perf: sort-merge join (SMJ) batch deferred filtering and move mark joins to bitwise stream. Near-unique LEFT and FULL SMJ 20-50x faster#21184
Merged
mbutrovich merged 29 commits intoapache:mainfrom Mar 30, 2026
Conversation
Contributor
Author
Member
|
run benchmarks sort_merge_join |
Contributor
Author
Note that the 2 queries I expect a speedup on in the |
|
🤖 Criterion benchmark running (GKE) | trigger File an issue against this benchmark runner |
|
Benchmark for this request failed. Last 20 lines of output: Click to expandFile an issue against this benchmark runner |
Contributor
Author
|
Also I'm now confused where I should add benchmarks. #20464 added Criterion SMJ benchmarks for sort-merge join , but it's missing scenarios from dfbench's |
Contributor
rluvaton
reviewed
Mar 26, 2026
rluvaton
reviewed
Mar 26, 2026
comphead
reviewed
Mar 27, 2026
…emiAntiMark) into the same SMJ folder.
mbutrovich
commented
Mar 27, 2026
| /// Calculated as sum of peak memory values across partitions | ||
| peak_mem_used: Gauge, | ||
| /// Metrics related to spilling | ||
| spill_metrics: SpillMetrics, |
Contributor
Author
There was a problem hiding this comment.
Moved SpillMetrics construction from SortMergeJoinMetrics into exec.rs where the SpillManager is now built (shared across both streams). The metrics are still registered into the same ExecutionPlanMetricsSet and reported via metrics() — just constructed in a different place.
mbutrovich
changed the title
comphead
approved these changes
Mar 27, 2026
Contributor
comphead
left a comment
comphead
left a comment
There was a problem hiding this comment.
Thanks @mbutrovich numbers looks great, lets wait some time if anyone wants to take a look as well
Contributor
|
run benchmark tpch tpcds |
|
🤖 Benchmark running (GKE) | trigger CPU Details (lscpu)Comparing simplify_smj_full_opt (fa4e99d) to a910b03 (merge-base) diff using: tpcds File an issue against this benchmark runner |
|
🤖 Benchmark running (GKE) | trigger CPU Details (lscpu)Comparing simplify_smj_full_opt (fa4e99d) to a910b03 (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 |
Contributor
|
Nice, this runs correctly with SMJ, but it doesn't impact those joins it seems. |
Contributor
|
run benchmark tpch |
Contributor
|
run benchmark tpch10 |
|
🤖 Benchmark running (GKE) | trigger CPU Details (lscpu)Comparing simplify_smj_full_opt (fa4e99d) to a910b03 (merge-base) diff using: tpch File an issue against this benchmark runner |
|
🤖 Benchmark running (GKE) | trigger CPU Details (lscpu)Comparing simplify_smj_full_opt (fa4e99d) to a910b03 (merge-base) diff using: tpch10 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 Usagetpch10 — base (merge-base)
tpch10 — branch
File an issue against this benchmark runner |
Contributor
|
No useful comments to add, just: thank you, love the direction! |
# Conflicts: # datafusion/physical-plan/src/joins/sort_merge_join/metrics.rs
Rich-T-kid
pushed a commit
to Rich-T-kid/datafusion
that referenced
this pull request
Apr 21, 2026
…#21200) ## Which issue does this PR close? <!-- We generally require a GitHub issue to be filed for all bug fixes and enhancements and this helps us generate change logs for our releases. You can link an issue to this PR using the GitHub syntax. For example `Closes apache#123` indicates that this PR will close issue apache#123. --> - Closes #. ## Rationale for this change <!-- Why are you proposing this change? If this is already explained clearly in the issue then this section is not needed. Explaining clearly why changes are proposed helps reviewers understand your changes and offer better suggestions for fixes. --> Our SMJ benchmark queries finish too quickly to demonstrate improvements that aren't massive. For example, I am working on an optimization that introduces `DynComparator` (part of apache#20910) and it's about a 10% improvement, but only when you actually make the queries run long enough. The new queries for apache#21184 are scaled enough to see improvements, but we need to scale the older queries. I am also continuing to see SMJ issues with Comet when running joins with billions (sometimes trillions) of rows. We can't do that for microbenchmarks, but we can at least start hitting millions of rows to look at more than a handful of batches. ## What changes are included in this PR? <!