Repository navigation
perf(arrow): fuse Boolean filter copying - #11918
Open
Rich-T-kid wants to merge 2 commits into
Open
Rich-T-kid wants to merge 2 commits into
Rich-T-kid wants to merge 2 commits into
Conversation
Contributor
Author
|
run benchmark coalesce_kernels |
|
🤖 Benchmark starting (GKE) | trigger Target: coalesce_kernels Sharding factor: 1 (1 workers). Comparing rich-T-kid/coalescer-boolean-filter-words (2637e5b) to 69be95c (merge-base) diff Run configurationrun benchmark coalesce_kernels
env:
BENCH_FILTER: single_boolean
shards: 1
Results will be posted when all workers finish. File an issue against this benchmark runner |
github-actions
Bot
added
the
arrow-kernels
|
🤖 Benchmark completed (GKE) | trigger Target: coalesce_kernels Comparing rich-T-kid/coalescer-boolean-filter-words (2637e5b) to 69be95c (merge-base) diff Run configurationrun benchmark coalesce_kernels
env:
BENCH_FILTER: single_boolean
shards: 1
DetailsPer-runner informationcoalesce_kernels — shard 1/1Node: gk3-benchmark-cluster-nap-2bx3iakq-2158138a-9hhg Instance: c4a-highmem-16 (12 vCPU / 65 GiB) uname: BENCH_COMMAND: cargo bench --features=arrow,async,test_common,experimental,object_store --bench coalesce_kernelsCPU Details (lscpu)Resource Usagebase (merge-base)
branch
File an issue against this benchmark runner |
Rich-T-kid
force-pushed
the
rich-T-kid/coalescer-boolean-filter-words
branch
3 times, most recently
from
October 26, 2026 14:58
154cace to
56c6e8c
Compare
Contributor
Author
|
run benchmark coalesce_kernels |
|
🤖 Benchmark starting (GKE) | trigger Target: coalesce_kernels Sharding factor: 1 (1 workers). Comparing rich-T-kid/coalescer-boolean-filter-words (967d3d0) to 69be95c (merge-base) diff Run configurationrun benchmark coalesce_kernels
env:
BENCH_FILTER: single_boolean
shards: 1
Results will be posted when all workers finish. File an issue against this benchmark runner |
|
🤖 Benchmark completed (GKE) | trigger Target: coalesce_kernels Comparing rich-T-kid/coalescer-boolean-filter-words (967d3d0) to 69be95c (merge-base) diff Run configurationrun benchmark coalesce_kernels
env:
BENCH_FILTER: single_boolean
shards: 1
DetailsPer-runner informationcoalesce_kernels — shard 1/1Node: gk3-benchmark-cluster-nap-1tw271hz-870aa374-gtwm Instance: c4a-highmem-16 (12 vCPU / 65 GiB) uname: BENCH_COMMAND: cargo bench --features=arrow,async,test_common,experimental,object_store --bench coalesce_kernelsCPU Details (lscpu)Resource Usagebase (merge-base)
branch
File an issue against this benchmark runner |
Rich-T-kid
force-pushed
the
rich-T-kid/coalescer-boolean-filter-words
branch
from
October 26, 2026 15:49
967d3d0 to
f66a20c
Compare
Contributor
Author
|
run benchmark coalesce_kernels |
|
🤖 Benchmark starting (GKE) | trigger Target: coalesce_kernels Sharding factor: 1 (1 workers). Comparing rich-T-kid/coalescer-boolean-filter-words (f66a20c) to 69be95c (merge-base) diff Run configurationrun benchmark coalesce_kernels
env:
BENCH_FILTER: single_boolean
shards: 1
Results will be posted when all workers finish. File an issue against this benchmark runner |
Rich-T-kid
force-pushed
the
rich-T-kid/coalescer-boolean-filter-words
branch
2 times, most recently
from
October 26, 2026 16:11
e9faa76 to
0efb530
Compare
|
🤖 Benchmark completed (GKE) | trigger Target: coalesce_kernels Comparing rich-T-kid/coalescer-boolean-filter-words (f66a20c) to 69be95c (merge-base) diff Run configurationrun benchmark coalesce_kernels
