Skip to content

[VL] Minor refactor to replace START_TIMEING with SCOPE_TIMER #11320

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
jinchengchenghh merged 1 commit into from
Dec 22, 2025
Merged

[VL] Minor refactor to replace START_TIMEING with SCOPE_TIMER #11320

Show file tree
Hide file tree
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
56 changes: 30 additions & 26 deletions cpp/velox/shuffle/VeloxHashShuffleWriter.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -269,13 +269,14 @@ arrow::Status VeloxHashShuffleWriter::write(std::shared_ptr<ColumnarBatch> cb, i
VELOX_CHECK(numColumns >= 2);
auto pidBatch = veloxColumnBatch->select(veloxPool_.get(), {0});
auto pidArr = getFirstColumn(*(pidBatch->getRowVector()));
START_TIMING(cpuWallTimingList_[CpuWallTimingCompute]);
std::fill(std::begin(partition2RowCount_), std::end(partition2RowCount_), 0);
RETURN_NOT_OK(partitioner_->compute(pidArr, pidBatch->numRows(), row2Partition_));
for (auto& pid : row2Partition_) {
partition2RowCount_[pid]++;
{
SCOPED_TIMER(cpuWallTimingList_[CpuWallTimingCompute]);
std::fill(std::begin(partition2RowCount_), std::end(partition2RowCount_), 0);
RETURN_NOT_OK(partitioner_->compute(pidArr, pidBatch->numRows(), row2Partition_));
for (auto& pid : row2Partition_) {
partition2RowCount_[pid]++;
}
}
END_TIMING();
std::vector<int32_t> range;
range.reserve(numColumns);
for (int32_t i = 1; i < numColumns; i++) {
Expand All @@ -289,9 +290,10 @@ arrow::Status VeloxHashShuffleWriter::write(std::shared_ptr<ColumnarBatch> cb, i
auto veloxColumnBatch = VeloxColumnarBatch::from(veloxPool_.get(), cb);
VELOX_CHECK_NOT_NULL(veloxColumnBatch);
facebook::velox::RowVectorPtr rv;
START_TIMING(cpuWallTimingList_[CpuWallTimingFlattenRV]);
rv = veloxColumnBatch->getFlattenedRowVector();
END_TIMING();
{
SCOPED_TIMER(cpuWallTimingList_[CpuWallTimingFlattenRV]);
rv = veloxColumnBatch->getFlattenedRowVector();
}
if (isExtremelyLargeBatch(rv)) {
auto numRows = rv->size();
int32_t offset = 0;
Expand All @@ -313,23 +315,25 @@ arrow::Status VeloxHashShuffleWriter::partitioningAndDoSplit(facebook::velox::Ro
std::fill(std::begin(partition2RowCount_), std::end(partition2RowCount_), 0);
if (partitioner_->hasPid()) {
auto pidArr = getFirstColumn(*rv);
START_TIMING(cpuWallTimingList_[CpuWallTimingCompute]);
RETURN_NOT_OK(partitioner_->compute(pidArr, rv->size(), row2Partition_));
for (auto& pid : row2Partition_) {
partition2RowCount_[pid]++;
{
SCOPED_TIMER(cpuWallTimingList_[CpuWallTimingCompute]);
RETURN_NOT_OK(partitioner_->compute(pidArr, rv->size(), row2Partition_));
for (auto& pid : row2Partition_) {
partition2RowCount_[pid]++;
}
}
END_TIMING();
auto strippedRv = getStrippedRowVector(*rv);
RETURN_NOT_OK(initFromRowVector(*strippedRv));
RETURN_NOT_OK(doSplit(*strippedRv, memLimit));
} else {
RETURN_NOT_OK(initFromRowVector(*rv));
START_TIMING(cpuWallTimingList_[CpuWallTimingCompute]);
RETURN_NOT_OK(partitioner_->compute(nullptr, rv->size(), row2Partition_));
for (auto& pid : row2Partition_) {
partition2RowCount_[pid]++;
{
SCOPED_TIMER(cpuWallTimingList_[CpuWallTimingCompute]);
RETURN_NOT_OK(partitioner_->compute(nullptr, rv->size(), row2Partition_));
for (auto& pid : row2Partition_) {
partition2RowCount_[pid]++;
}
}
END_TIMING();
RETURN_NOT_OK(doSplit(*rv, memLimit));
}
return arrow::Status::OK();
Expand Down Expand Up @@ -416,13 +420,13 @@ arrow::Status VeloxHashShuffleWriter::doSplit(const facebook::velox::RowVector&
RETURN_NOT_OK(buildPartition2Row(rowNum));
RETURN_NOT_OK(updateInputHasNull(rv));

START_TIMING(cpuWallTimingList_[CpuWallTimingIteratePartitions]);

setSplitState(SplitState::kPreAlloc);
// Calculate buffer size based on available offheap memory, history average bytes per row and options_.bufferSize.
auto preAllocBufferSize = calculatePartitionBufferSize(rv, memLimit);
RETURN_NOT_OK(preAllocPartitionBuffers(preAllocBufferSize));
END_TIMING();
{
SCOPED_TIMER(cpuWallTimingList_[CpuWallTimingIteratePartitions]);
setSplitState(SplitState::kPreAlloc);
// Calculate buffer size based on available offheap memory, history average bytes per row and options_.bufferSize.
auto preAllocBufferSize = calculatePartitionBufferSize(rv, memLimit);
RETURN_NOT_OK(preAllocPartitionBuffers(preAllocBufferSize));
}

printPartitionBuffer();