-- There is no need to duplicate the description in the issue here but it is sometimes worth providing a summary of the individual changes in this PR. --> Bring our SMJ queries into alignment with some of the newer ones (Q21-23) to demonstrate further performance wins. ## Are these changes tested? <!-- We typically require tests for all PRs in order to: 1. Prevent the code from being accidentally broken by subsequent changes 2. Serve as another way to document the expected behavior of the code If tests are not included in your PR, please explain why (for example, are they covered by existing tests)? --> I ran the benchmark. On my M3 Max, here's how long it takes: | Query | Join Type | Rows | Keys | Filter | Median (ms) | |-------|-----------|------|------|--------|-------------| | Q1 | INNER | 1M×1M | 1:1 | — | 16.3 | | Q2 | INNER | 1M×10M | 1:10 | — | 117.4 | | Q3 | INNER | 1M×1M | 1:100 | — | 74.2 | | Q4 | INNER | 1M×10M | 1:10 | 1% | 17.1 | | Q5 | INNER | 1M×1M | 1:100 | 10% | 18.4 | | Q6 | LEFT | 1M×10M | 1:10 | — | 129.3 | | Q7 | LEFT | 1M×10M | 1:10 | 50% | 150.2 | | Q8 | FULL | 1M×1M | 1:10 | — | 16.6 | | Q9 | FULL | 1M×10M | 1:10 | 10% | 153.5 | | Q10 | LEFT SEMI | 1M×10M | 1:10 | — | 53.1 | | Q11 | LEFT SEMI | 1M×10M | 1:10 | 1% | 15.5 | | Q12 | LEFT SEMI | 1M×10M | 1:10 | 50% | 65.0 | | Q13 | LEFT SEMI | 1M×10M | 1:10 | 90% | 105.7 | | Q14 | LEFT ANTI | 1M×10M | 1:10 | — | 54.3 | | Q15 | LEFT ANTI | 1M×10M | 1:10 | partial | 51.5 | | Q16 | LEFT ANTI | 1M×1M | 1:1 | — | 10.3 | | Q17 | INNER | 1M×50M | 1:50 | 5% | 75.9 | | Q18 | LEFT SEMI | 1M×50M | 1:50 | 2% | 50.2 | | Q19 | LEFT ANTI | 1M×50M | 1:50 | partial | 336.4 | | Q20 | INNER | 1M×10M | 1:100 | GROUP BY | 763.7 | | Q21 | INNER | 10M×10M | 1:1 | 50% | 186.1 | | Q22 | LEFT | 10M×10M | 1:1 | 50% | 10,193.8 | | Q23 | FULL | 10M×10M | 1:1 | 50% | 10,194.7 | Note that Q22 and Q23 will be about 20x faster when apache#21184 merges, so taking 10 seconds to run is just a short-term issue. ## Are there any user-facing changes? <!-- If there are user-facing changes then we may require documentation to be updated before approving the PR. --> <!-- If there are any breaking changes to public APIs, please add the `api change` label. --> No.
Rich-T-kid
pushed a commit
to Rich-T-kid/datafusion
that referenced
this pull request
Apr 21, 2026
…ins to bitwise stream. Near-unique LEFT and FULL SMJ 20-50x faster (apache#21184) ## Which issue does this PR close? Partially addresses apache#20910. Fixes apache#21197. ## Rationale for this change Sort-merge join with a filter on outer joins (LEFT/RIGHT/FULL) runs `process_filtered_batches()` on every key transition in the Init state. With near-unique keys (1:1 cardinality), this means running the full deferred filtering pipeline (concat + `get_corrected_filter_mask` + `filter_record_batch_by_join_type`) once per row — making filtered LEFT/RIGHT/FULL **55x slower** than INNER for 10M unique keys. Additionally, mark join logic in `MaterializingSortMergeJoinStream` materializes full `(streamed, buffered)` pairs only to discard most of them via `get_corrected_filter_mask()`. Mark joins are structurally identical to semi joins (one output row per outer row with a boolean result) and belong in `BitwiseSortMergeJoinStream`, which avoids pair materialization entirely using a per-outer-batch bitset. ## What changes are included in this PR? Three areas of improvement, building on the specialized semi/anti stream from apache#20806: **1. Move mark joins to `BitwiseSortMergeJoinStream`** - Match on join type; `emit_outer_batch()` emits all rows with the match bitset as a boolean column (vs semi's filter / anti's invert-and-filter) - Route `LeftMark`/`RightMark` from `SortMergeJoinExec::execute()` to the bitwise stream - Remove all mark-specific logic from `MaterializingSortMergeJoinStream` (`mark_row_as_match`, `is_not_null` column generation, mark arms in filter correction) **2. Batch filter evaluation in `freeze_streamed()`** - Split `freeze_streamed()` into null-joined classification + `freeze_streamed_matched()` for batched materialization - Collect indices across chunks, materialize left/right columns once using tiered Arrow kernels (`slice` → `take` → `interleave`) - Single `RecordBatch` construction and single `expression.evaluate()` per freeze instead of per chunk - Vectorize `append_filter_metadata()` using builder `extend()` instead of per-element loop **3. Batch