Skip to content

[ML] Opt.: Load the consecutive clusters of a lazy buffer in one event loop #23793

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

Merged
siliataider merged 1 commit into from
Oct 22, 2026
Merged

[ML] Opt.: Load the consecutive clusters of a lazy buffer in one event loop #23793

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
24 changes: 24 additions & 0 deletions tree/ml/src/RDataLoaderEngine.cxx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,6 +11,23 @@

namespace ROOT::Experimental::Internal::ML {

namespace {
/// \brief Merge the clusters that follow each other in the same dataframe
std::vector<RClusterRange> MergeConsecutiveClusters(const std::vector<RClusterRange> &clusters)
{
std::vector<RClusterRange> merged;
for (const auto &cluster : clusters) {
if (!merged.empty() && merged.back().rdfIdx == cluster.rdfIdx && merged.back().end == cluster.start) {
merged.back().end = cluster.end;
merged.back().SetNumEntries(merged.back().GetNumEntries() + cluster.GetNumEntries());
} else {
merged.push_back(cluster);
}
}
return merged;
}
} // namespace

/// \brief Describe how the loader's columns map onto a batch-tensor row.
std::vector<RColumnLayout> RDataLoaderEngine::MakeColumnLayout()
{
Expand Down Expand Up @@ -332,6 +349,12 @@ void RDataLoaderEngine::LoadData()
stagingBuffer.Reserve(accumulatedEntries, fClusterLoader->GetNumChunkCols());
std::size_t rowOffset = 0;

// without shuffling, the clusters are contiguous, so we merge them to run in the same event loop;
// otherwise (shuffling is enabled), it is one event loop per cluster
if (!discovering) {
trainClustersToLoad = MergeConsecutiveClusters(trainClustersToLoad);
}

for (auto &cluster : trainClustersToLoad) {
auto loadedEntries =
fClusterLoader->LoadTrainingClusterInto(stagingBuffer, cluster.rdfIdx, cluster.start, cluster.end);
Expand Down Expand Up @@ -407,6 +430,7 @@ void RDataLoaderEngine::LoadData()
stagingBuffer.Reserve(accumulatedEntries, fClusterLoader->GetNumChunkCols());
std::size_t rowOffset = 0;

valClustersToLoad = MergeConsecutiveClusters(valClustersToLoad);
for (const auto &cluster : valClustersToLoad) {
fClusterLoader->LoadValidationClusterInto(stagingBuffer, cluster.rdfIdx, cluster.start, cluster.end);
rowOffset += cluster.GetNumEntries();
Expand Down
Loading