ryujinx-archive/Ryujinx-git
2
0
Fork 0
Code Issues Pull requests 10 Projects Releases Packages Wiki Activity Actions
Ryujinx-git/Ryujinx.Audio/Renderer/Server/CommandGenerator.cs

1014 lines
45 KiB
C#
Raw Normal View History

Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
using Ryujinx.Audio.Common;
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
using Ryujinx.Audio.Renderer.Common;
using Ryujinx.Audio.Renderer.Dsp.Command;
using Ryujinx.Audio.Renderer.Dsp.State;
using Ryujinx.Audio.Renderer.Parameter;
using Ryujinx.Audio.Renderer.Server.Effect;
using Ryujinx.Audio.Renderer.Server.Mix;
using Ryujinx.Audio.Renderer.Server.Performance;
using Ryujinx.Audio.Renderer.Server.Sink;
using Ryujinx.Audio.Renderer.Server.Splitter;
using Ryujinx.Audio.Renderer.Server.Voice;
using Ryujinx.Audio.Renderer.Utils;
using System;
using System.Diagnostics;
namespace Ryujinx.Audio.Renderer.Server
{
public class CommandGenerator
{
private CommandBuffer _commandBuffer;
private RendererSystemContext _rendererContext;
private VoiceContext _voiceContext;
private MixContext _mixContext;
private EffectContext _effectContext;
private SinkContext _sinkContext;
private SplitterContext _splitterContext;
private PerformanceManager _performanceManager;
public CommandGenerator(CommandBuffer commandBuffer, RendererSystemContext rendererContext, VoiceContext voiceContext, MixContext mixContext, EffectContext effectContext, SinkContext sinkContext, SplitterContext splitterContext, PerformanceManager performanceManager)
{
_commandBuffer = commandBuffer;
_rendererContext = rendererContext;
_voiceContext = voiceContext;
_mixContext = mixContext;
_effectContext = effectContext;
_sinkContext = sinkContext;
_splitterContext = splitterContext;
_performanceManager = performanceManager;
Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
_commandBuffer.GenerateClearMixBuffer(Constants.InvalidNodeId);
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
}
private void GenerateDataSource(ref VoiceState voiceState, Memory<VoiceUpdateState> dspState, int channelIndex)
{
Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
if (voiceState.MixId != Constants.UnusedMixId)
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
{
ref MixState mix = ref _mixContext.GetState(voiceState.MixId);
_commandBuffer.GenerateDepopPrepare(dspState,
_rendererContext.DepopBuffer,
mix.BufferCount,
mix.BufferOffset,
voiceState.NodeId,
voiceState.WasPlaying);
}
Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
else if (voiceState.SplitterId != Constants.UnusedSplitterId)
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
{
int destinationId = 0;
while (true)
{
Span<SplitterDestination> destinationSpan = _splitterContext.GetDestination((int)voiceState.SplitterId, destinationId++);
if (destinationSpan.IsEmpty)
{
break;
}
ref SplitterDestination destination = ref destinationSpan[0];
if (destination.IsConfigured())
{
int mixId = destination.DestinationId;
Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
if (mixId < _mixContext.GetCount() && mixId != Constants.UnusedSplitterIdInt)
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
{
ref MixState mix = ref _mixContext.GetState(mixId);
_commandBuffer.GenerateDepopPrepare(dspState,
_rendererContext.DepopBuffer,
mix.BufferCount,
mix.BufferOffset,
voiceState.NodeId,
voiceState.WasPlaying);
destination.MarkAsNeedToUpdateInternalState();
}
}
}
}
if (!voiceState.WasPlaying)
{
Amadeus: Fix multi-channel PCM sources on REV8 (#1536) This add a missing offset on the output buffer of the DataSourceVersion2Command. This fix music only playing on the left channel on Fairy Tail, Family Mysteries: Poisonous Promises, SEGA AGES Sonic the Hedgehog 2 and probably more.
