Skip to content

perf: avoid FFI import/export when passing batches between native plans #3930

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

Draft
andygrove wants to merge 22 commits into
base: main
Choose a base branch
from
Draft

perf: avoid FFI import/export when passing batches between native plans #3930

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
22 commits
Select commit Hold shift + click to select a range
86a4989
docs: add design spec for batch stash shuffle optimization
andygrove Apr 12, 2026
488522c
docs: add implementation plan for batch stash shuffle optimization
andygrove Apr 12, 2026
ca8ee6a
feat: add BatchStash registry for native batch handle passing
andygrove Apr 12, 2026
e68d777
feat: add CometHandleBatchIterator Java class and JNI bridge
andygrove Apr 12, 2026
07c86c7
feat: add executePlanBatchHandle JNI function for stash-mode output
andygrove Apr 12, 2026
d2f87ab
feat: add stash mode to CometExecIterator for batch handle output
andygrove Apr 12, 2026
0a49ca6
feat: add handle-mode input path to ScanExec for batch stash retrieval
andygrove Apr 12, 2026
84862ef
feat: detect CometHandleBatchIterator in planner and enable handle mode
andygrove Apr 12, 2026
3588543
feat: preserve CometExecIterator reference through shuffle dependency…
andygrove Apr 12, 2026
db137ad
feat: integrate batch stash mode in CometNativeShuffleWriter
andygrove Apr 12, 2026
ed95e83
style: apply formatting fixes
andygrove Apr 12, 2026
a827437
fix: use scan source name for handle mode detection instead of JNI is…
andygrove Apr 12, 2026
3af0e9a
chore: remove JVM crash log files
andygrove Apr 12, 2026
e6c3a42
refactor: deduplicate executePlan logic and address code review findings
andygrove Apr 12, 2026
6cc0e78
refactor: use protobuf field for batch stash handle mode instead of s…
andygrove Apr 12, 2026
c42be90
chore: remove design docs from PR
andygrove Apr 12, 2026
f70119a
fix: add batch_stash_handle field to Scan struct literals in tests
andygrove Apr 12, 2026
c64d482
fix: use InputBatch::Complete for stashed batches to bypass schema re…
andygrove Apr 12, 2026
d37752a
feat: add spark.comet.exec.shuffle.batchStash.enabled config (default…
andygrove Apr 12, 2026
38af7d9
fix: apply schema reconciliation for stashed batches with type mismat…
andygrove Apr 12, 2026
ba07bf7
fix: address PR review feedback for batch stash
andygrove Apr 13, 2026
008d7cf
refactor: move batch stash from global static into ExecutionContext
andygrove Apr 16, 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
10 changes: 10 additions & 0 deletions common/src/main/scala/org/apache/comet/CometConf.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -448,6 +448,16 @@ object CometConf extends ShimCometConf {
.intConf
.createWithDefault(1)

val COMET_SHUFFLE_BATCH_STASH_ENABLED: ConfigEntry[Boolean] =
Copy link
Copy Markdown
Member Author

@andygrove andygrove Apr 12, 2026

Choose a reason for hiding this comment

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

Added config for now just in case we discover bugs, but I plan on removing this config in the future.

conf(s"$COMET_EXEC_CONFIG_PREFIX.shuffle.batchStash.enabled")
.category(CATEGORY_SHUFFLE)
.doc(
"When enabled, batches passed between a native child plan and a native shuffle " +
"writer are transferred via an opaque handle instead of Arrow FFI, avoiding " +
"unnecessary serialization overhead.")
.booleanConf
.createWithDefault(true)

val COMET_COLUMNAR_SHUFFLE_ASYNC_ENABLED: ConfigEntry[Boolean] =
conf("spark.comet.columnar.shuffle.async.enabled")
.category(CATEGORY_SHUFFLE)
Expand Down
4 changes: 1 addition & 3 deletions native/Cargo.lock
View file
Open in desktop

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

Loading
Loading