niansa-misc/dxvk
1
0
Fork 0
mirror of https://github.com/doitsujin/dxvk.git synced 2025-03-06 20:58:37 +01:00
Code Issues Projects Releases Packages Wiki Activity

[dxvk] Improve handling of cpu-frametime related stutters for low-latency frame pacing

Browse source 
Stutters less this way because we increase the sensitivity to mark frames as outliers, so that they don't get used for predicting the next frame. The actual "optimal" threshold is still to be fine-tuned, but this one worked really well.
This commit is contained in:
netborg 2025-02-28 08:48:36 +01:00
parent 988dcdc08c
commit fd68a08572
1 changed files with 12 additions and 11 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

23
src/dxvk/framepacer/dxvk_framepacer_mode_low_latency.h
View file

@ -116,9 +116,7 @@ namespace dxvk {
// where gpuSubmit[i] <= gpuRun[i] for all i // where gpuSubmit[i] <= gpuRun[i] for all i
std::vector<int32_t>& gpuRun = m_tempGpuRun; std::vector<int32_t>& gpuRun = m_tempGpuRun;
std::vector<int32_t>& gpuRunDurations = m_tempGpuRunDurations;
gpuRun.clear(); gpuRun.clear();
gpuRunDurations.clear();
int32_t optimizedGpuTime = 0; int32_t optimizedGpuTime = 0;
gpuRun.push_back(optimizedGpuTime); gpuRun.push_back(optimizedGpuTime);
@ -127,7 +125,6 @@ namespace dxvk {
int32_t duration = duration_cast<microseconds>( m->gpuReady[i+1] - _gpuRun ).count(); int32_t duration = duration_cast<microseconds>( m->gpuReady[i+1] - _gpuRun ).count();
optimizedGpuTime += duration; optimizedGpuTime += duration;
gpuRun.push_back(optimizedGpuTime); gpuRun.push_back(optimizedGpuTime);
gpuRunDurations.push_back(duration);
} }
int32_t alignment = duration_cast<microseconds>( m->gpuSubmit[numLoop-1] - m->gpuSubmit[0] ).count() int32_t alignment = duration_cast<microseconds>( m->gpuSubmit[numLoop-1] - m->gpuSubmit[0] ).count()
@ -177,7 +174,7 @@ namespace dxvk {
struct SyncProps { struct SyncProps {
int32_t optimizedGpuTime; // gpu executing packed submits in one go int32_t optimizedGpuTime; // gpu executing packed submits in one go
int32_t gpuSync; // us after gpuStart int32_t gpuSync; // gpuStart to this sync point, in microseconds
int32_t cpuUntilGpuSync; int32_t cpuUntilGpuSync;
int32_t cpuUntilGpuStart; int32_t cpuUntilGpuStart;
int32_t csStart; int32_t csStart;
@ -187,9 +184,14 @@ namespace dxvk {
SyncProps getSyncPrediction() { SyncProps getSyncPrediction() {
// in the future we might use more samples to get a prediction // In the future we might use more samples to get a prediction.
// however, simple averaging gives a slightly artificial mouse input // Possibly this will be optional, as until now, basing it on
// more advanced methods will be investigated // just the previous frame gave us the best mouse input feel.
// Simple averaging or median filtering is surely not the way
// to go, but more advanced methods will be investigated.
// The best place to filter should be on the Present() timeline,
// so not sure if we really will do any filtering here other
// than outlier removal, which will dampen stuttering effects.
SyncProps res = {}; SyncProps res = {};
uint64_t id = m_propsFinished; uint64_t id = m_propsFinished;
if (id < DXGI_MAX_SWAP_CHAIN_BUFFERS+7) if (id < DXGI_MAX_SWAP_CHAIN_BUFFERS+7)
@ -210,14 +212,14 @@ namespace dxvk {
bool isOutlier( uint64_t frameId ) { bool isOutlier( uint64_t frameId ) {
constexpr size_t numLoop = 7; constexpr size_t numLoop = 7;
int32_t totalCpuTime = 0; int32_t totalCpuTime = 0;
for (size_t i=0; i<numLoop; ++i) { for (size_t i=1; i<numLoop; ++i) {
const LatencyMarkers* m = m_latencyMarkersStorage->getConstMarkers(frameId-i); const LatencyMarkers* m = m_latencyMarkersStorage->getConstMarkers(frameId-i);
totalCpuTime += m->cpuFinished; totalCpuTime += m->cpuFinished;
} }
int32_t avgCpuTime = totalCpuTime / numLoop; int32_t avgCpuTime = totalCpuTime / (numLoop-1);
const LatencyMarkers* m = m_latencyMarkersStorage->getConstMarkers(frameId); const LatencyMarkers* m = m_latencyMarkersStorage->getConstMarkers(frameId);
if (m->cpuFinished > 1.7*avgCpuTime || m->gpuSubmit.empty() || m->gpuReady.size() != (m->gpuSubmit.size()+1) ) if (m->cpuFinished > 1.3*avgCpuTime || m->gpuSubmit.empty() || m->gpuReady.size() != (m->gpuSubmit.size()+1) )
return true; return true;
return false; return false;
@ -232,7 +234,6 @@ namespace dxvk {
std::atomic<uint64_t> m_propsFinished = { 0 }; std::atomic<uint64_t> m_propsFinished = { 0 };
std::vector<int32_t> m_tempGpuRun; std::vector<int32_t> m_tempGpuRun;
std::vector<int32_t> m_tempGpuRunDurations;
}; };