2020-09-06 19:54:33 +00:00
Debug.Assert(voiceState.SampleFormat != SampleFormat.Adpcm || channelIndex == 0);
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
if (_rendererContext.BehaviourContext.IsWaveBufferVersion2Supported())
{
_commandBuffer.GenerateDataSourceVersion2(ref voiceState,
dspState,
(ushort)_rendererContext.MixBufferCount,
(ushort)channelIndex,
voiceState.NodeId);
}
else
{
switch (voiceState.SampleFormat)
{
case SampleFormat.PcmInt16:
_commandBuffer.GeneratePcmInt16DataSourceVersion1(ref voiceState,
dspState,
(ushort)_rendererContext.MixBufferCount,
(ushort)channelIndex,
voiceState.NodeId);
break;
case SampleFormat.PcmFloat:
_commandBuffer.GeneratePcmFloatDataSourceVersion1(ref voiceState,
dspState,
(ushort)_rendererContext.MixBufferCount,
(ushort)channelIndex,
voiceState.NodeId);
break;
case SampleFormat.Adpcm:
_commandBuffer.GenerateAdpcmDataSourceVersion1(ref voiceState,
dspState,
(ushort)_rendererContext.MixBufferCount,
voiceState.NodeId);
break;
default:
throw new NotImplementedException($"Unsupported data source {voiceState.SampleFormat}");
}
}
}
}
private void GenerateBiquadFilterForVoice(ref VoiceState voiceState, Memory<VoiceUpdateState> state, int baseIndex, int bufferOffset, int nodeId)
{
amadeus: Update to REV10 (#2654) * amadeus: Update to REV10 This implements all the changes made with REV10 on 13.0.0. * Address Ack's comment * Address gdkchan's comment
2021-09-19 10:29:19 +00:00
bool supportsOptimizedPath = _rendererContext.BehaviourContext.IsBiquadFilterGroupedOptimizationSupported();
if (supportsOptimizedPath && voiceState.BiquadFilters[0].Enable && voiceState.BiquadFilters[1].Enable)
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
{
amadeus: Update to REV10 (#2654) * amadeus: Update to REV10 This implements all the changes made with REV10 on 13.0.0. * Address Ack's comment * Address gdkchan's comment
2021-09-19 10:29:19 +00:00
Memory<byte> biquadStateRawMemory = SpanMemoryManager<byte>.Cast(state).Slice(VoiceUpdateState.BiquadStateOffset, VoiceUpdateState.BiquadStateSize * Constants.VoiceBiquadFilterCount);
Memory<BiquadFilterState> stateMemory = SpanMemoryManager<BiquadFilterState>.Cast(biquadStateRawMemory);
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
amadeus: Update to REV10 (#2654) * amadeus: Update to REV10 This implements all the changes made with REV10 on 13.0.0. * Address Ack's comment * Address gdkchan's comment
2021-09-19 10:29:19 +00:00
_commandBuffer.GenerateGroupedBiquadFilter(baseIndex, voiceState.BiquadFilters.ToSpan(), stateMemory, bufferOffset, bufferOffset, voiceState.BiquadFilterNeedInitialization, nodeId);
}
else
{
for (int i = 0; i < voiceState.BiquadFilters.Length; i++)
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
{
amadeus: Update to REV10 (#2654) * amadeus: Update to REV10 This implements all the changes made with REV10 on 13.0.0. * Address Ack's comment * Address gdkchan's comment
2021-09-19 10:29:19 +00:00
ref BiquadFilterParameter filter = ref voiceState.BiquadFilters[i];
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
amadeus: Update to REV10 (#2654) * amadeus: Update to REV10 This implements all the changes made with REV10 on 13.0.0. * Address Ack's comment * Address gdkchan's comment
2021-09-19 10:29:19 +00:00
if (filter.Enable)
{
Memory<byte> biquadStateRawMemory = SpanMemoryManager<byte>.Cast(state).Slice(VoiceUpdateState.BiquadStateOffset, VoiceUpdateState.BiquadStateSize * Constants.VoiceBiquadFilterCount);
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
amadeus: Update to REV10 (#2654) * amadeus: Update to REV10 This implements all the changes made with REV10 on 13.0.0. * Address Ack's comment * Address gdkchan's comment
2021-09-19 10:29:19 +00:00
Memory<BiquadFilterState> stateMemory = SpanMemoryManager<BiquadFilterState>.Cast(biquadStateRawMemory);
_commandBuffer.GenerateBiquadFilter(baseIndex,
ref filter,
stateMemory.Slice(i, 1),
bufferOffset,
bufferOffset,
!voiceState.BiquadFilterNeedInitialization[i],
nodeId);
}
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
}
}
}
private void GenerateVoiceMix(Span<float> mixVolumes, Span<float> previousMixVolumes, Memory<VoiceUpdateState> state, uint bufferOffset, uint bufferCount, uint bufferIndex, int nodeId)
{
Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
if (bufferCount > Constants.VoiceChannelCountMax)
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
{
_commandBuffer.GenerateMixRampGrouped(bufferCount,
bufferIndex,
bufferOffset,
previousMixVolumes,
mixVolumes,
state,
nodeId);
}
else
{
for (int i = 0; i < bufferCount; i++)
{
float previousMixVolume = previousMixVolumes[i];
float mixVolume = mixVolumes[i];
if (mixVolume != 0.0f || previousMixVolume != 0.0f)
{
_commandBuffer.GenerateMixRamp(previousMixVolume,
mixVolume,
bufferIndex,
bufferOffset + (uint)i,
i,
state,
nodeId);
}
}
}
}
private void GenerateVoice(ref VoiceState voiceState)
{
int nodeId = voiceState.NodeId;
uint channelsCount = voiceState.ChannelsCount;
for (int channelIndex = 0; channelIndex < channelsCount; channelIndex++)
{
Memory<VoiceUpdateState> dspStateMemory = _voiceContext.GetUpdateStateForDsp(voiceState.ChannelResourceIds[channelIndex]);
ref VoiceChannelResource channelResource = ref _voiceContext.GetChannelResource(voiceState.ChannelResourceIds[channelIndex]);
PerformanceDetailType dataSourceDetailType = PerformanceDetailType.Adpcm;
if (voiceState.SampleFormat == SampleFormat.PcmInt16)
{
dataSourceDetailType = PerformanceDetailType.PcmInt16;
}
else if (voiceState.SampleFormat == SampleFormat.PcmFloat)