env:
BENCH_FILTER: single_boolean
shards: 1
DetailsPer-runner informationcoalesce_kernels — shard 1/1Node: gk3-benchmark-cluster-nap-1ken38iy-4504e04d-ct9z Instance: c4a-highmem-16 (12 vCPU / 65 GiB) uname: BENCH_COMMAND: cargo bench --features=arrow,async,test_common,experimental,object_store --bench coalesce_kernelsCPU Details (lscpu)Resource Usagebase (merge-base)
branch
File an issue against this benchmark runner |
Rich-T-kid
force-pushed
the
rich-T-kid/coalescer-boolean-filter-words
branch
from
October 26, 2026 16:33
0efb530 to
2502bea
Compare
Contributor
Author
|
run benchmark coalesce_kernels |
|
🤖 Benchmark starting (GKE) | trigger Target: coalesce_kernels Sharding factor: 1 (1 workers). Comparing rich-T-kid/coalescer-boolean-filter-words (2502bea) to 69be95c (merge-base) diff Run configurationrun benchmark coalesce_kernels
env:
BENCH_FILTER: single_boolean
shards: 1
Results will be posted when all workers finish. File an issue against this benchmark runner |
|
🤖 Benchmark completed (GKE) | trigger Target: coalesce_kernels Comparing rich-T-kid/coalescer-boolean-filter-words (2502bea) to 69be95c (merge-base) diff Run configurationrun benchmark coalesce_kernels
env:
BENCH_FILTER: single_boolean
shards: 1
DetailsPer-runner informationcoalesce_kernels — shard 1/1Node: gk3-benchmark-cluster-nap-1exwvsr2-39029fda-8h8s Instance: c4a-highmem-16 (12 vCPU / 65 GiB) uname: BENCH_COMMAND: cargo bench --features=arrow,async,test_common,experimental,object_store --bench coalesce_kernelsCPU Details (lscpu)Resource Usagebase (merge-base)
branch
File an issue against this benchmark runner |
Rich-T-kid
force-pushed
the
rich-T-kid/coalescer-boolean-filter-words
branch
from
October 26, 2026 17:32
2502bea to
a88e90c
Compare
Contributor
Author
created #11921 for this |
Rich-T-kid
force-pushed
the
rich-T-kid/coalescer-boolean-filter-words
branch
from
October 26, 2026 17:44
a88e90c to
05de52d
Compare
Pack selected Boolean values and validity bits directly into the in-progress builders, avoiding temporary filtered arrays. Check filter masks before reading source words so sparse inputs skip empty chunks. Use separate loops for sources with and without actual nulls, and carry compressed fragments across chunk boundaries while preserving lazy validity allocation.
Rich-T-kid
force-pushed
the
rich-T-kid/coalescer-boolean-filter-words
branch
from
October 26, 2026 18:09
05de52d to
16e65c3
Compare
Contributor
Author
|
run benchmark coalesce_kernels |
|
🤖 Benchmark starting (GKE) | trigger Target: coalesce_kernels Sharding factor: 1 (1 workers). Comparing rich-T-kid/coalescer-boolean-filter-words (16e65c3) to 69be95c (merge-base) diff Run configurationrun benchmark coalesce_kernels
env:
BENCH_FILTER: single_boolean
shards: 1
Results will be posted when all workers finish. File an issue against this benchmark runner |
Rich-T-kid
marked this pull request as ready for review
October 26, 2026 18:12
Contributor
Author
|
@mbutrovich if your interested this is ready for review. Should be the final PR related to booleans in the coalescer kernel & it implements the ideas you mentioned here #11732 (comment) 🚀 |
mbutrovich
reviewed
Oct 26, 2026
mbutrovich
left a comment
mbutrovich
left a comment
Contributor
There was a problem hiding this comment.
Thanks @Rich-T-kid, this is the order we talked about in #11732, and the sparse-filter numbers from the bot look good. Filing #11921 for the dense cases is the right call.