deferred filtering in Init state** (this is the big win for Q22 and Q23) - Gate `process_filtered_batches()` on accumulated rows >= `batch_size` instead of running on every Init entry - Accumulated data bounded to ~2×batch_size (one from `freeze_dequeuing_buffered`, one accumulating toward next freeze) — does not reintroduce unbounded buffering fixed by PR apache#20482 - `Exhausted` state flushes any remainder **Cleanup:** - Rename `SortMergeJoinStream` → `MaterializingSortMergeJoinStream` (materializes explicit row pairs for join output) and `SemiAntiMarkSortMergeJoinStream` → `BitwiseSortMergeJoinStream` (tracks matches via boolean bitset) - Consolidate `semi_anti_mark_sort_merge_join/` into `sort_merge_join/` as `bitwise_stream.rs` / `bitwise_tests.rs`; rename `stream.rs` → `materializing_stream.rs` and `tests.rs` → `materializing_tests.rs` - Consolidate `SpillManager` construction into `SortMergeJoinExec::execute()` (shared across both streams); move `peak_mem_used` gauge into `BitwiseSortMergeJoinStream::try_new` - `MaterializingSortMergeJoinStream` now handles only Inner/Left/Right/Full — all semi/anti/mark branching removed - `get_corrected_filter_mask()`: merge identical Left/Right/Full branches; add null-metadata passthrough for already-null-joined rows - `filter_record_batch_by_join_type()`: rewrite from `filter(true) + filter(false) + concat` to `zip()` for in-place null-joining — preserves row ordering and removes `create_null_joined_batch()` entirely; add early return for empty batches - `filter_record_batch_by_join_type()`: use `compute::filter()` directly on `BooleanArray` instead of wrapping in temporary `RecordBatch` ## Benchmarks `cargo run --release --bin dfbench -- smj` | Query | Join Type | Rows | Keys | Filter | Main (ms) | PR (ms) | Speedup | |-------|-----------|------|------|--------|-----------|---------|---------| | Q1 | INNER | 1M×1M | 1:1 | — | 16.3 | 14.4 | 1.1x | | Q2 | INNER | 1M×10M | 1:10 | — | 117.4 | 120.1 | 1.0x | | Q3 | INNER | 1M×1M | 1:100 | — | 74.2 | 66.6 | 1.1x | | Q4 | INNER | 1M×10M | 1:10 | 1% | 17.1 | 15.1 | 1.1x | | Q5 | INNER | 1M×1M | 1:100 | 10% | 18.4 | 14.4 | 1.3x | | Q6 | LEFT | 1M×10M | 1:10 | — | 129.3 | 122.7 | 1.1x | | Q7 | LEFT | 1M×10M | 1:10 | 50% | 150.2 | 142.2 | 1.1x | | Q8 | FULL | 1M×1M | 1:10 | — | 16.6 | 16.7 | 1.0x | | Q9 | FULL | 1M×10M | 1:10 | 10% | 153.5 | 136.2 | 1.1x | | Q10 | LEFT SEMI | 1M×10M | 1:10 | — | 53.1 | 53.1 | 1.0x | | Q11 | LEFT SEMI | 1M×10M | 1:10 | 1% | 15.5 | 14.7 | 1.1x | | Q12 | LEFT SEMI | 1M×10M | 1:10 | 50% | 65.0 | 67.3 | 1.0x | | Q13 | LEFT SEMI | 1M×10M | 1:10 | 90% | 105.7 | 109.8 | 1.0x | | Q14 | LEFT ANTI | 1M×10M | 1:10 | — | 54.3 | 53.9 | 1.0x | | Q15 | LEFT ANTI | 1M×10M | 1:10 | partial | 51.5 | 50.5 | 1.0x | | Q16 | LEFT ANTI | 1M×1M | 1:1 | — | 10.3 | 11.3 | 0.9x | | Q17 | INNER | 1M×50M | 1:50 | 5% | 75.9 | 79.0 | 1.0x | | Q18 | LEFT SEMI | 1M×50M | 1:50 | 2% | 50.2 | 49.0 | 1.0x | | Q19 | LEFT ANTI | 1M×50M | 1:50 | partial | 336.4 | 344.2 | 1.0x | | Q20 | INNER | 1M×10M | 1:100 | GROUP BY | 763.7 | 803.9 | 1.0x | | Q21 | INNER | 10M×10M | 1:1 | 50% | 186.1 | 187.8 | 1.0x | | Q22 | LEFT | 10M×10M | 1:1 | 50% | 10,193.8 | 185.8 | **54.9x** | | Q23 | FULL | 10M×10M | 1:1 | 50% | 10,194.7 | 233.6 | **43.6x** | | Q24 | LEFT MARK | 1M×10M | 1:10 | 1% | FAILS | 15.1 | — | | Q25 | LEFT MARK | 1M×10M | 1:10 | 50% | FAILS | 67.3 | — | | Q26 | LEFT MARK | 1M×10M | 1:10 | 90% | FAILS | 110.0 | — | General workload (Q1-Q20, various join types/cardinalities/selectivities): no regressions. ## Are these changes tested? In addition to existing unit and sqllogictests: - I ran 50 iterations of the fuzz tests (modified to only test against hash join as the baseline because nested loop join takes too long) `cargo test -p datafusion --features extended_tests --test fuzz -- join_fuzz` - One new sqllogictest for apache#21197 that fails on main - Four new unit tests: three for full join with filter that spills - One new fuzz test to exercise full join with filter that spills - New benchmark queries Q21-Q23: 10M×10M unique keys with 50% join filter for INNER/LEFT/FULL — exercises the degenerate case this PR fixes - New benchmark queries Q24-Q26 duplicated Q11-Q13 but for Mark joins, showing that they have the same performance as other joins (`LeftSemi`) that use this stream ## Are there any user-facing changes? No.