{
dataSourceDetailType = PerformanceDetailType.PcmFloat;
}
bool performanceInitialized = false;
PerformanceEntryAddresses performanceEntry = new PerformanceEntryAddresses();
if (_performanceManager != null && _performanceManager.IsTargetNodeId(nodeId) && _performanceManager.GetNextEntry(out performanceEntry, dataSourceDetailType, PerformanceEntryType.Voice, nodeId))
{
performanceInitialized = true;
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.Start, nodeId);
}
GenerateDataSource(ref voiceState, dspStateMemory, channelIndex);
if (performanceInitialized)
{
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.End, nodeId);
}
if (voiceState.WasPlaying)
{
voiceState.PreviousVolume = 0.0f;
}
else if (voiceState.HasAnyDestination())
{
performanceInitialized = false;
if (_performanceManager != null && _performanceManager.IsTargetNodeId(nodeId) && _performanceManager.GetNextEntry(out performanceEntry, PerformanceDetailType.BiquadFilter, PerformanceEntryType.Voice, nodeId))
{
performanceInitialized = true;
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.Start, nodeId);
}
GenerateBiquadFilterForVoice(ref voiceState, dspStateMemory, (int)_rendererContext.MixBufferCount, channelIndex, nodeId);
if (performanceInitialized)
{
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.End, nodeId);
}
performanceInitialized = false;
if (_performanceManager != null && _performanceManager.IsTargetNodeId(nodeId) && _performanceManager.GetNextEntry(out performanceEntry, PerformanceDetailType.VolumeRamp, PerformanceEntryType.Voice, nodeId))
{
performanceInitialized = true;
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.Start, nodeId);
}
_commandBuffer.GenerateVolumeRamp(voiceState.PreviousVolume,
voiceState.Volume,
_rendererContext.MixBufferCount + (uint)channelIndex,
nodeId);
if (performanceInitialized)
{
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.End, nodeId);
}
voiceState.PreviousVolume = voiceState.Volume;
Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
if (voiceState.MixId == Constants.UnusedMixId)
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
{
Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
if (voiceState.SplitterId != Constants.UnusedSplitterId)
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
{
int destinationId = channelIndex;
while (true)
{
Span<SplitterDestination> destinationSpan = _splitterContext.GetDestination((int)voiceState.SplitterId, destinationId);
if (destinationSpan.IsEmpty)
{
break;
}
ref SplitterDestination destination = ref destinationSpan[0];
destinationId += (int)channelsCount;
if (destination.IsConfigured())
{
int mixId = destination.DestinationId;
Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
if (mixId < _mixContext.GetCount() && mixId != Constants.UnusedSplitterIdInt)
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
{
ref MixState mix = ref _mixContext.GetState(mixId);
GenerateVoiceMix(destination.MixBufferVolume,
destination.PreviousMixBufferVolume,
dspStateMemory,
mix.BufferOffset,
mix.BufferCount,
_rendererContext.MixBufferCount + (uint)channelIndex,
nodeId);
destination.MarkAsNeedToUpdateInternalState();
}
}
}
}
}
else
{
ref MixState mix = ref _mixContext.GetState(voiceState.MixId);
performanceInitialized = false;
if (_performanceManager != null && _performanceManager.IsTargetNodeId(nodeId) && _performanceManager.GetNextEntry(out performanceEntry, PerformanceDetailType.Mix, PerformanceEntryType.Voice, nodeId))
{
performanceInitialized = true;
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.Start, nodeId);
}
GenerateVoiceMix(channelResource.Mix.ToSpan(),
channelResource.PreviousMix.ToSpan(),
dspStateMemory,
mix.BufferOffset,
mix.BufferCount,
_rendererContext.MixBufferCount + (uint)channelIndex,
nodeId);
if (performanceInitialized)
{
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.End, nodeId);
}
channelResource.UpdateState();
}
for (int i = 0; i < voiceState.BiquadFilterNeedInitialization.Length; i++)
{
voiceState.BiquadFilterNeedInitialization[i] = voiceState.BiquadFilters[i].Enable;
}
}
}
}
public void GenerateVoices()
{
for (int i = 0; i < _voiceContext.GetCount(); i++)
{
ref VoiceState sortedState = ref _voiceContext.GetSortedState(i);
if (!sortedState.ShouldSkip() && sortedState.UpdateForCommandGeneration(_voiceContext))
{
int nodeId = sortedState.NodeId;
PerformanceEntryAddresses performanceEntry = new PerformanceEntryAddresses();
bool performanceInitialized = false;
if (_performanceManager != null && _performanceManager.GetNextEntry(out performanceEntry, PerformanceEntryType.Voice, nodeId))
{
performanceInitialized = true;
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.Start, nodeId);
}
GenerateVoice(ref sortedState);
if (performanceInitialized)
{
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.End, nodeId);
}
}
}
_splitterContext.UpdateInternalState();
}
public void GeneratePerformance(ref PerformanceEntryAddresses performanceEntryAddresses, PerformanceCommand.Type type, int nodeId)
{
_commandBuffer.GeneratePerformance(ref performanceEntryAddresses, type, nodeId);
}
private void GenerateBufferMixerEffect(int bufferOffset, BufferMixEffect effect, int nodeId)
{
Debug.Assert(effect.Type == EffectType.BufferMix);
if (effect.IsEnabled)
{
for (int i = 0; i < effect.Parameter.MixesCount; i++)
{
if (effect.Parameter.Volumes[i] != 0.0f)
{
_commandBuffer.GenerateMix((uint)bufferOffset + effect.Parameter.Input[i],
(uint)bufferOffset + effect.Parameter.Output[i],
nodeId,
effect.Parameter.Volumes[i]);
}
}
}
}