Expand Down
35 changes: 19 additions & 16 deletions cpp/velox/shuffle/VeloxRssSortShuffleWriter.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,7 +21,6 @@
#include "shuffle/ShuffleSchema.h"
#include "utils/Common.h"
#include "utils/Macros.h"
#include "utils/VeloxArrowUtils.h"

#include "velox/common/base/Nulls.h"
#include "velox/type/Type.h"
Expand Down Expand Up @@ -79,10 +78,11 @@ arrow::Status VeloxRssSortShuffleWriter::write(std::shared_ptr<ColumnarBatch> cb
VELOX_CHECK(numColumns >= 2);
auto pidBatch = veloxColumnBatch->select(veloxPool_.get(), {0});
auto pidArr = getFirstColumn(*(pidBatch->getRowVector()));
START_TIMING(cpuWallTimingList_[CpuWallTimingCompute]);
setSortState(RssSortState::kSort);
RETURN_NOT_OK(partitioner_->compute(pidArr, pidBatch->numRows(), batches_.size(), rowVectorIndexMap_));
END_TIMING();
{
SCOPED_TIMER(cpuWallTimingList_[CpuWallTimingCompute]);
setSortState(RssSortState::kSort);
RETURN_NOT_OK(partitioner_->compute(pidArr, pidBatch->numRows(), batches_.size(), rowVectorIndexMap_));
}
std::vector<int32_t> range;
range.reserve(numColumns);
for (int32_t i = 1; i < numColumns; i++) {
Expand All @@ -96,24 +96,27 @@ arrow::Status VeloxRssSortShuffleWriter::write(std::shared_ptr<ColumnarBatch> cb
auto veloxColumnBatch = VeloxColumnarBatch::from(veloxPool_.get(), cb);
VELOX_CHECK_NOT_NULL(veloxColumnBatch);
facebook::velox::RowVectorPtr rv;
START_TIMING(cpuWallTimingList_[CpuWallTimingFlattenRV]);
rv = veloxColumnBatch->getFlattenedRowVector();
END_TIMING();
{
SCOPED_TIMER(cpuWallTimingList_[CpuWallTimingFlattenRV]);
rv = veloxColumnBatch->getFlattenedRowVector();
}
if (partitioner_->hasPid()) {
auto pidArr = getFirstColumn(*rv);
START_TIMING(cpuWallTimingList_[CpuWallTimingCompute]);
setSortState(RssSortState::kSort);
RETURN_NOT_OK(partitioner_->compute(pidArr, rv->size(), batches_.size(), rowVectorIndexMap_));
END_TIMING();
{
SCOPED_TIMER(cpuWallTimingList_[CpuWallTimingCompute]);
setSortState(RssSortState::kSort);
RETURN_NOT_OK(partitioner_->compute(pidArr, rv->size(), batches_.size(), rowVectorIndexMap_));
}
auto strippedRv = getStrippedRowVector(*rv);
RETURN_NOT_OK(initFromRowVector(*strippedRv));
RETURN_NOT_OK(doSort(strippedRv, sortBufferMaxSize_));
} else {
RETURN_NOT_OK(initFromRowVector(*rv));
START_TIMING(cpuWallTimingList_[CpuWallTimingCompute]);
setSortState(RssSortState::kSort);
RETURN_NOT_OK(partitioner_->compute(nullptr, rv->size(), batches_.size(), rowVectorIndexMap_));
END_TIMING();
{
SCOPED_TIMER(cpuWallTimingList_[CpuWallTimingCompute]);
setSortState(RssSortState::kSort);
RETURN_NOT_OK(partitioner_->compute(nullptr, rv->size(), batches_.size(), rowVectorIndexMap_));
}
RETURN_NOT_OK(doSort(rv, sortBufferMaxSize_));
}
}
Expand Down
18 changes: 6 additions & 12 deletions cpp/velox/utils/Common.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -35,17 +35,11 @@ static inline void fastCopy(void* dst, const void* src, size_t n) {
facebook::velox::simd::memcpy(dst, src, n);
}

#define START_TIMING(timing) \
{ \
auto ptiming = &timing; \
facebook::velox::DeltaCpuWallTimer timer{ \
[ptiming](const facebook::velox::CpuWallTiming& delta) { ptiming->add(delta); }};

#define END_TIMING() }

#define SCOPED_TIMER(timing) \
auto ptiming = &timing; \
facebook::velox::DeltaCpuWallTimer timer{ \
[ptiming](const facebook::velox::CpuWallTiming& delta) { ptiming->add(delta); }};
#define SCOPED_TIMER(timing) \
do { \
auto ptiming = &timing; \
facebook::velox::DeltaCpuWallTimer timer{ \
[ptiming](const facebook::velox::CpuWallTiming& delta) { ptiming->add(delta); }}; \
} while (0)

} // namespace gluten