1
0
Fork 0
mirror of https://github.com/Ryujinx/Ryujinx.git synced 2024-12-27 07:46:02 +00:00
Ryujinx/Ryujinx.Audio.Backends.SoundIo/SoundIoHardwareDeviceSession.cs

439 lines
17 KiB
C#
Raw Normal View History

using Ryujinx.Audio.Backends.Common;
using Ryujinx.Audio.Backends.SoundIo.Native;
using Ryujinx.Audio.Common;
using Ryujinx.Memory;
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
using System;
using System.Collections.Concurrent;
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
using System.Runtime.CompilerServices;
using System.Threading;
using static Ryujinx.Audio.Backends.SoundIo.Native.SoundIo;
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
namespace Ryujinx.Audio.Backends.SoundIo
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
{
class SoundIoHardwareDeviceSession : HardwareDeviceSessionOutputBase
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
{
private SoundIoHardwareDeviceDriver _driver;
private ConcurrentQueue<SoundIoAudioBuffer> _queuedBuffers;
private SoundIoOutStreamContext _outputStream;
private DynamicRingBuffer _ringBuffer;
private ulong _playedSampleCount;
private ManualResetEvent _updateRequiredEvent;
private int _disposeState;
public SoundIoHardwareDeviceSession(SoundIoHardwareDeviceDriver driver, IVirtualMemoryManager memoryManager, SampleFormat requestedSampleFormat, uint requestedSampleRate, uint requestedChannelCount, float requestedVolume) : base(memoryManager, requestedSampleFormat, requestedSampleRate, requestedChannelCount)
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
{
_driver = driver;
_updateRequiredEvent = _driver.GetUpdateRequiredEvent();
_queuedBuffers = new ConcurrentQueue<SoundIoAudioBuffer>();
_ringBuffer = new DynamicRingBuffer();
SetupOutputStream(requestedVolume);
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
}
private void SetupOutputStream(float requestedVolume)
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
{
_outputStream = _driver.OpenStream(RequestedSampleFormat, RequestedSampleRate, RequestedChannelCount);
_outputStream.WriteCallback += Update;
_outputStream.Volume = requestedVolume;
// TODO: Setup other callbacks (errors, ect).
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
_outputStream.Open();
}
public override ulong GetPlayedSampleCount()
{
return Interlocked.Read(ref _playedSampleCount);
}
public override float GetVolume()
{
return _outputStream.Volume;
}
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
public override void PrepareToClose() { }
public override void QueueBuffer(AudioBuffer buffer)
{
SoundIoAudioBuffer driverBuffer = new SoundIoAudioBuffer(buffer.DataPointer, GetSampleCount(buffer));
_ringBuffer.Write(buffer.Data, 0, buffer.Data.Length);
_queuedBuffers.Enqueue(driverBuffer);
}
public override void SetVolume(float volume)
{
_outputStream.SetVolume(volume);
}
public override void Start()
{
_outputStream.Start();
_outputStream.Pause(false);
_driver.FlushContextEvents();
}
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
public override void Stop()
{
_outputStream.Pause(true);
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
_driver.FlushContextEvents();
}
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
public override void UnregisterBuffer(AudioBuffer buffer) {}
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
public override bool WasBufferFullyConsumed(AudioBuffer buffer)
{
if (!_queuedBuffers.TryPeek(out SoundIoAudioBuffer driverBuffer))
{
return true;
}
return driverBuffer.DriverIdentifier != buffer.DataPointer;
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
}
private unsafe void Update(int minFrameCount, int maxFrameCount)
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
{
int bytesPerFrame = _outputStream.BytesPerFrame;
uint bytesPerSample = (uint)_outputStream.BytesPerSample;
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
int bufferedFrames = _ringBuffer.Length / bytesPerFrame;
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
int frameCount = Math.Min(bufferedFrames, maxFrameCount);
if (frameCount == 0)
{
return;
}
Span<SoundIoChannelArea> areas = _outputStream.BeginWrite(ref frameCount);
int channelCount = areas.Length;
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
byte[] samples = new byte[frameCount * bytesPerFrame];
_ringBuffer.Read(samples, 0, samples.Length);
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
// This is a huge ugly block of code, but we save
// a significant amount of time over the generic
// loop that handles other channel counts.