private void GenerateAuxEffect(uint bufferOffset, AuxiliaryBufferEffect effect, int nodeId)
{
Debug.Assert(effect.Type == EffectType.AuxiliaryBuffer);
if (effect.IsEnabled)
{
effect.GetWorkBuffer(0);
effect.GetWorkBuffer(1);
}
if (effect.State.SendBufferInfoBase != 0 && effect.State.ReturnBufferInfoBase != 0)
{
int i = 0;
uint writeOffset = 0;
for (uint channelIndex = effect.Parameter.ChannelCount; channelIndex != 0; channelIndex--)
{
uint newUpdateCount = writeOffset + _commandBuffer.CommandList.SampleCount;
uint updateCount;
amadeus: Update to REV10 (#2654) * amadeus: Update to REV10 This implements all the changes made with REV10 on 13.0.0. * Address Ack's comment * Address gdkchan's comment
2021-09-19 10:29:19 +00:00
if (channelIndex != 1)
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
{
updateCount = 0;
}
else
{
updateCount = newUpdateCount;
}
_commandBuffer.GenerateAuxEffect(bufferOffset,
effect.Parameter.Input[i],
effect.Parameter.Output[i],
ref effect.State,
effect.IsEnabled,
effect.Parameter.BufferStorageSize,
effect.State.SendBufferInfoBase,
effect.State.ReturnBufferInfoBase,
updateCount,
writeOffset,
nodeId);
writeOffset = newUpdateCount;
i++;
}
}
}
amadeus: Update to REV11 (#3230) This should implement all ABI changes from REV11 on 14.0.0 As Nintendo changed the channel disposition for "legacy" effects (Delay, Reverb and Reverb 3D) to match the standard channel mapping, I took the liberty to just remap to the old disposition for now. The proper changes will be handled at a later date with a complete rewriting of those 3 effects to be more readable (see https://github.com/Ryujinx/Ryujinx/pull/3205 for the first iteration of it).
2022-04-06 07:12:38 +00:00
private void GenerateDelayEffect(uint bufferOffset, DelayEffect effect, int nodeId, bool newEffectChannelMappingSupported)
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
{
Debug.Assert(effect.Type == EffectType.Delay);
ulong workBuffer = effect.GetWorkBuffer(-1);
amadeus: Update to REV11 (#3230) This should implement all ABI changes from REV11 on 14.0.0 As Nintendo changed the channel disposition for "legacy" effects (Delay, Reverb and Reverb 3D) to match the standard channel mapping, I took the liberty to just remap to the old disposition for now. The proper changes will be handled at a later date with a complete rewriting of those 3 effects to be more readable (see https://github.com/Ryujinx/Ryujinx/pull/3205 for the first iteration of it).
2022-04-06 07:12:38 +00:00
_commandBuffer.GenerateDelayEffect(bufferOffset, effect.Parameter, effect.State, effect.IsEnabled, workBuffer, nodeId, newEffectChannelMappingSupported);
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
}
amadeus: Update to REV11 (#3230) This should implement all ABI changes from REV11 on 14.0.0 As Nintendo changed the channel disposition for "legacy" effects (Delay, Reverb and Reverb 3D) to match the standard channel mapping, I took the liberty to just remap to the old disposition for now. The proper changes will be handled at a later date with a complete rewriting of those 3 effects to be more readable (see https://github.com/Ryujinx/Ryujinx/pull/3205 for the first iteration of it).
2022-04-06 07:12:38 +00:00
private void GenerateReverbEffect(uint bufferOffset, ReverbEffect effect, int nodeId, bool isLongSizePreDelaySupported, bool newEffectChannelMappingSupported)
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
{
Debug.Assert(effect.Type == EffectType.Reverb);
ulong workBuffer = effect.GetWorkBuffer(-1);
amadeus: Update to REV11 (#3230) This should implement all ABI changes from REV11 on 14.0.0 As Nintendo changed the channel disposition for "legacy" effects (Delay, Reverb and Reverb 3D) to match the standard channel mapping, I took the liberty to just remap to the old disposition for now. The proper changes will be handled at a later date with a complete rewriting of those 3 effects to be more readable (see https://github.com/Ryujinx/Ryujinx/pull/3205 for the first iteration of it).
2022-04-06 07:12:38 +00:00
_commandBuffer.GenerateReverbEffect(bufferOffset, effect.Parameter, effect.State, effect.IsEnabled, workBuffer, nodeId, isLongSizePreDelaySupported, newEffectChannelMappingSupported);
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
}
amadeus: Update to REV11 (#3230) This should implement all ABI changes from REV11 on 14.0.0 As Nintendo changed the channel disposition for "legacy" effects (Delay, Reverb and Reverb 3D) to match the standard channel mapping, I took the liberty to just remap to the old disposition for now. The proper changes will be handled at a later date with a complete rewriting of those 3 effects to be more readable (see https://github.com/Ryujinx/Ryujinx/pull/3205 for the first iteration of it).
2022-04-06 07:12:38 +00:00
private void GenerateReverb3dEffect(uint bufferOffset, Reverb3dEffect effect, int nodeId, bool newEffectChannelMappingSupported)
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
{
Debug.Assert(effect.Type == EffectType.Reverb3d);
ulong workBuffer = effect.GetWorkBuffer(-1);
amadeus: Update to REV11 (#3230) This should implement all ABI changes from REV11 on 14.0.0 As Nintendo changed the channel disposition for "legacy" effects (Delay, Reverb and Reverb 3D) to match the standard channel mapping, I took the liberty to just remap to the old disposition for now. The proper changes will be handled at a later date with a complete rewriting of those 3 effects to be more readable (see https://github.com/Ryujinx/Ryujinx/pull/3205 for the first iteration of it).
