dolphin/Source/Core/VideoCommon/MainBase.cpp

357 lines
7.8 KiB
C++
Raw Normal View History

#include "Common/Event.h"
#include "Core/ConfigManager.h"
#include "VideoCommon/BoundingBox.h"
#include "VideoCommon/BPStructs.h"
#include "VideoCommon/CommandProcessor.h"
#include "VideoCommon/Fifo.h"
#include "VideoCommon/FramebufferManagerBase.h"
#include "VideoCommon/MainBase.h"
#include "VideoCommon/OnScreenDisplay.h"
#include "VideoCommon/PixelEngine.h"
#include "VideoCommon/RenderBase.h"
#include "VideoCommon/TextureCacheBase.h"
#include "VideoCommon/VertexLoaderManager.h"
#include "VideoCommon/VideoBackendBase.h"
#include "VideoCommon/VideoConfig.h"
#include "VideoCommon/VideoState.h"
bool s_BackendInitialized = false;
Common::Flag s_swapRequested;
static Common::Flag s_FifoShuttingDown;
static Common::Flag s_efbAccessRequested;
static Common::Event s_efbAccessReadyEvent;
static Common::Flag s_perfQueryRequested;
static Common::Event s_perfQueryReadyEvent;
static Common::Flag s_BBoxRequested;
static Common::Event s_BBoxReadyEvent;
static int s_BBoxIndex;
static u16 s_BBoxResult;
static volatile struct
{
u32 xfbAddr;
u32 fbWidth;
2014-05-02 01:08:44 -06:00
u32 fbStride;
u32 fbHeight;
} s_beginFieldArgs;
static struct
{
EFBAccessType type;
u32 x;
u32 y;
u32 Data;
} s_accessEFBArgs;
static u32 s_AccessEFBResult = 0;
void VideoBackendHardware::EmuStateChange(EMUSTATE_CHANGE newState)
{
EmulatorState((newState == EMUSTATE_CHANGE_PLAY) ? true : false);
}
// Enter and exit the video loop
void VideoBackendHardware::Video_EnterLoop()
{
RunGpuLoop();
}
void VideoBackendHardware::Video_ExitLoop()
{
ExitGpuLoop();
s_FifoShuttingDown.Set();
s_efbAccessReadyEvent.Set();
s_perfQueryReadyEvent.Set();
}
void VideoBackendHardware::Video_SetRendering(bool bEnabled)
{
Fifo_SetRendering(bEnabled);
}
// Run from the graphics thread (from Fifo.cpp)
2014-07-08 06:29:26 -06:00
static void VideoFifo_CheckSwapRequest()
{
if (g_ActiveConfig.bUseXFB)
{
if (s_swapRequested.IsSet())
{
EFBRectangle rc;
2014-05-02 01:08:44 -06:00
Renderer::Swap(s_beginFieldArgs.xfbAddr, s_beginFieldArgs.fbWidth, s_beginFieldArgs.fbStride, s_beginFieldArgs.fbHeight, rc);
s_swapRequested.Clear();
}
}
}
// Run from the graphics thread (from Fifo.cpp)
void VideoFifo_CheckSwapRequestAt(u32 xfbAddr, u32 fbWidth, u32 fbHeight)
{
if (g_ActiveConfig.bUseXFB)
{
if (s_swapRequested.IsSet())
{
u32 aLower = xfbAddr;
u32 aUpper = xfbAddr + 2 * fbWidth * fbHeight;
u32 bLower = s_beginFieldArgs.xfbAddr;
2014-05-02 01:08:44 -06:00
u32 bUpper = s_beginFieldArgs.xfbAddr + 2 * s_beginFieldArgs.fbStride * s_beginFieldArgs.fbHeight;
if (AddressRangesOverlap(aLower, aUpper, bLower, bUpper))
VideoFifo_CheckSwapRequest();
}
}
}
// Run from the CPU thread (from VideoInterface.cpp)
2014-05-02 01:08:44 -06:00
void VideoBackendHardware::Video_BeginField(u32 xfbAddr, u32 fbWidth, u32 fbStride, u32 fbHeight)
{
if (s_BackendInitialized && g_ActiveConfig.bUseXFB)
{
if (!SConfig::GetInstance().m_LocalCoreStartupParameter.bCPUThread)
VideoFifo_CheckSwapRequest();
s_beginFieldArgs.xfbAddr = xfbAddr;
s_beginFieldArgs.fbWidth = fbWidth;
2014-05-02 01:08:44 -06:00
s_beginFieldArgs.fbStride = fbStride;
s_beginFieldArgs.fbHeight = fbHeight;
}
}
// Run from the CPU thread (from VideoInterface.cpp)
void VideoBackendHardware::Video_EndField()
{
if (s_BackendInitialized)
{
