cache * Fix host shader save. * Fixup to work with new renderer stuff * Make command Run static, use array of delegates as lookup Profile says this takes less time than the previous way. * Bring up to date * Add settings toggle. Fix Muiltithreading Off mode. * Fix warning. * Release tracking lock for flushes * Fix Conditional Render fast path with threaded gal * Make handle iteration safe when releasing the lock This is mostly temporary. * Attempt to set backend threading on driver Only really works on nvidia before launching a game. * Fix race condition with BufferModifiedRangeList, exceptions in tracking actions * Update buffer set commands * Some cleanup * Only use stutter workaround when using opengl renderer non-threaded * Add host-conditional reservation of counter events There has always been the possibility that conditional rendering could use a query object just as it is disposed by the counter queue. This change makes it so that when the host decides to use host conditional rendering, the query object is reserved so that it cannot be deleted. Counter events can optionally start reserved, as the threaded implementation can reserve them before the backend creates them, and there would otherwise be a short amount of time where the counter queue could dispose the event before a call to reserve it could be made. * Address Feedback * Make counter flush tracked again. Hopefully does not cause any issues this time. * Wait for FlushTo on the main queue thread. Currently assumes only one thread will want to FlushTo (in this case, the GPU thread) * Add SDL2 headless integration * Add HLE macro commands. Co-authored-by: Mary <mary@mary.zone>
2021-08-26 23:31:29 +01:00
_wakeSignal.Set();
while (!evt.Disposed)
{
_eventConsumed.WaitOne(1);
}
Add a Multithreading layer for the GAL, multi-thread shader compilation at runtime (#2501) * Initial Implementation About as fast as nvidia GL multithreading, can be improved with faster command queuing. * Struct based command list Speeds up a bit. Still a lot of time lost to resource copy. * Do shader init while the render thread is active. * Introduce circular span pool V1 Ideally should be able to use structs instead of references for storing these spans on commands. Will try that next. * Refactor SpanRef some more Use a struct to represent SpanRef, rather than a reference. * Flush buffers on background thread * Use a span for UpdateRenderScale. Much faster than copying the array. * Calculate command size using reflection * WIP parallel shaders * Some minor optimisation * Only 2 max refs per command now. The command with 3 refs is gone. :relieved: * Don't cast on the GPU side * Remove redundant casts, force sync on window present * Fix Shader Cache * Fix host shader save. * Fixup to work with new renderer stuff * Make command Run static, use array of delegates as lookup Profile says this takes less time than the previous way. * Bring up to date * Add settings toggle. Fix Muiltithreading Off mode. * Fix warning. * Release tracking lock for flushes * Fix Conditional Render fast path with threaded gal * Make handle iteration safe when releasing the lock This is mostly temporary. * Attempt to set backend threading on driver Only really works on nvidia before launching a game. * Fix race condition with BufferModifiedRangeList, exceptions in tracking actions * Update buffer set commands * Some cleanup * Only use stutter workaround when using opengl renderer non-threaded * Add host-conditional reservation of counter events There has always been the possibility that conditional rendering could use a query object just as it is disposed by the counter queue. This change makes it so that when the host decides to use host conditional rendering, the query object is reserved so that it cannot be deleted. Counter events can optionally start reserved, as the threaded implementation can reserve them before the backend creates them, and there would otherwise be a short amount of time where the counter queue could dispose the event before a call to reserve it could be made. * Address Feedback * Make counter flush tracked again. Hopefully does not cause any issues this time. * Wait for FlushTo on the main queue thread. Currently assumes only one thread will want to FlushTo (in this case, the GPU thread) * Add SDL2 headless integration * Add HLE macro commands. Co-authored-by: Mary <mary@mary.zone>
2021-08-26 23:31:29 +01:00
Interlocked.Decrement(ref _waiterCount);
}
public void Dispose()
{
lock (_lock)
{
while (_events.Count > 0)
{
CounterQueueEvent evt = _events.Dequeue();
evt.Dispose();
}
Disposed = true;
}
_queuedEvent.Set();
_consumerThread.Join();
foreach (BufferedQuery query in _queryPool)
{
query.Dispose();
}
Add a Multithreading layer for the GAL, multi-thread shader compilation at runtime (#2501) * Initial Implementation About as fast as nvidia GL multithreading, can be improved with faster command queuing. * Struct based command list Speeds up a bit. Still a lot of time lost to resource copy. * Do shader init while the render thread is active. * Introduce circular span pool V1 Ideally should be able to use structs instead of references for storing these spans on commands. Will try that next. * Refactor SpanRef some more Use a struct to represent SpanRef, rather than a reference. * Flush buffers on background thread * Use a span for UpdateRenderScale. Much faster than copying the array. * Calculate command size using reflection * WIP parallel shaders * Some minor optimisation * Only 2 max refs per command now. The command with 3 refs is gone. :relieved: * Don't cast on the GPU side * Remove redundant casts, force sync on window present * Fix Shader Cache * Fix host shader save. * Fixup to work with new renderer stuff * Make command Run static, use array of delegates as lookup Profile says this takes less time than the previous way. * Bring up to date * Add settings toggle. Fix Muiltithreading Off mode. * Fix warning. * Release tracking lock for flushes * Fix Conditional Render fast path with threaded gal * Make handle iteration safe when releasing the lock This is mostly temporary. * Attempt to set backend threading on driver Only really works on nvidia before launching a game. * Fix race condition with BufferModifiedRangeList, exceptions in tracking actions * Update buffer set commands * Some cleanup * Only use stutter workaround when using opengl renderer non-threaded * Add host-conditional reservation of counter events There has always been the possibility that conditional rendering could use a query object just as it is disposed by the counter queue. This change makes it so that when the host decides to use host conditional rendering, the query object is reserved so that it cannot be deleted. Counter events can optionally start reserved, as the threaded implementation can reserve them before the backend creates them, and there would otherwise be a short amount of time where the counter queue could dispose the event before a call to reserve it could be made. * Address Feedback * Make counter flush tracked again. Hopefully does not cause any issues this time. * Wait for FlushTo on the main queue thread. Currently assumes only one thread will want to FlushTo (in this case, the GPU thread) * Add SDL2 headless integration * Add HLE macro commands. Co-authored-by: Mary <mary@mary.zone>
2021-08-26 23:31:29 +01:00
_queuedEvent.Dispose();
_wakeSignal.Dispose();
_eventConsumed.Dispose();
}
}
}