2022-04-06 07:12:38 +00:00
_commandBuffer.GenerateReverb3dEffect(bufferOffset, effect.Parameter, effect.State, effect.IsEnabled, workBuffer, nodeId, newEffectChannelMappingSupported);
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
}
private void GenerateBiquadFilterEffect(uint bufferOffset, BiquadFilterEffect effect, int nodeId)
{
Debug.Assert(effect.Type == EffectType.BiquadFilter);
if (effect.IsEnabled)
{
bool needInitialization = effect.Parameter.Status == UsageState.Invalid ||
(effect.Parameter.Status == UsageState.New && !_rendererContext.BehaviourContext.IsBiquadFilterEffectStateClearBugFixed());
BiquadFilterParameter parameter = new BiquadFilterParameter();
parameter.Enable = true;
effect.Parameter.Denominator.ToSpan().CopyTo(parameter.Denominator.ToSpan());
effect.Parameter.Numerator.ToSpan().CopyTo(parameter.Numerator.ToSpan());
for (int i = 0; i < effect.Parameter.ChannelCount; i++)
{
_commandBuffer.GenerateBiquadFilter((int)bufferOffset, ref parameter, effect.State.Slice(i, 1),
effect.Parameter.Input[i],
effect.Parameter.Output[i],
needInitialization,
nodeId);
}
}
else
{
for (int i = 0; i < effect.Parameter.ChannelCount; i++)
{
uint inputBufferIndex = bufferOffset + effect.Parameter.Input[i];
uint outputBufferIndex = bufferOffset + effect.Parameter.Output[i];
// If the input and output isn't the same, generate a command.
if (inputBufferIndex != outputBufferIndex)
{
_commandBuffer.GenerateCopyMixBuffer(inputBufferIndex, outputBufferIndex, nodeId);
}
}
}
}
amadeus: Update to REV9 (#2309) * amadeus: Update to REV9 This implements all the changes made with REV9 on 12.0.0. * Address Ac_k's comments
2021-05-25 17:01:09 +00:00
private void GenerateLimiterEffect(uint bufferOffset, LimiterEffect effect, int nodeId, int effectId)
{
Debug.Assert(effect.Type == EffectType.Limiter);
ulong workBuffer = effect.GetWorkBuffer(-1);
if (_rendererContext.BehaviourContext.IsEffectInfoVersion2Supported())
{
amadeus: Fix limiter correctness (#3126) This fixes missing audio on Nintendo Switch Sports Online Play Test.
2022-02-16 20:38:45 +00:00
Memory<EffectResultState> dspResultState;
if (effect.Parameter.StatisticsEnabled)
{
dspResultState = _effectContext.GetDspStateMemory(effectId);
}
else
{
dspResultState = Memory<EffectResultState>.Empty;
}
amadeus: Update to REV9 (#2309) * amadeus: Update to REV9 This implements all the changes made with REV9 on 12.0.0. * Address Ac_k's comments
2021-05-25 17:01:09 +00:00
_commandBuffer.GenerateLimiterEffectVersion2(bufferOffset, effect.Parameter, effect.State, dspResultState, effect.IsEnabled, workBuffer, nodeId);
}
else
{
_commandBuffer.GenerateLimiterEffectVersion1(bufferOffset, effect.Parameter, effect.State, effect.IsEnabled, workBuffer, nodeId);
}
}
amadeus: Update to REV10 (#2654) * amadeus: Update to REV10 This implements all the changes made with REV10 on 13.0.0. * Address Ack's comment * Address gdkchan's comment
2021-09-19 10:29:19 +00:00
private void GenerateCaptureEffect(uint bufferOffset, CaptureBufferEffect effect, int nodeId)
{
Debug.Assert(effect.Type == EffectType.CaptureBuffer);
if (effect.IsEnabled)
{
effect.GetWorkBuffer(0);
}
if (effect.State.SendBufferInfoBase != 0)
{
int i = 0;
uint writeOffset = 0;
for (uint channelIndex = effect.Parameter.ChannelCount; channelIndex != 0; channelIndex--)
{
uint newUpdateCount = writeOffset + _commandBuffer.CommandList.SampleCount;
uint updateCount;
if (channelIndex != 1)
{
updateCount = 0;
}
else
{
updateCount = newUpdateCount;
}
_commandBuffer.GenerateCaptureEffect(bufferOffset,
effect.Parameter.Input[i],
effect.State.SendBufferInfo,
effect.IsEnabled,
effect.Parameter.BufferStorageSize,
effect.State.SendBufferInfoBase,
updateCount,
writeOffset,
nodeId);
writeOffset = newUpdateCount;
i++;
}
}
}
amadeus: Update to REV9 (#2309) * amadeus: Update to REV9 This implements all the changes made with REV9 on 12.0.0. * Address Ac_k's comments
2021-05-25 17:01:09 +00:00
private void GenerateEffect(ref MixState mix, int effectId, BaseEffect effect)
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
{
int nodeId = mix.NodeId;
Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
bool isFinalMix = mix.MixId == Constants.FinalMixId;
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
PerformanceEntryAddresses performanceEntry = new PerformanceEntryAddresses();
bool performanceInitialized = false;
if (_performanceManager != null && _performanceManager.GetNextEntry(out performanceEntry, effect.GetPerformanceDetailType(),
isFinalMix ? PerformanceEntryType.FinalMix : PerformanceEntryType.SubMix, nodeId))
{
performanceInitialized = true;
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.Start, nodeId);
}
switch (effect.Type)
{
case EffectType.BufferMix:
GenerateBufferMixerEffect((int)mix.BufferOffset, (BufferMixEffect)effect, nodeId);
break;
case EffectType.AuxiliaryBuffer:
GenerateAuxEffect(mix.BufferOffset, (AuxiliaryBufferEffect)effect, nodeId);
break;
case EffectType.Delay:
amadeus: Update to REV11 (#3230) This should implement all ABI changes from REV11 on 14.0.0 As Nintendo changed the channel disposition for "legacy" effects (Delay, Reverb and Reverb 3D) to match the standard channel mapping, I took the liberty to just remap to the old disposition for now. The proper changes will be handled at a later date with a complete rewriting of those 3 effects to be more readable (see https://github.com/Ryujinx/Ryujinx/pull/3205 for the first iteration of it).