Add the 'desynced GPU thread' mode. It's a relatively big commit (less big with -w), but it's hard to test any of this separately... The basic problem is that in netplay or movies, the state of the CPU must be deterministic, including when the game receives notification that the GPU has processed FIFO data. Dual core mode notifies the game whenever the GPU thread actually gets around to doing the work, so it isn't deterministic. Single core mode is because it notifies the game 'instantly' (after processing the data synchronously), but it's too slow for many systems and games. My old dc-netplay branch worked as follows: everything worked as normal except the state of the CP registers was a lie, and the CPU thread only delivered results when idle detection triggered (waiting for the GPU if they weren't ready at that point). Usually, a game is idle iff all the work for the frame has been done, except for a small amount of work depending on the GPU result, so neither the CPU or the GPU waiting on the other affected performance much. However, it's possible that the game could be waiting for some earlier interrupt, and any of several games which, for whatever reason, never went into a detectable idle (even when I tried to improve the detection) would never receive results at all. (The current method should have better compatibility, but it also has slightly higher overhead and breaks some other things, so I want to reimplement this, hopefully with less impact on the code, in the future.) With this commit, the basic idea is that the CPU thread acts as if the work has been done instantly, like single core mode, but actually hands it off asynchronously to the GPU thread (after backing up some data that the game might change in memory before it's actually done). Since the work isn't done, any feedback from the GPU to the CPU, such as real XFB/EFB copies (virtual are OK), EFB pokes, performance queries, etc. is broken; but most games work with these options disabled, and there is no need to try to detect what the CPU thread is doing. Technically: when the flag g_use_deterministic_gpu_thread (currently stuck on) is on, the CPU thread calls RunGpu like in single core mode. This function synchronously copies the data from the FIFO to the internal video buffer and updates the CP registers, interrupts, etc. However, instead of the regular ReadDataFromFifo followed by running the opcode decoder, it runs ReadDataFromFifoOnCPU -> OpcodeDecoder_Preprocess, which relatively quickly scans through the FIFO data, detects SetFinish calls etc., which are immediately fired, and saves certain associated data from memory (e.g. display lists) in AuxBuffers (a parallel stream to the main FIFO, which is a bit slow at the moment), before handing the data off to the GPU thread to actually render. That makes up the bulk of this commit. In various circumstances, including the aforementioned EFB pokes and performance queries as well as swap requests (i.e. the end of a frame - we don't want the CPU potentially pumping out frames too quickly and the GPU falling behind*), SyncGPU is called to wait for actual completion. The overhead mainly comes from OpcodeDecoder_Preprocess (which is, again, synchronous), as well as the actual copying. Currently, display lists and such are escrowed from main memory even though they usually won't change over the course of a frame, and textures are not even though they might, resulting in a small chance of graphical glitches. When the texture locking (i.e. fault on write) code lands, I can make this all correct and maybe a little faster. * This suggests an alternate determinism method of just delaying results until a short time before the end of each frame. For all I know this might mostly work - I haven't tried it - but if any significant work hinges on the competion of render to texture etc., the frame will be missed.