// TODO: Is this still right in 2022?
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
// Mono
if (channelCount == 1)
{
ref SoundIoChannelArea area = ref areas[0];
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
fixed (byte* srcptr = samples)
{
if (bytesPerSample == 1)
{
for (int frame = 0; frame < frameCount; frame++)
{
((byte*)area.Pointer)[0] = srcptr[frame * bytesPerFrame];
area.Pointer += area.Step;
}
}
else if (bytesPerSample == 2)
{
for (int frame = 0; frame < frameCount; frame++)
{
((short*)area.Pointer)[0] = ((short*)srcptr)[frame * bytesPerFrame >> 1];
area.Pointer += area.Step;
}
}
else if (bytesPerSample == 4)
{
for (int frame = 0; frame < frameCount; frame++)
{
((int*)area.Pointer)[0] = ((int*)srcptr)[frame * bytesPerFrame >> 2];
area.Pointer += area.Step;
}
}
else
{
for (int frame = 0; frame < frameCount; frame++)
{
Unsafe.CopyBlockUnaligned((byte*)area.Pointer, srcptr + (frame * bytesPerFrame), bytesPerSample);
area.Pointer += area.Step;
}
}
}
}
// Stereo
else if (channelCount == 2)
{
ref SoundIoChannelArea area1 = ref areas[0];
ref SoundIoChannelArea area2 = ref areas[1];
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
fixed (byte* srcptr = samples)
{
if (bytesPerSample == 1)
{
for (int frame = 0; frame < frameCount; frame++)
{
// Channel 1
((byte*)area1.Pointer)[0] = srcptr[(frame * bytesPerFrame) + 0];
// Channel 2
((byte*)area2.Pointer)[0] = srcptr[(frame * bytesPerFrame) + 1];
area1.Pointer += area1.Step;
area2.Pointer += area2.Step;
}
}
else if (bytesPerSample == 2)
{
for (int frame = 0; frame < frameCount; frame++)
{
// Channel 1
((short*)area1.Pointer)[0] = ((short*)srcptr)[(frame * bytesPerFrame >> 1) + 0];
// Channel 2
((short*)area2.Pointer)[0] = ((short*)srcptr)[(frame * bytesPerFrame >> 1) + 1];
area1.Pointer += area1.Step;
area2.Pointer += area2.Step;
}
}
else if (bytesPerSample == 4)
{
for (int frame = 0; frame < frameCount; frame++)
{
// Channel 1
((int*)area1.Pointer)[0] = ((int*)srcptr)[(frame * bytesPerFrame >> 2) + 0];
// Channel 2
((int*)area2.Pointer)[0] = ((int*)srcptr)[(frame * bytesPerFrame >> 2) + 1];
area1.Pointer += area1.Step;
area2.Pointer += area2.Step;
}
}
else
{
for (int frame = 0; frame < frameCount; frame++)
{
// Channel 1
Unsafe.CopyBlockUnaligned((byte*)area1.Pointer, srcptr + (frame * bytesPerFrame) + (0 * bytesPerSample), bytesPerSample);
// Channel 2
Unsafe.CopyBlockUnaligned((byte*)area2.Pointer, srcptr + (frame * bytesPerFrame) + (1 * bytesPerSample), bytesPerSample);
area1.Pointer += area1.Step;
area2.Pointer += area2.Step;
}
}
}
}
// Surround
else if (channelCount == 6)
{
ref SoundIoChannelArea area1 = ref areas[0];
ref SoundIoChannelArea area2 = ref areas[1];
ref SoundIoChannelArea area3 = ref areas[2];
ref SoundIoChannelArea area4 = ref areas[3];
ref SoundIoChannelArea area5 = ref areas[4];
ref SoundIoChannelArea area6 = ref areas[5];
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
fixed (byte* srcptr = samples)
{
if (bytesPerSample == 1)
{
for (int frame = 0; frame < frameCount; frame++)
{
// Channel 1
((byte*)area1.Pointer)[0] = srcptr[(frame * bytesPerFrame) + 0];
// Channel 2
((byte*)area2.Pointer)[0] = srcptr[(frame * bytesPerFrame) + 1];
// Channel 3
((byte*)area3.Pointer)[0] = srcptr[(frame * bytesPerFrame) + 2];
// Channel 4
((byte*)area4.Pointer)[0] = srcptr[(frame * bytesPerFrame) + 3];