|
🤖 Benchmark completed (GKE) | trigger Target: coalesce_kernels Comparing rich-T-kid/coalescer-boolean-filter-words (16e65c3) to 69be95c (merge-base) diff Run configurationrun benchmark coalesce_kernels
env:
BENCH_FILTER: single_boolean
shards: 1
DetailsPer-runner informationcoalesce_kernels — shard 1/1Node: gk3-benchmark-cluster-nap-s8dlj0s2-749b0757-kqjj Instance: c4a-highmem-16 (12 vCPU / 65 GiB) uname: BENCH_COMMAND: cargo bench --features=arrow,async,test_common,experimental,object_store --bench coalesce_kernelsCPU Details (lscpu)Resource Usagebase (merge-base)
branch
File an issue against this benchmark runner |
Contributor
Author
|
run benchmark coalesce_kernels |
1 similar comment
Contributor
Author
|
run benchmark coalesce_kernels |
|
🤖 Benchmark starting (GKE) | trigger Target: coalesce_kernels Sharding factor: 1 (1 workers). Comparing rich-T-kid/coalescer-boolean-filter-words (176af76) to 69be95c (merge-base) diff Run configurationrun benchmark coalesce_kernels
env:
BENCH_FILTER: single_boolean
shards: 1
Results will be posted when all workers finish. File an issue against this benchmark runner |
|
🤖 Benchmark starting (GKE) | trigger Target: coalesce_kernels Sharding factor: 1 (1 workers). Comparing rich-T-kid/coalescer-boolean-filter-words (176af76) to 69be95c (merge-base) diff Run configurationrun benchmark coalesce_kernels
env:
BENCH_FILTER: single_boolean
shards: 1
Results will be posted when all workers finish. File an issue against this benchmark runner |
|
🤖 Benchmark completed (GKE) | trigger Target: coalesce_kernels Comparing rich-T-kid/coalescer-boolean-filter-words (176af76) to 69be95c (merge-base) diff Run configurationrun benchmark coalesce_kernels
env:
BENCH_FILTER: single_boolean
shards: 1
DetailsPer-runner informationcoalesce_kernels — shard 1/1Node: gk3-benchmark-cluster-nap-s8dlj0s2-9d742a00-t68b Instance: c4a-highmem-16 (12 vCPU / 65 GiB) uname: BENCH_COMMAND: cargo bench --features=arrow,async,test_common,experimental,object_store --bench coalesce_kernelsCPU Details (lscpu)Resource Usagebase (merge-base)
branch
File an issue against this benchmark runner |
|
🤖 Benchmark completed (GKE) | trigger Target: coalesce_kernels Comparing rich-T-kid/coalescer-boolean-filter-words (176af76) to 69be95c (merge-base) diff Run configurationrun benchmark coalesce_kernels
env:
BENCH_FILTER: single_boolean
shards: 1
DetailsPer-runner informationcoalesce_kernels — shard 1/1Node: gk3-benchmark-cluster-nap-zknmidoc-9e5bd0bb-8jts Instance: c4a-highmem-16 (12 vCPU / 65 GiB) uname: BENCH_COMMAND: cargo bench --features=arrow,async,test_common,experimental,object_store --bench coalesce_kernelsCPU Details (lscpu)Resource Usagebase (merge-base)
branch
File an issue against this benchmark runner |
Rich-T-kid
force-pushed
the
rich-T-kid/coalescer-boolean-filter-words
branch
from
October 26, 2026 20:42
176af76 to
9b48802
Compare
Contributor
Author
|
run benchmark coalesce_kernels |
|
🤖 Benchmark starting (GKE) | trigger Target: coalesce_kernels Sharding factor: 1 (1 workers). Comparing rich-T-kid/coalescer-boolean-filter-words (9b48802) to 69be95c (merge-base) diff Run configurationrun benchmark coalesce_kernels
env:
BENCH_FILTER: single_boolean
shards: 1
Results will be posted when all workers finish. File an issue against this benchmark runner |
|
🤖 Benchmark completed (GKE) | trigger Target: coalesce_kernels Comparing rich-T-kid/coalescer-boolean-filter-words (9b48802) to 69be95c (merge-base) diff Run configurationrun benchmark coalesce_kernels
env:
BENCH_FILTER: single_boolean
shards: 1
DetailsPer-runner informationcoalesce_kernels — shard 1/1Node: gk3-benchmark-cluster-nap-1bd6zsy7-b5970f85-bs68 Instance: c4a-highmem-16 (12 vCPU / 65 GiB) uname: BENCH_COMMAND: cargo bench --features=arrow,async,test_common,experimental,object_store --bench coalesce_kernelsCPU Details (lscpu)Resource Usagebase (merge-base)
branch
File an issue against this benchmark runner |
Rich-T-kid
force-pushed
the
rich-T-kid/coalescer-boolean-filter-words
branch
from
October 26, 2026 21:44
9b48802 to
e22fa80
Compare
Contributor
Author
|
run benchmark coalesce_kernels |
|
🤖 Benchmark starting (GKE) | trigger Target: coalesce_kernels Sharding factor: 1 (1 workers). Comparing rich-T-kid/coalescer-boolean-filter-words (e22fa80) to 69be95c (merge-base) diff Run configurationrun benchmark coalesce_kernels
env:
BENCH_FILTER: single_boolean
shards: 1
Results will be posted when all workers finish. File an issue against this benchmark runner |
Let the Boolean coalescer append words produced by the existing compression kernel instead of maintaining a second copy of its packing loop. Support values and validity in one mask traversal, and cover shorter filters and exact word boundaries.
