Skip to content

ITS: GPU: fix perPrimaryVertexProcessing #15522

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
f3sch wants to merge 1 commit into
base: dev
Choose a base branch
from
+10 −4
Open

ITS: GPU: fix perPrimaryVertexProcessing #15522

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 10 additions & 4 deletions Detectors/ITSMFT/ITS/tracking/GPU/cuda/TrackerTraitsGPU.cxx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -52,8 +52,6 @@ void TrackerTraitsGPU<NLayers>::initialiseTimeFrame(const int iteration)
if (this->mTrkParams[iteration].PassFlags[IterationStep::FirstPass] || this->mTrkParams[iteration].PassFlags[IterationStep::UseUPCMask]) {
mTimeFrameGPU->loadROFCutMask(iteration);
}
// push every create artefact on the stack
mTimeFrameGPU->pushMemoryStack(iteration);
}

template <int NLayers>
Expand All @@ -68,8 +66,9 @@ void TrackerTraitsGPU<NLayers>::computeLayerTracklets(const int iteration, int i
{
const auto topology = mTimeFrameGPU->getDeviceTrackingTopologyView();
const auto hostTopology = mTimeFrameGPU->getTrackingTopologyView();
const bool loadFirstPassData = this->mTrkParams[iteration].PassFlags[IterationStep::FirstPass] && iVertex <= 0; // load data only on first pass and first vertex
for (int iLayer{0}; iLayer < this->mTrkParams[iteration].NLayers; ++iLayer) {
if (this->mTrkParams[iteration].PassFlags[IterationStep::FirstPass]) {
if (loadFirstPassData) {
mTimeFrameGPU->createUsedClustersDevice(iLayer);
mTimeFrameGPU->loadClustersDevice(iLayer);
mTimeFrameGPU->loadClustersIndexTables(iLayer);
Expand All @@ -78,9 +77,16 @@ void TrackerTraitsGPU<NLayers>::computeLayerTracklets(const int iteration, int i
mTimeFrameGPU->recordEvent(iLayer);
}

for (int linkId{0}; linkId < hostTopology.nLinks; ++linkId) {
mTimeFrameGPU->createTrackletsLUTDevice(loadFirstPassData, linkId); // on first pass allocates, then only clears memory
}

// Stack allocations created from trackleting through road finding are scoped to one tracker pass.
// With per-primary-vertex processing, the chain is called once per vertex while initialisation is only done once.
mTimeFrameGPU->pushMemoryStack(iteration);

for (int linkId{0}; linkId < hostTopology.nLinks; ++linkId) {
const auto link = hostTopology.getLink(linkId);
mTimeFrameGPU->createTrackletsLUTDevice(this->mTrkParams[iteration].PassFlags[IterationStep::FirstPass], linkId);
mTimeFrameGPU->waitEvent(linkId, link.fromLayer);
mTimeFrameGPU->waitEvent(linkId, link.toLayer);
countTrackletsInROFsHandler<NLayers>(mTimeFrameGPU->getDeviceIndexTableUtils(),
Expand Down
Loading