// Channel 5
((byte*)area5.Pointer)[0] = srcptr[(frame * bytesPerFrame) + 4];
// Channel 6
((byte*)area6.Pointer)[0] = srcptr[(frame * bytesPerFrame) + 5];
area1.Pointer += area1.Step;
area2.Pointer += area2.Step;
area3.Pointer += area3.Step;
area4.Pointer += area4.Step;
area5.Pointer += area5.Step;
area6.Pointer += area6.Step;
}
}
else if (bytesPerSample == 2)
{
for (int frame = 0; frame < frameCount; frame++)
{
// Channel 1
((short*)area1.Pointer)[0] = ((short*)srcptr)[(frame * bytesPerFrame >> 1) + 0];
// Channel 2
((short*)area2.Pointer)[0] = ((short*)srcptr)[(frame * bytesPerFrame >> 1) + 1];
// Channel 3
((short*)area3.Pointer)[0] = ((short*)srcptr)[(frame * bytesPerFrame >> 1) + 2];
// Channel 4
((short*)area4.Pointer)[0] = ((short*)srcptr)[(frame * bytesPerFrame >> 1) + 3];
// Channel 5
((short*)area5.Pointer)[0] = ((short*)srcptr)[(frame * bytesPerFrame >> 1) + 4];
// Channel 6
((short*)area6.Pointer)[0] = ((short*)srcptr)[(frame * bytesPerFrame >> 1) + 5];
area1.Pointer += area1.Step;
area2.Pointer += area2.Step;
area3.Pointer += area3.Step;
area4.Pointer += area4.Step;
area5.Pointer += area5.Step;
area6.Pointer += area6.Step;
}
}
else if (bytesPerSample == 4)
{
for (int frame = 0; frame < frameCount; frame++)
{
// Channel 1
((int*)area1.Pointer)[0] = ((int*)srcptr)[(frame * bytesPerFrame >> 2) + 0];
// Channel 2
((int*)area2.Pointer)[0] = ((int*)srcptr)[(frame * bytesPerFrame >> 2) + 1];
// Channel 3
((int*)area3.Pointer)[0] = ((int*)srcptr)[(frame * bytesPerFrame >> 2) + 2];
// Channel 4
((int*)area4.Pointer)[0] = ((int*)srcptr)[(frame * bytesPerFrame >> 2) + 3];
// Channel 5
((int*)area5.Pointer)[0] = ((int*)srcptr)[(frame * bytesPerFrame >> 2) + 4];
// Channel 6
((int*)area6.Pointer)[0] = ((int*)srcptr)[(frame * bytesPerFrame >> 2) + 5];
area1.Pointer += area1.Step;
area2.Pointer += area2.Step;
area3.Pointer += area3.Step;
area4.Pointer += area4.Step;
area5.Pointer += area5.Step;
area6.Pointer += area6.Step;
}
}
else
{
for (int frame = 0; frame < frameCount; frame++)
{
// Channel 1
Unsafe.CopyBlockUnaligned((byte*)area1.Pointer, srcptr + (frame * bytesPerFrame) + (0 * bytesPerSample), bytesPerSample);
// Channel 2
Unsafe.CopyBlockUnaligned((byte*)area2.Pointer, srcptr + (frame * bytesPerFrame) + (1 * bytesPerSample), bytesPerSample);
// Channel 3
Unsafe.CopyBlockUnaligned((byte*)area3.Pointer, srcptr + (frame * bytesPerFrame) + (2 * bytesPerSample), bytesPerSample);
// Channel 4
Unsafe.CopyBlockUnaligned((byte*)area4.Pointer, srcptr + (frame * bytesPerFrame) + (3 * bytesPerSample), bytesPerSample);
// Channel 5
Unsafe.CopyBlockUnaligned((byte*)area5.Pointer, srcptr + (frame * bytesPerFrame) + (4 * bytesPerSample), bytesPerSample);
// Channel 6
Unsafe.CopyBlockUnaligned((byte*)area6.Pointer, srcptr + (frame * bytesPerFrame) + (5 * bytesPerSample), bytesPerSample);
area1.Pointer += area1.Step;
area2.Pointer += area2.Step;
area3.Pointer += area3.Step;
area4.Pointer += area4.Step;
area5.Pointer += area5.Step;
area6.Pointer += area6.Step;
}
}
}
}
// Every other channel count
else
{
fixed (byte* srcptr = samples)
{
for (int frame = 0; frame < frameCount; frame++)
{
for (int channel = 0; channel < areas.Length; channel++)
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
{
// Copy channel by channel, frame by frame. This is slow!