Rich-T-kid
force-pushed
the
rich-T-kid/coalescer-boolean-filter-words
branch
from
October 26, 2026 21:45
e22fa80 to
4d091cd
Compare
Contributor
Author
|
run benchmark coalesce_kernels |
|
🤖 Benchmark starting (GKE) | trigger Target: coalesce_kernels Sharding factor: 1 (1 workers). Comparing rich-T-kid/coalescer-boolean-filter-words (4d091cd) to 69be95c (merge-base) diff Run configurationrun benchmark coalesce_kernels
env:
BENCH_FILTER: single_boolean
shards: 1
Results will be posted when all workers finish. File an issue against this benchmark runner |
Contributor
Author
|
run benchmark filter_kernels |
|
🤖 Benchmark starting (GKE) | trigger Target: filter_kernels Sharding factor: 1 (1 workers). Comparing rich-T-kid/coalescer-boolean-filter-words (4d091cd) to 69be95c (merge-base) diff Run configurationrun benchmark filter_kernels
shards: 1
Results will be posted when all workers finish. File an issue against this benchmark runner |
|
🤖 Benchmark completed (GKE) | trigger Target: coalesce_kernels Comparing rich-T-kid/coalescer-boolean-filter-words (e22fa80) to 69be95c (merge-base) diff Run configurationrun benchmark coalesce_kernels
env:
BENCH_FILTER: single_boolean
shards: 1
DetailsPer-runner informationcoalesce_kernels — shard 1/1Node: gk3-benchmark-cluster-nap-a7pbla7y-9f7e0324-m8jd Instance: c4a-highmem-16 (12 vCPU / 65 GiB) uname: BENCH_COMMAND: cargo bench --features=arrow,async,test_common,experimental,object_store --bench coalesce_kernelsCPU Details (lscpu)Resource Usagebase (merge-base)
branch
File an issue against this benchmark runner |
|
🤖 Benchmark completed (GKE) | trigger Target: coalesce_kernels Comparing rich-T-kid/coalescer-boolean-filter-words (4d091cd) to 69be95c (merge-base) diff Run configurationrun benchmark coalesce_kernels
env:
BENCH_FILTER: single_boolean
shards: 1
DetailsPer-runner informationcoalesce_kernels — shard 1/1Node: gk3-benchmark-cluster-nap-1rom390t-571e000f-ggq5 Instance: c4a-highmem-16 (12 vCPU / 65 GiB) uname: BENCH_COMMAND: cargo bench --features=arrow,async,test_common,experimental,object_store --bench coalesce_kernelsCPU Details (lscpu)Resource Usagebase (merge-base)
branch
File an issue against this benchmark runner |
|
🤖 Benchmark completed (GKE) | trigger Target: filter_kernels Comparing rich-T-kid/coalescer-boolean-filter-words (4d091cd) to 69be95c (merge-base) diff Run configurationrun benchmark filter_kernels
shards: 1
DetailsPer-runner informationfilter_kernels — shard 1/1Node: gk3-benchmark-cluster-nap-1rom390t-571e000f-jrjb Instance: c4a-highmem-16 (12 vCPU / 65 GiB) uname: BENCH_COMMAND: cargo bench --features=arrow,async,test_common,experimental,object_store --bench filter_kernelsCPU Details (lscpu)Resource Usagebase (merge-base)
branch
File an issue against this benchmark runner |
This branch has not been deployed
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
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.
Closes
Why
Follow-up to #11732 and #11867.
copy_rows_by_filter_frompreviously ran the Boolean filter kernel, producing temporary value and validity buffers. The coalescer immediately copied those buffers into its builders and discarded them. Now that the builders can append packed words, the filter kernel can write straight into the coalescer output.What changed
The Boolean coalescer now uses the existing word-at-a-time compression kernel. Non-null and all-valid sources append values and advance validity lazily. Nullable sources compress values and validity together in one pass over the filter mask.
Empty filter words are skipped before reading the corresponding source words, which avoids touching most of the source for sparse filters. Sliced bitmaps, a filter shorter than the source, and an empty remainder are handled by the same chunked path.
Note on
filter.rsfilter_bits_compressnow emits packed words through a callback instead of owning its output buffer. The regular Boolean filter still collects those words into aBufferthroughcollect_filter_bits_compress, so its strategy selection and result are unchanged. The paired form lets the coalescer compress values and validity with one mask traversal. These helpers remain crate-private.Example
Take this eight-row input:
The null predicate at row 3 is normalized to
false, so rows 1, 2, 4, and 6 are selected. With row 0 in the least-significant bit:The compression kernel produces:
Those four bits are appended directly to the output builders. The result is:
Larger inputs use the same operation on each 64-row chunk, carrying partial words between chunks.
Results
Test coverage
The tests compare the coalescer output with the Boolean filter kernel for nullable, non-null, and all-valid sources. They cover sliced inputs, null predicate entries, word boundaries, carry between words, a filter shorter than the source, an exact multiple of 64, and optimized and unoptimized predicates. A filter longer than the source is checked separately.