2022-04-06 07:12:38 +00:00
GenerateDelayEffect(mix.BufferOffset, (DelayEffect)effect, nodeId, _rendererContext.BehaviourContext.IsNewEffectChannelMappingSupported());
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
break;
case EffectType.Reverb:
amadeus: Update to REV11 (#3230) This should implement all ABI changes from REV11 on 14.0.0 As Nintendo changed the channel disposition for "legacy" effects (Delay, Reverb and Reverb 3D) to match the standard channel mapping, I took the liberty to just remap to the old disposition for now. The proper changes will be handled at a later date with a complete rewriting of those 3 effects to be more readable (see https://github.com/Ryujinx/Ryujinx/pull/3205 for the first iteration of it).
2022-04-06 07:12:38 +00:00
GenerateReverbEffect(mix.BufferOffset, (ReverbEffect)effect, nodeId, mix.IsLongSizePreDelaySupported, _rendererContext.BehaviourContext.IsNewEffectChannelMappingSupported());
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
break;
case EffectType.Reverb3d:
amadeus: Update to REV11 (#3230) This should implement all ABI changes from REV11 on 14.0.0 As Nintendo changed the channel disposition for "legacy" effects (Delay, Reverb and Reverb 3D) to match the standard channel mapping, I took the liberty to just remap to the old disposition for now. The proper changes will be handled at a later date with a complete rewriting of those 3 effects to be more readable (see https://github.com/Ryujinx/Ryujinx/pull/3205 for the first iteration of it).
2022-04-06 07:12:38 +00:00
GenerateReverb3dEffect(mix.BufferOffset, (Reverb3dEffect)effect, nodeId, _rendererContext.BehaviourContext.IsNewEffectChannelMappingSupported());
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
break;
case EffectType.BiquadFilter:
GenerateBiquadFilterEffect(mix.BufferOffset, (BiquadFilterEffect)effect, nodeId);
break;
amadeus: Update to REV9 (#2309) * amadeus: Update to REV9 This implements all the changes made with REV9 on 12.0.0. * Address Ac_k's comments
2021-05-25 17:01:09 +00:00
case EffectType.Limiter:
GenerateLimiterEffect(mix.BufferOffset, (LimiterEffect)effect, nodeId, effectId);
break;
amadeus: Update to REV10 (#2654) * amadeus: Update to REV10 This implements all the changes made with REV10 on 13.0.0. * Address Ack's comment * Address gdkchan's comment
2021-09-19 10:29:19 +00:00
case EffectType.CaptureBuffer:
GenerateCaptureEffect(mix.BufferOffset, (CaptureBufferEffect)effect, nodeId);
break;
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
default:
throw new NotImplementedException($"Unsupported effect type {effect.Type}");
}
if (performanceInitialized)
{
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.End, nodeId);
}
effect.UpdateForCommandGeneration();
}
private void GenerateEffects(ref MixState mix)
{
ReadOnlySpan<int> effectProcessingOrderArray = mix.EffectProcessingOrderArray;
Debug.Assert(_effectContext.GetCount() == 0 || !effectProcessingOrderArray.IsEmpty);
for (int i = 0; i < _effectContext.GetCount(); i++)
{
int effectOrder = effectProcessingOrderArray[i];
Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
if (effectOrder == Constants.InvalidProcessingOrder)
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
{
break;
}
// BaseEffect is a class, we don't need to pass it by ref
BaseEffect effect = _effectContext.GetEffect(effectOrder);
Debug.Assert(effect.Type != EffectType.Invalid);
Debug.Assert(effect.MixId == mix.MixId);
if (!effect.ShouldSkip())
{
amadeus: Update to REV9 (#2309) * amadeus: Update to REV9 This implements all the changes made with REV9 on 12.0.0. * Address Ac_k's comments
2021-05-25 17:01:09 +00:00
GenerateEffect(ref mix, effectOrder, effect);
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
}
}
}
private void GenerateMix(ref MixState mix)
{
if (mix.HasAnyDestination())
{
Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
Debug.Assert(mix.DestinationMixId != Constants.UnusedMixId || mix.DestinationSplitterId != Constants.UnusedSplitterId);
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
if (mix.DestinationMixId == Constants.UnusedMixId)
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
{
Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