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
7 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.
Which issue does this PR close?
Partially addresses #20910. Fixes #21197.
Rationale for this change
Sort-merge join with a filter on outer joins (LEFT/RIGHT/FULL) runs
process_filtered_batches()on every key transition in the Init state. With near-unique keys (1:1 cardinality), this means running the full deferred filtering pipeline (concat +get_corrected_filter_mask+filter_record_batch_by_join_type) once per row — making filtered LEFT/RIGHT/FULL 55x slower than INNER for 10M unique keys.Additionally, mark join logic in
MaterializingSortMergeJoinStreammaterializes full(streamed, buffered)pairs only to discard most of them viaget_corrected_filter_mask(). Mark joins are structurally identical to semi joins (one output row per outer row with a boolean result) and belong inBitwiseSortMergeJoinStream, which avoids pair materialization entirely using a per-outer-batch bitset.What changes are included in this PR?
Three areas of improvement, building on the specialized semi/anti stream from #20806:
1. Move mark joins to
BitwiseSortMergeJoinStreamemit_outer_batch()emits all rows with the match bitset as a boolean column (vs semi's filter / anti's invert-and-filter)LeftMark/RightMarkfromSortMergeJoinExec::execute()to the bitwise streamMaterializingSortMergeJoinStream(mark_row_as_match,is_not_nullcolumn generation, mark arms in filter correction)2. Batch filter evaluation in
freeze_streamed()freeze_streamed()into null-joined classification +freeze_streamed_matched()for batched materializationslice→take→interleave)RecordBatchconstruction and singleexpression.evaluate()per freeze instead of per chunkappend_filter_metadata()using builderextend()instead of per-element loop3. Batch deferred filtering in Init state (this is the big win for Q22 and Q23)
process_filtered_batches()on accumulated rows >=batch_sizeinstead of running on every Init entryfreeze_dequeuing_buffered, one accumulating toward next freeze) — does not reintroduce unbounded buffering fixed by PR fix: SortMergeJoin don't wait for all input before emitting #20482Exhaustedstate flushes any remainderCleanup:
SortMergeJoinStream→MaterializingSortMergeJoinStream(materializes explicit row pairs for join output) andSemiAntiMarkSortMergeJoinStream→BitwiseSortMergeJoinStream(tracks matches via boolean bitset)semi_anti_mark_sort_merge_join/intosort_merge_join/asbitwise_stream.rs/bitwise_tests.rs; renamestream.rs→materializing_stream.rsandtests.rs→materializing_tests.rsSpillManagerconstruction intoSortMergeJoinExec::execute()(shared across both streams); movepeak_mem_usedgauge intoBitwiseSortMergeJoinStream::try_newMaterializingSortMergeJoinStreamnow handles only Inner/Left/Right/Full — all semi/anti/mark branching removedget_corrected_filter_mask(): merge identical Left/Right/Full branches; add null-metadata passthrough for already-null-joined rowsfilter_record_batch_by_join_type(): rewrite fromfilter(true) + filter(false) + concattozip()for in-place null-joining — preserves row ordering and removescreate_null_joined_batch()entirely; add early return for empty batchesfilter_record_batch_by_join_type(): usecompute::filter()directly onBooleanArrayinstead of wrapping in temporaryRecordBatchBenchmarks
cargo run --release --bin dfbench -- smjGeneral workload (Q1-Q20, various join types/cardinalities/selectivities): no regressions.
Are these changes tested?
In addition to existing unit and sqllogictests:
cargo test -p datafusion --features extended_tests --test fuzz -- join_fuzzLeftSemi) that use this streamAre there any user-facing changes?
No.