2014-08-27 20:56:19 -06:00
SyncGPU(SYNC_GPU_SWAP);
s_swapRequested.Set();
}
}
void VideoBackendHardware::Video_AddMessage(const std::string& msg, u32 milliseconds)
{
OSD::AddMessage(msg, milliseconds);
}
void VideoBackendHardware::Video_ClearMessages()
{
OSD::ClearMessages();
}
// Screenshot
bool VideoBackendHardware::Video_Screenshot(const std::string& filename)
{
Renderer::SetScreenshot(filename.c_str());
return true;
}
void VideoFifo_CheckEFBAccess()
{
if (s_efbAccessRequested.IsSet())
{
s_AccessEFBResult = g_renderer->AccessEFB(s_accessEFBArgs.type, s_accessEFBArgs.x, s_accessEFBArgs.y, s_accessEFBArgs.Data);
s_efbAccessRequested.Clear();
s_efbAccessReadyEvent.Set();
}
}
u32 VideoBackendHardware::Video_AccessEFB(EFBAccessType type, u32 x, u32 y, u32 InputData)
{
if (s_BackendInitialized && g_ActiveConfig.bEFBAccessEnable)
{
Add the 'desynced GPU thread' mode. It's a relatively big commit (less big with -w), but it's hard to test any of this separately... The basic problem is that in netplay or movies, the state of the CPU must be deterministic, including when the game receives notification that the GPU has processed FIFO data. Dual core mode notifies the game whenever the GPU thread actually gets around to doing the work, so it isn't deterministic. Single core mode is because it notifies the game 'instantly' (after processing the data synchronously), but it's too slow for many systems and games. My old dc-netplay branch worked as follows: everything worked as normal except the state of the CP registers was a lie, and the CPU thread only delivered results when idle detection triggered (waiting for the GPU if they weren't ready at that point). Usually, a game is idle iff all the work for the frame has been done, except for a small amount of work depending on the GPU result, so neither the CPU or the GPU waiting on the other affected performance much. However, it's possible that the game could be waiting for some earlier interrupt, and any of several games which, for whatever reason, never went into a detectable idle (even when I tried to improve the detection) would never receive results at all. (The current method should have better compatibility, but it also has slightly higher overhead and breaks some other things, so I want to reimplement this, hopefully with less impact on the code, in the future.) With this commit, the basic idea is that the CPU thread acts as if the work has been done instantly, like single core mode, but actually hands it off asynchronously to the GPU thread (after backing up some data that the game might change in memory before it's actually done). Since the work isn't done, any feedback from the GPU to the CPU, such as real XFB/EFB copies (virtual are OK), EFB pokes, performance queries, etc. is broken; but most games work with these options disabled, and there is no need to try to detect what the CPU thread is doing. Technically: when the flag g_use_deterministic_gpu_thread (currently stuck on) is on, the CPU thread calls RunGpu like in single core mode. This function synchronously copies the data from the FIFO to the internal video buffer and updates the CP registers, interrupts, etc. However, instead of the regular ReadDataFromFifo followed by running the opcode decoder, it runs ReadDataFromFifoOnCPU -> OpcodeDecoder_Preprocess, which relatively quickly scans through the FIFO data, detects SetFinish calls etc., which are immediately fired, and saves certain associated data from memory (e.g. display lists) in AuxBuffers (a parallel stream to the main FIFO, which is a bit slow at the moment), before handing the data off to the GPU thread to actually render. That makes up the bulk of this commit. In various circumstances, including the aforementioned EFB pokes and performance queries as well as swap requests (i.e. the end of a frame - we don't want the CPU potentially pumping out frames too quickly and the GPU falling behind*), SyncGPU is called to wait for actual completion. The overhead mainly comes from OpcodeDecoder_Preprocess (which is, again, synchronous), as well as the actual copying. Currently, display lists and such are escrowed from main memory even though they usually won't change over the course of a frame, and textures are not even though they might, resulting in a small chance of graphical glitches. When the texture locking (i.e. fault on write) code lands, I can make this all correct and maybe a little faster. * This suggests an alternate determinism method of just delaying results until a short time before the end of each frame. For all I know this might mostly work - I haven't tried it - but if any significant work hinges on the competion of render to texture etc., the frame will be missed.
