feat: precomputed-hash column for Partial→Hash-Repartition shuffle (draft)#21816
Closed
Dandandan wants to merge 2 commits intoapache:mainfrom
Closed
feat: precomputed-hash column for Partial→Hash-Repartition shuffle (draft)#21816Dandandan wants to merge 2 commits intoapache:mainfrom
Dandandan wants to merge 2 commits intoapache:mainfrom
Conversation
Adds an opt-in `datafusion.execution.emit_aggregate_group_hash` config option. When enabled, a Partial `AggregateExec` whose immediate consumer is `RepartitionExec` with `Partitioning::Hash` over the same group columns appends a trailing `__datafusion_precomputed_hash: UInt64` field to its output. The hash is computed once with `REPARTITION_RANDOM_STATE` and `RepartitionExec` consumes it directly via a new fast path, eliminating a full rehashing pass on the shuffle — biggest wins on string/binary group keys (e.g. Clickbench regexp_replace keys). Pieces: - `AggregateExec::with_emit_group_hash(bool)` rebuilds schema + cache. - `create_schema` appends the tagged `UInt64` field with source-column indices in field metadata so multi-column groups work. - `row_hash::emit` / `transform_to_states` hash the group arrays with the repartition seed and append as the last column. - `RepartitionExec::Hash` scans its input schema for the precomputed marker; matches both "partitioning expr IS the hash column" and "partitioning exprs are the recorded source columns in order". - New optimizer rule `EmitPartialAggregateHash` flips the flag when the Partial→Hash-Repartition pattern is present; config-gated so default behavior is unchanged. - Field sits at the end of the Partial's output; Final's indexing into group/state columns is unaffected and its own output schema is clean. Out of scope for this PR: - FinalPartitioned `GroupValues` reuse (requires GroupValues trait extension and a second agg-seed hash column — noted as follow-up). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Contributor
Author
|
run benchmarks |
github-actions
Bot
added
documentation
|
🤖 Benchmark running (GKE) | trigger CPU Details (lscpu)Comparing precomputed-group-hash-for-repartition (de92c7a) to 067ba4b (merge-base) diff using: tpch File an issue against this benchmark runner |
|
🤖 Benchmark running (GKE) | trigger CPU Details (lscpu)Comparing precomputed-group-hash-for-repartition (de92c7a) to 067ba4b (merge-base) diff using: clickbench_partitioned File an issue against this benchmark runner |
|
🤖 Benchmark running (GKE) | trigger CPU Details (lscpu)Comparing precomputed-group-hash-for-repartition (de92c7a) to 067ba4b (merge-base) diff using: tpcds File an issue against this benchmark runner |
Extends the precomputed-hash pipeline through the Final side: - Unify `AGGREGATION_HASH_SEED` with `REPARTITION_RANDOM_STATE` (seed 0) so a single hash value is valid for the Partial's hash table, RepartitionExec routing, and the Final's hash-table probe. Hashbrown selects buckets from high bits while routing uses low bits, so reuse is safe in practice. - Add `GroupValues::intern_with_hashes(cols, hashes, groups)` with a default implementation that falls back to `intern`. Override on `GroupValuesRows` (the row-based general path) to skip the `create_hashes` call; factored the probe/insert loop into `GroupValuesRows::intern_rows`. - Wire `GroupedHashAggregateStream`: when the Final-side input schema carries a `UInt64` field tagged with the precomputed-hash metadata, extract the array and feed its values into `intern_with_hashes`. - Flip `datafusion.execution.emit_aggregate_group_hash` default to `true`. The rule is still gated by the config so it's trivially disabled per session if needed. Other `GroupValues` specializations (single-column bytes/primitive/ boolean, multi-column composite) pick up the fallback behavior — they still work correctly, just without the Final-side saving until each implementation opts in. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Contributor
Author
|
run benchmarks |
|
🤖 Benchmark running (GKE) | trigger CPU Details (lscpu)Comparing precomputed-group-hash-for-repartition (ac295b2) to 067ba4b (merge-base) diff using: clickbench_partitioned File an issue against this benchmark runner |
|
🤖 Benchmark running (GKE) | trigger CPU Details (lscpu)Comparing precomputed-group-hash-for-repartition (ac295b2) to 067ba4b (merge-base) diff using: tpcds File an issue against this benchmark runner |
|
🤖 Benchmark running (GKE) | trigger CPU Details (lscpu)Comparing precomputed-group-hash-for-repartition (ac295b2) to 067ba4b (merge-base) diff using: tpch File an issue against this benchmark runner |
|
🤖 Benchmark completed (GKE) | trigger Instance: CPU Details (lscpu)Details
Resource Usagetpch — base (merge-base)
tpch — branch
File an issue against this benchmark runner |
|
🤖 Benchmark completed (GKE) | trigger Instance: CPU Details (lscpu)Details
Resource Usagetpcds — base (merge-base)
tpcds — branch
File an issue against this benchmark runner |
|
🤖 Benchmark completed (GKE) | trigger Instance: CPU Details (lscpu)Details
Resource Usageclickbench_partitioned — base (merge-base)
clickbench_partitioned — branch
File an issue against this benchmark runner |
|
🤖 Benchmark completed (GKE) | trigger Instance: CPU Details (lscpu)Details
Resource Usagetpch — base (merge-base)
tpch — branch
File an issue against this benchmark runner |
|
🤖 Benchmark completed (GKE) | trigger Instance: CPU Details (lscpu)Details
Resource Usageclickbench_partitioned — base (merge-base)
clickbench_partitioned — branch
File an issue against this benchmark runner |
|
🤖 Benchmark completed (GKE) | trigger Instance: CPU Details (lscpu)Details
Resource Usagetpcds — base (merge-base)
tpcds — branch
File an issue against this benchmark runner |
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
2 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.
Summary
datafusion.execution.emit_aggregate_group_hash(default false). When enabled, a PartialAggregateExecfeeding a HashRepartitionExecover the same group columns emits a trailingUInt64column of precomputed row hashes (seeded withREPARTITION_RANDOM_STATE).RepartitionExec::Hashconsumes the column directly via a new fast path, eliminating one full rehashing pass on the shuffle. Biggest wins on string/binary group keys (e.g. Clickbenchregexp_replacekeys).EmitPartialAggregateHashflips the flag on matchingPartial → Hash-Repartitionpairs. Config-gated so default behavior is unchanged.What the rule does
datafusion.precomputed_hash = "repartition_seed_0"plus adatafusion.precomputed_hash_colsCSV of source indices).Out of scope / follow-ups
GroupValuesreuse. Would need aGroupValuestrait extension + a second hash column seeded withAGGREGATION_HASH_SEEDto preserve the intentional seed difference between the shuffle hash and the agg-side probe hash.Test plan
create_hashes(&group_arrays, REPARTITION_RANDOM_STATE)with_emit_group_hash(true)is a no-op on non-Partial modesdetect_precomputed_hash_columnmatches (a) hash-column-as-partitioning-expr and (b) multi-column group with source indices, and rejects subset/reorderingcargo fmt --all+cargo clippy -p datafusion-physical-plan -p datafusion-physical-optimizer -p datafusion-common --all-targets -- -D warnings🤖 Generated with Claude Code