[SPARK-58080][PYTHON][TESTS] Add ASV microbenchmark for SQL_GROUPED_MAP_PANDAS_UDF_WITH_STATE#57189
Closed
Yicong-Huang wants to merge 1 commit into
Closed
Conversation
HyukjinKwon
approved these changes
Jul 13, 2026
uros-b
approved these changes
Jul 14, 2026
uros-b
left a comment
uros-b
left a comment
Member
There was a problem hiding this comment.
Thank you @Yicong-Huang and @HyukjinKwon!
Yicong-Huang
force-pushed
the
bench-apply-in-pandas-with-state
branch
from
1512ec8 to
6f97f04
Compare
Contributor
Author
|
Thanks @HyukjinKwon and @uros-b, merging now |
Yicong-Huang
added a commit
that referenced
this pull request
Jul 15, 2026
…P_PANDAS_UDF_WITH_STATE ### What changes were proposed in this pull request? Add an ASV micro-benchmark for `SQL_GROUPED_MAP_PANDAS_UDF_WITH_STATE` (`applyInPandasWithState`) to `bench_eval_type.py`. Unlike TransformWithState, no state server socket is involved: `ApplyInPandasWithStateSerializer` reconstructs each `GroupState` entirely from a metadata column carried inline in the Arrow stream. The benchmark emits the data columns followed by one trailing struct column (`__state`, matching the JVM `ApplyInPandasWithStateWriter.STATE_METADATA_SCHEMA`), and faithfully reproduces the JVM writer's bin-packing and cross-batch chunking (`isLastChunk` set only on a group's final chunk), so a group split across batches reassembles into one `GroupState`. Scenarios cover few/many groups, small/large group sizes, wide columns, mixed value types (string/binary/boolean) and a nested struct column. UDFs: `identity_udf`, `sort_udf`, `count_udf`. `identity_udf`/`sort_udf` pass values through; `count_udf` exercises the per-group state read/write path (reads the running count via `getOption`, writes it back via `update`) and re-emits the key plus the count. ### Why are the changes needed? Part of [SPARK-55724](https://issues.apache.org/jira/browse/SPARK-55724). Establishes a performance baseline before refactoring `SQL_GROUPED_MAP_PANDAS_UDF_WITH_STATE`. ### Does this PR introduce _any_ user-facing change? No ### How was this patch tested? `COLUMNS=120 asv run --python=same --bench "ApplyInPandasWithState" -a "repeat=3"` (one of two stable runs): `ApplyInPandasWithStateUDFTimeBench`: ```text ================ ============== ============ ============ -- udf ---------------- ---------------------------------------- scenario identity_udf sort_udf count_udf ================ ============== ============ ============ few_groups_sm 1.02+-0s 1.06+-0.01s 128+-1ms few_groups_lg 6.98+-0.05s 7.18+-0.05s 762+-3ms many_groups_sm 6.76+-0.04s 7.62+-0.06s 3.14+-0.01s many_groups_lg 4.80+-0.1s 4.95+-0.08s 934+-3ms wide_cols 5.92+-0.02s 6.17+-0.1s 878+-20ms mixed_cols 4.49+-0.01s 4.76+-0.04s 704+-5ms nested_struct 9.17+-0.1s 10.8+-0.07s 3.32+-0.02s ================ ============== ============ ============ ``` `ApplyInPandasWithStateUDFPeakmemBench`: ```text ================ ============== ========== =========== -- udf ---------------- ------------------------------------- scenario identity_udf sort_udf count_udf ================ ============== ========== =========== few_groups_sm 118M 119M 106M few_groups_lg 272M 280M 221M many_groups_sm 172M 173M 144M many_groups_lg 172M 178M 140M wide_cols 284M 282M 233M mixed_cols 187M 188M 182M nested_struct 227M 228M 211M ================ ============== ========== =========== ``` ### Was this patch authored or co-authored using generative AI tooling? No Closes #57189 from Yicong-Huang/bench-apply-in-pandas-with-state. Authored-by: Yicong Huang <17627829+Yicong-Huang@users.noreply.github.com> Signed-off-by: Yicong-Huang <17627829+Yicong-Huang@users.noreply.github.com> (cherry picked from commit 4bea4ad) Signed-off-by: Yicong-Huang <17627829+Yicong-Huang@users.noreply.github.com>
Contributor
Author
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What changes were proposed in this pull request?
Add an ASV micro-benchmark for
SQL_GROUPED_MAP_PANDAS_UDF_WITH_STATE(applyInPandasWithState) tobench_eval_type.py.Unlike TransformWithState, no state server socket is involved:
ApplyInPandasWithStateSerializerreconstructs eachGroupStateentirely from a metadata column carried inline in the Arrow stream. The benchmark emits the data columns followed by one trailing struct column (__state, matching the JVMApplyInPandasWithStateWriter.STATE_METADATA_SCHEMA), and faithfully reproduces the JVM writer's bin-packing and cross-batch chunking (isLastChunkset only on a group's final chunk), so a group split across batches reassembles into oneGroupState.Scenarios cover few/many groups, small/large group sizes, wide columns, mixed value types (string/binary/boolean) and a nested struct column. UDFs:
identity_udf,sort_udf,count_udf.identity_udf/sort_udfpass values through;count_udfexercises the per-group state read/write path (reads the running count viagetOption, writes it back viaupdate) and re-emits the key plus the count.Why are the changes needed?
Part of SPARK-55724. Establishes a performance baseline before refactoring
SQL_GROUPED_MAP_PANDAS_UDF_WITH_STATE.Does this PR introduce any user-facing change?
No
How was this patch tested?
COLUMNS=120 asv run --python=same --bench "ApplyInPandasWithState" -a "repeat=3"(one of two stable runs):ApplyInPandasWithStateUDFTimeBench:ApplyInPandasWithStateUDFPeakmemBench:Was this patch authored or co-authored using generative AI tooling?
No