2014-08-27 20:56:19 -06:00
SyncGPU(SYNC_GPU_EFB_POKE);
s_accessEFBArgs.type = type;
s_accessEFBArgs.x = x;
s_accessEFBArgs.y = y;
s_accessEFBArgs.Data = InputData;
s_efbAccessRequested.Set();
if (SConfig::GetInstance().m_LocalCoreStartupParameter.bCPUThread)
{
s_efbAccessReadyEvent.Reset();
if (s_FifoShuttingDown.IsSet())
return 0;
s_efbAccessRequested.Set();
s_efbAccessReadyEvent.Wait();
}
else
VideoFifo_CheckEFBAccess();
return s_AccessEFBResult;
}
return 0;
}
2014-07-08 06:29:26 -06:00
static void VideoFifo_CheckPerfQueryRequest()
{
if (s_perfQueryRequested.IsSet())
{
g_perf_query->FlushResults();
s_perfQueryRequested.Clear();
s_perfQueryReadyEvent.Set();
}
}
u32 VideoBackendHardware::Video_GetQueryResult(PerfQueryType type)
{
if (!g_perf_query->ShouldEmulate())
{
return 0;
}
Add the 'desynced GPU thread' mode. It's a relatively big commit (less big with -w), but it's hard to test any of this separately... The basic problem is that in netplay or movies, the state of the CPU must be deterministic, including when the game receives notification that the GPU has processed FIFO data. Dual core mode notifies the game whenever the GPU thread actually gets around to doing the work, so it isn't deterministic. Single core mode is because it notifies the game 'instantly' (after processing the data synchronously), but it's too slow for many systems and games. My old dc-netplay branch worked as follows: everything worked as normal except the state of the CP registers was a lie, and the CPU thread only delivered results when idle detection triggered (waiting for the GPU if they weren't ready at that point). Usually, a game is idle iff all the work for the frame has been done, except for a small amount of work depending on the GPU result, so neither the CPU or the GPU waiting on the other affected performance much. However, it's possible that the game could be waiting for some earlier interrupt, and any of several games which, for whatever reason, never went into a detectable idle (even when I tried to improve the detection) would never receive results at all. (The current method should have better compatibility, but it also has slightly higher overhead and breaks some other things, so I want to reimplement this, hopefully with less impact on the code, in the future.) With this commit, the basic idea is that the CPU thread acts as if the work has been done instantly, like single core mode, but actually hands it off asynchronously to the GPU thread (after backing up some data that the game might change in memory before it's actually done). Since the work isn't done, any feedback from the GPU to the CPU, such as real XFB/EFB copies (virtual are OK), EFB pokes, performance queries, etc. is broken; but most games work with these options disabled, and there is no need to try to detect what the CPU thread is doing. Technically: when the flag g_use_deterministic_gpu_thread (currently stuck on) is on, the CPU thread calls RunGpu like in single core mode. This function synchronously copies the data from the FIFO to the internal video buffer and updates the CP registers, interrupts, etc. However, instead of the regular ReadDataFromFifo followed by running the opcode decoder, it runs ReadDataFromFifoOnCPU -> OpcodeDecoder_Preprocess, which relatively quickly scans through the FIFO data, detects SetFinish calls etc., which are immediately fired, and saves certain associated data from memory (e.g. display lists) in AuxBuffers (a parallel stream to the main FIFO, which is a bit slow at the moment), before handing the data off to the GPU thread to actually render. That makes up the bulk of this commit. In various circumstances, including the aforementioned EFB pokes and performance queries as well as swap requests (i.e. the end of a frame - we don't want the CPU potentially pumping out frames too quickly and the GPU falling behind*), SyncGPU is called to wait for actual completion. The overhead mainly comes from OpcodeDecoder_Preprocess (which is, again, synchronous), as well as the actual copying. Currently, display lists and such are escrowed from main memory even though they usually won't change over the course of a frame, and textures are not even though they might, resulting in a small chance of graphical glitches. When the texture locking (i.e. fault on write) code lands, I can make this all correct and maybe a little faster. * This suggests an alternate determinism method of just delaying results until a short time before the end of each frame. For all I know this might mostly work - I haven't tried it - but if any significant work hinges on the competion of render to texture etc., the frame will be missed.