if (mix.DestinationSplitterId != Constants.UnusedSplitterId)
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
{
int destinationId = 0;
while (true)
{
int destinationIndex = destinationId++;
Span<SplitterDestination> destinationSpan = _splitterContext.GetDestination((int)mix.DestinationSplitterId, destinationIndex);
if (destinationSpan.IsEmpty)
{
break;
}
ref SplitterDestination destination = ref destinationSpan[0];
if (destination.IsConfigured())
{
int mixId = destination.DestinationId;
Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
if (mixId < _mixContext.GetCount() && mixId != Constants.UnusedSplitterIdInt)
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
{
ref MixState destinationMix = ref _mixContext.GetState(mixId);
uint inputBufferIndex = mix.BufferOffset + ((uint)destinationIndex % mix.BufferCount);
for (uint bufferDestinationIndex = 0; bufferDestinationIndex < destinationMix.BufferCount; bufferDestinationIndex++)
{
float volume = mix.Volume * destination.GetMixVolume((int)bufferDestinationIndex);
if (volume != 0.0f)
{
_commandBuffer.GenerateMix(inputBufferIndex,
destinationMix.BufferOffset + bufferDestinationIndex,
mix.NodeId,
volume);
}
}
}
}
}
}
}
else
{
ref MixState destinationMix = ref _mixContext.GetState(mix.DestinationMixId);
for (uint bufferIndex = 0; bufferIndex < mix.BufferCount; bufferIndex++)
{
for (uint bufferDestinationIndex = 0; bufferDestinationIndex < destinationMix.BufferCount; bufferDestinationIndex++)
{
float volume = mix.Volume * mix.GetMixBufferVolume((int)bufferIndex, (int)bufferDestinationIndex);
if (volume != 0.0f)
{
_commandBuffer.GenerateMix(mix.BufferOffset + bufferIndex,
destinationMix.BufferOffset + bufferDestinationIndex,
mix.NodeId,
volume);
}
}
}
}
}
}
private void GenerateSubMix(ref MixState subMix)
{
_commandBuffer.GenerateDepopForMixBuffersCommand(_rendererContext.DepopBuffer,
subMix.BufferOffset,
subMix.BufferCount,
subMix.NodeId,
subMix.SampleRate);
GenerateEffects(ref subMix);
PerformanceEntryAddresses performanceEntry = new PerformanceEntryAddresses();
int nodeId = subMix.NodeId;
bool performanceInitialized = false;
if (_performanceManager != null && _performanceManager.IsTargetNodeId(nodeId) && _performanceManager.GetNextEntry(out performanceEntry, PerformanceDetailType.Mix, PerformanceEntryType.SubMix, nodeId))
{
performanceInitialized = true;
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.Start, nodeId);
}
GenerateMix(ref subMix);
if (performanceInitialized)
{
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.End, nodeId);
}
}
public void GenerateSubMixes()
{
for (int id = 0; id < _mixContext.GetCount(); id++)
{
ref MixState sortedState = ref _mixContext.GetSortedState(id);
Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
if (sortedState.IsUsed && sortedState.MixId != Constants.FinalMixId)
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
{
int nodeId = sortedState.NodeId;
PerformanceEntryAddresses performanceEntry = new PerformanceEntryAddresses();
bool performanceInitialized = false;
if (_performanceManager != null && _performanceManager.GetNextEntry(out performanceEntry, PerformanceEntryType.SubMix, nodeId))
{
performanceInitialized = true;
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.Start, nodeId);
}
GenerateSubMix(ref sortedState);
if (performanceInitialized)
{
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.End, nodeId);
}
}
}
}
private void GenerateFinalMix()
{
ref MixState finalMix = ref _mixContext.GetFinalState();
_commandBuffer.GenerateDepopForMixBuffersCommand(_rendererContext.DepopBuffer,
finalMix.BufferOffset,
finalMix.BufferCount,
finalMix.NodeId,
finalMix.SampleRate);
GenerateEffects(ref finalMix);
PerformanceEntryAddresses performanceEntry = new PerformanceEntryAddresses();
int nodeId = finalMix.NodeId;
bool performanceInitialized = false;
if (_performanceManager != null && _performanceManager.IsTargetNodeId(nodeId) && _performanceManager.GetNextEntry(out performanceEntry, PerformanceDetailType.Mix, PerformanceEntryType.FinalMix, nodeId))
{
performanceInitialized = true;
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.Start, nodeId);
}
// Only generate volume command if the volume isn't 100%.
