Skip to content

Proposed get_output_mr() refactor #6

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
simoneves wants to merge 5 commits into
base: simoneves/decimal_pr3
Choose a base branch
from
Open

Proposed get_output_mr() refactor #6

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
7322c78
Use passed-in MR for all cuDF operations that take one
simoneves May 19, 2026
9ec9295
Move get_output_mr() calls up one level
simoneves May 19, 2026
c8c1470
Move get_output_mr() calls up one more level
simoneves May 19, 2026
abd221a
Format
simoneves May 19, 2026
6bdbc94
Fix breakage
simoneves May 19, 2026
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
164 changes: 99 additions & 65 deletions velox/experimental/cudf/exec/CudfGroupby.cpp
View file
Open in desktop

Large diffs are not rendered by default.

11 changes: 8 additions & 3 deletions velox/experimental/cudf/exec/CudfGroupby.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,6 +20,8 @@

#include <cudf/groupby.hpp>

#include <rmm/resource_ref.hpp>

namespace facebook::velox::cudf_velox {

struct GroupbyAggregator {
Expand All @@ -31,11 +33,13 @@ struct GroupbyAggregator {
virtual void addGroupbyRequest(
cudf::table_view const& tbl,
std::vector<cudf::groupby::aggregation_request>& requests,
rmm::cuda_stream_view stream) = 0;
rmm::cuda_stream_view stream,
rmm::device_async_resource_ref mr) = 0;

virtual std::unique_ptr<cudf::column> makeOutputColumn(
std::vector<cudf::groupby::aggregation_result>& results,
rmm::cuda_stream_view stream) = 0;
rmm::cuda_stream_view stream,
rmm::device_async_resource_ref mr) = 0;

virtual ~GroupbyAggregator() = default;

Expand Down Expand Up @@ -101,7 +105,8 @@ class CudfGroupby : public CudfOperatorBase {
std::vector<column_index_t> const& groupByKeys,
std::vector<std::unique_ptr<GroupbyAggregator>>& aggregators,
TypePtr const& outputType,
rmm::cuda_stream_view stream);
rmm::cuda_stream_view stream,
rmm::device_async_resource_ref mr);

CudfVectorPtr releaseAndResetBufferedResult();

Expand Down
Loading
Loading