2014-08-27 20:56:19 -06:00
SyncGPU(SYNC_GPU_PERFQUERY);
// TODO: Is this check sane?
if (!g_perf_query->IsFlushed())
{
if (SConfig::GetInstance().m_LocalCoreStartupParameter.bCPUThread)
{
s_perfQueryReadyEvent.Reset();
if (s_FifoShuttingDown.IsSet())
return 0;
s_perfQueryRequested.Set();
s_perfQueryReadyEvent.Wait();
}
else
g_perf_query->FlushResults();
}
return g_perf_query->GetQueryResult(type);
}
u16 VideoBackendHardware::Video_GetBoundingBox(int index)
{
if (!g_ActiveConfig.backend_info.bSupportsBBox)
return BoundingBox::coords[index];
SyncGPU(SYNC_GPU_BBOX);
if (SConfig::GetInstance().m_LocalCoreStartupParameter.bCPUThread)
{
s_BBoxReadyEvent.Reset();
if (s_FifoShuttingDown.IsSet())
return 0;
s_BBoxIndex = index;
s_BBoxRequested.Set();
s_BBoxReadyEvent.Wait();
return s_BBoxResult;
}
else
{
return g_renderer->BBoxRead(index);
}
}
static void VideoFifo_CheckBBoxRequest()
{
if (s_BBoxRequested.IsSet())
{
s_BBoxResult = g_renderer->BBoxRead(s_BBoxIndex);
s_BBoxRequested.Clear();
s_BBoxReadyEvent.Set();
}
}
2012-01-02 03:20:22 -07:00
void VideoBackendHardware::InitializeShared()
{
VideoCommon_Init();
s_swapRequested.Clear();
s_efbAccessRequested.Clear();
s_perfQueryRequested.Clear();
s_FifoShuttingDown.Clear();
2012-01-02 03:20:22 -07:00
memset((void*)&s_beginFieldArgs, 0, sizeof(s_beginFieldArgs));
memset(&s_accessEFBArgs, 0, sizeof(s_accessEFBArgs));
s_AccessEFBResult = 0;
m_invalid = false;
2012-01-02 03:20:22 -07:00
}
// Run from the CPU thread
void VideoBackendHardware::DoState(PointerWrap& p)
{
bool software = false;
p.Do(software);
if (p.GetMode() == PointerWrap::MODE_READ && software == true)
{
// change mode to abort load of incompatible save state.
p.SetMode(PointerWrap::MODE_VERIFY);
}
VideoCommon_DoState(p);
p.DoMarker("VideoCommon");
p.Do(s_swapRequested);
p.Do(s_efbAccessRequested);
p.Do(s_beginFieldArgs);
p.Do(s_accessEFBArgs);
p.Do(s_AccessEFBResult);
p.DoMarker("VideoBackendHardware");
// Refresh state.
if (p.GetMode() == PointerWrap::MODE_READ)
{
m_invalid = true;
RecomputeCachedArraybases();
// Clear all caches that touch RAM
// (? these don't appear to touch any emulation state that gets saved. moved to on load only.)
VertexLoaderManager::MarkAllDirty();
}
}
void VideoBackendHardware::CheckInvalidState()
{
if (m_invalid)
{
m_invalid = false;
BPReload();
TextureCache::Invalidate();
}
}
void VideoBackendHardware::PauseAndLock(bool doLock, bool unpauseOnUnlock)
{
Fifo_PauseAndLock(doLock, unpauseOnUnlock);
}
void VideoBackendHardware::RunLoop(bool enable)
{
VideoCommon_RunLoop(enable);
}
void VideoFifo_CheckAsyncRequest()
{
VideoFifo_CheckSwapRequest();
VideoFifo_CheckEFBAccess();
VideoFifo_CheckPerfQueryRequest();
VideoFifo_CheckBBoxRequest();
}
void VideoBackendHardware::Video_GatherPipeBursted()
{
CommandProcessor::GatherPipeBursted();
}
bool VideoBackendHardware::Video_IsPossibleWaitingSetDrawDone()
{
return CommandProcessor::isPossibleWaitingSetDrawDone;
}
void VideoBackendHardware::RegisterCPMMIO(MMIO::Mapping* mmio, u32 base)
{
CommandProcessor::RegisterMMIO(mmio, base);
}
void VideoBackendHardware::UpdateWantDeterminism(bool want)
{
Fifo_UpdateWantDeterminism(want);
}