Skip to content

[GLUTEN-10933][VL]fix: Fix the CudfVector childreSize is not correct #11326

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
Jan 6, 2026
Merged

[GLUTEN-10933][VL]fix: Fix the CudfVector childreSize is not correct #11326

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
7 changes: 7 additions & 0 deletions cpp/velox/memory/VeloxColumnarBatch.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -29,6 +29,13 @@ class VeloxColumnarBatch final : public ColumnarBatch {
VeloxColumnarBatch(facebook::velox::RowVectorPtr rowVector)
: ColumnarBatch(rowVector->childrenSize(), rowVector->size()), rowVector_(rowVector) {}

#ifdef GLUTEN_ENABLE_GPU
// The batch may be CudfVector, it's childrenSize is not correct, many tests failed if correcting the value
// https://github.com/facebookincubator/velox/pull/15629#discussion_r2581655771
VeloxColumnarBatch(facebook::velox::RowVectorPtr rowVector, int32_t numColumns)
: ColumnarBatch(numColumns, rowVector->size()), rowVector_(rowVector) {}
#endif
Comment on lines +32 to +37
Copy link
Member

@zhztheplayer zhztheplayer Jan 5, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Compared to adding a new ctor, what if we just change the old ctor to call rowVector->rowType()->size()?

By the way. Why childrenSize() is inaccurate for cuDF row vector?

Copy link
Contributor Author

@jinchengchenghh jinchengchenghh Jan 6, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Because the childrenSize() always returns 0, RowVector contains flatVector, but CudfVector contains cudf::table(contains cudf::column)

Copy link
Contributor Author

@jinchengchenghh jinchengchenghh Jan 6, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

As this discussion https://github.com/facebookincubator/velox/pull/15629/files#r2572687025


std::string getType() const override {
return kType;
}
Expand Down
2 changes: 1 addition & 1 deletion cpp/velox/utils/GpuBufferBatchResizer.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -159,7 +159,7 @@ std::shared_ptr<VeloxColumnarBatch> makeCudfTable(
auto cudfTable = std::make_unique<cudf::table>(std::move(cudfColumns));
stream.synchronize();
return std::make_shared<VeloxColumnarBatch>(
std::make_shared<cudf_velox::CudfVector>(pool, type, numRows, std::move(cudfTable), stream));
std::make_shared<cudf_velox::CudfVector>(pool, type, numRows, std::move(cudfTable), stream), type->size());
}

} // namespace
Expand Down