if (finalMix.Volume != 1.0f)
{
for (uint bufferIndex = 0; bufferIndex < finalMix.BufferCount; bufferIndex++)
{
bool performanceSubInitialized = false;
if (_performanceManager != null && _performanceManager.IsTargetNodeId(nodeId) && _performanceManager.GetNextEntry(out performanceEntry, PerformanceDetailType.VolumeRamp, PerformanceEntryType.FinalMix, nodeId))
{
performanceSubInitialized = true;
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.Start, nodeId);
}
_commandBuffer.GenerateVolume(finalMix.Volume,
finalMix.BufferOffset + bufferIndex,
nodeId);
if (performanceSubInitialized)
{
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.End, nodeId);
}
}
}
if (performanceInitialized)
{
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.End, nodeId);
}
}
public void GenerateFinalMixes()
{
int nodeId = _mixContext.GetFinalState().NodeId;
PerformanceEntryAddresses performanceEntry = new PerformanceEntryAddresses();
bool performanceInitialized = false;
if (_performanceManager != null && _performanceManager.GetNextEntry(out performanceEntry, PerformanceEntryType.FinalMix, nodeId))
{
performanceInitialized = true;
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.Start, nodeId);
}
GenerateFinalMix();
if (performanceInitialized)
{
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.End, nodeId);
}
}
private void GenerateCircularBuffer(CircularBufferSink sink, ref MixState finalMix)
{
Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
_commandBuffer.GenerateCircularBuffer(finalMix.BufferOffset, sink, Constants.InvalidNodeId);
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
}
private void GenerateDevice(DeviceSink sink, ref MixState finalMix)
{
if (_commandBuffer.CommandList.SampleRate != 48000 && sink.UpsamplerState == null)
{
sink.UpsamplerState = _rendererContext.UpsamplerManager.Allocate();
}
bool useCustomDownMixingCommand = _rendererContext.ChannelCount == 2 && sink.Parameter.DownMixParameterEnabled;
if (useCustomDownMixingCommand)
{
_commandBuffer.GenerateDownMixSurroundToStereo(finalMix.BufferOffset,
sink.Parameter.Input.ToSpan(),
sink.Parameter.Input.ToSpan(),
sink.DownMixCoefficients,
Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
Constants.InvalidNodeId);
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
}
Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
// NOTE: We do the downmixing at the DSP level as it's easier that way.
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
else if (_rendererContext.ChannelCount == 2 && sink.Parameter.InputCount == 6)
{
_commandBuffer.GenerateDownMixSurroundToStereo(finalMix.BufferOffset,
sink.Parameter.Input.ToSpan(),
sink.Parameter.Input.ToSpan(),
Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
Constants.DefaultSurroundToStereoCoefficients,
Constants.InvalidNodeId);
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
}
CommandList commandList = _commandBuffer.CommandList;
if (sink.UpsamplerState != null)
{
_commandBuffer.GenerateUpsample(finalMix.BufferOffset,
sink.UpsamplerState,
sink.Parameter.InputCount,
sink.Parameter.Input.ToSpan(),
commandList.BufferCount,
commandList.SampleCount,
commandList.SampleRate,
Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
Constants.InvalidNodeId);
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
}
_commandBuffer.GenerateDeviceSink(finalMix.BufferOffset,
sink,
_rendererContext.SessionId,
commandList.Buffers,
Haydn: Part 1 (#2007) * Haydn: Part 1 Based on my reverse of audio 11.0.0. As always, core implementation under LGPLv3 for the same reasons as for Amadeus. This place the bases of a more flexible audio system while making audout & audin accurate. This have the following improvements: - Complete reimplementation of audout and audin. - Audin currently only have a dummy backend. - Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL). - Audio Renderer now can output to 5.1 devices when supported. - Audio Renderer init its backend on demand instead of keeping two up all the time. - All backends implementation are now in their own project. - Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this. As a note, games having issues with OpenAL haven't improved and will not because of OpenAL design (stopping when buffers finish playing causing possible audio "pops" when buffers are very small). * Update for latest hexkyz's edits on Switchbrew * audren: Rollback channel configuration changes * Address gdkchan's comments * Fix typo in OpenAL backend driver * Address last comments * Fix a nit * Address gdkchan's comments
2021-02-26 00:11:56 +00:00
Constants.InvalidNodeId);
Amadeus: Final Act (#1481) * Amadeus: Final Act This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer! This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months. This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware. Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3. - FE3H voices in videos and chapter intro are not present. - Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time). - Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed. * Make timing more precise when sleeping on Windows Improve precision to a 1ms resolution on Windows NT based OS. This is used to avoid having totally erratic timings and unify all Windows users to the same resolution. NOTE: This is only active when emulation is running.
2020-08-18 01:49:37 +00:00
}
private void GenerateSink(BaseSink sink, ref MixState finalMix)
{
bool performanceInitialized = false;
PerformanceEntryAddresses performanceEntry = new PerformanceEntryAddresses();
if (_performanceManager != null && _performanceManager.GetNextEntry(out performanceEntry, PerformanceEntryType.Sink, sink.NodeId))
{
performanceInitialized = true;
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.Start, sink.NodeId);
}
if (!sink.ShouldSkip)
{
switch (sink.Type)
{
case SinkType.CircularBuffer:
GenerateCircularBuffer((CircularBufferSink)sink, ref finalMix);
break;
case SinkType.Device:
GenerateDevice((DeviceSink)sink, ref finalMix);
break;
default:
throw new NotImplementedException($"Unsupported sink type {sink.Type}");
}
sink.UpdateForCommandGeneration();
}
if (performanceInitialized)
{
GeneratePerformance(ref performanceEntry, PerformanceCommand.Type.End, sink.NodeId);
}
}
public void GenerateSinks()
{
ref MixState finalMix = ref _mixContext.GetFinalState();
for (int i = 0; i < _sinkContext.GetCount(); i++)
{
// BaseSink is a class, we don't need to pass it by ref
BaseSink sink = _sinkContext.GetSink(i);
if (sink.IsUsed)
{
GenerateSink(sink, ref finalMix);
}
}
}
}
misc: Reformat Ryujinx.Audio with dotnet-format (#3485) This is the first commit of a series of reformat around the codebase as discussed internally some weeks ago. This project being one that isn't touched that much, it shouldn't cause conflict with any opened PRs.
2022-07-25 18:46:33 +00:00
}
Reference in a new issue Copy permalink