Unsafe.CopyBlockUnaligned((byte*)areas[channel].Pointer, srcptr + (frame * bytesPerFrame) + (channel * bytesPerSample), bytesPerSample);
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
areas[channel].Pointer += areas[channel].Step;
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
}
}
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
}
}
_outputStream.EndWrite();
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
ulong sampleCount = (ulong)(samples.Length / bytesPerSample / channelCount);
ulong availaibleSampleCount = sampleCount;
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
bool needUpdate = false;
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
while (availaibleSampleCount > 0 && _queuedBuffers.TryPeek(out SoundIoAudioBuffer driverBuffer))
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
{
ulong sampleStillNeeded = driverBuffer.SampleCount - Interlocked.Read(ref driverBuffer.SamplePlayed);
ulong playedAudioBufferSampleCount = Math.Min(sampleStillNeeded, availaibleSampleCount);
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
Interlocked.Add(ref driverBuffer.SamplePlayed, playedAudioBufferSampleCount);
availaibleSampleCount -= playedAudioBufferSampleCount;
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
if (Interlocked.Read(ref driverBuffer.SamplePlayed) == driverBuffer.SampleCount)
{
_queuedBuffers.TryDequeue(out _);
needUpdate = true;
}
Interlocked.Add(ref _playedSampleCount, playedAudioBufferSampleCount);
}
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
// Notify the output if needed.
if (needUpdate)
{
_updateRequiredEvent.Set();
}
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
}
protected virtual void Dispose(bool disposing)
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
{
if (disposing && _driver.Unregister(this))
{
PrepareToClose();
Stop();
_outputStream.Dispose();
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
}
}
public override void Dispose()
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
{
if (Interlocked.CompareExchange(ref _disposeState, 1, 0) == 0)
{
Dispose(true);
}
Implement libsoundio as an alternative audio backend (#406) * Audio: Implement libsoundio as an alternative audio backend libsoundio will be preferred over OpenAL if it is available on the machine. If neither are available, it will fallback to a dummy audio renderer that outputs no sound. * Audio: Fix SoundIoRingBuffer documentation * Audio: Unroll and optimize the audio write callback Copying one sample at a time is slow, this unrolls the most common audio channel layouts and manually copies the bytes between source and destination. This is over 2x faster than calling CopyBlockUnaligned every sample. * Audio: Optimize the write callback further This dramatically reduces the audio buffer copy time. When the sample size is one of handled sample sizes the buffer copy operation is almost 10x faster than CopyBlockAligned. This works by copying full samples at a time, rather than the individual bytes that make up the sample. This allows for 2x or 4x faster copy operations depending on sample size. * Audio: Fix typo in Stereo write callback * Audio: Fix Surround (5.1) audio write callback * Audio: Update Documentation * Audio: Use built-in Unsafe.SizeOf<T>() Built-in `SizeOf<T>()` is 10x faster than our `TypeSize<T>` helper. This also helps reduce code surface area. * Audio: Keep fixed buffer style consistent * Audio: Address styling nits * Audio: More style nits * Audio: Add additional documentation * Audio: Move libsoundio bindings internal As per discussion, moving the libsoundio native bindings into Ryujinx.Audio * Audio: Bump Target Framework back up to .NET Core 2.1 * Audio: Remove voice mixing optimizations. Leaves Saturation optimizations in place.
2018-11-15 02:22:50 +00:00
}
}
}