Add has_true() and has_false() to BooleanArray by adriangb · Pull Request #9511 · apache/arrow-rs

adriangb · 2026-03-05T10:26:39Z

Motivation

When working with BooleanArray, a common pattern is checking whether any true or false value exists — e.g.
arr.true_count() > 0 or arr.false_count() == 0. This currently requires true_count() / false_count(), which scan the entire bitmap to count every set bit (via popcount), even though we only need to know if at least one exists.

This PR adds has_true() and has_false() methods that short-circuit as soon as they find a matching value, providing both:

Better performance — faster on large arrays in the best case
More ergonomic API — arr.has_true() expresses intent more clearly than arr.true_count() > 0

Callsites in DataFusion

There are several places in DataFusion that would benefit from these methods:

datafusion/functions-nested/src/array_has.rs — eq_array.true_count() > 0 → eq_array.has_true()
datafusion/physical-plan/src/topk/mod.rs — filter.true_count() == 0 check → !filter.has_true()
datafusion/datasource-parquet/src/metadata.rs — exactness.true_count() == 0 and combined_mask.true_count() > 0
datafusion/physical-plan/src/joins/nested_loop_join.rs — bitmap.true_count() == 0 checks
datafusion/physical-expr-common/src/physical_expr.rs — selection_count == 0 from selection.true_count()
datafusion/physical-expr/src/expressions/binary.rs — short-circuit checks for AND/OR

Benchmark Results

Scenario                          true_count     has_true       has_false      Speedup (best)
─────────────────────────────────────────────────────────────────────────────────────────────
all_true, 64                      4.32 ns        4.08 ns        4.76 ns        ~1.1x
all_false, 64                     4.30 ns        4.25 ns        4.52 ns        ~1.0x
all_true, 1024                    5.15 ns        4.52 ns        4.99 ns        ~1.1x
all_false, 1024                   5.17 ns        4.55 ns        5.00 ns        ~1.1x
mixed_early, 1024                 5.22 ns        —              5.04 ns        ~1.0x
nulls_all_true, 1024              12.84 ns       4.10 ns        12.92 ns       ~3.1x
all_true, 65536                   100.06 ns      5.96 ns        49.70 ns       ~16.8x (has_true)
all_false, 65536                  99.33 ns       49.30 ns       6.19 ns        ~16.0x (has_false)
mixed_early, 65536                100.10 ns      —              6.20 ns        ~16.1x (has_false)
nulls_all_true, 65536             522.94 ns      4.05 ns        521.82 ns      ~129x (has_true)

The key wins are on larger arrays (65,536 elements), where has_true/has_false are up to 16-129x faster than
true_count() in best-case scenarios (early short-circuit). Even in worst case (must scan entire array), performance is
comparable to true_count.

Implementation

The implementation processes bits in 64-bit chunks using UnalignedBitChunk, which handles arbitrary bit offsets and aligns
data for SIMD-friendly processing.

has_true (no nulls): OR-folds 64-bit chunks, short-circuits when any bit is set
has_false (no nulls): AND-folds 64-bit chunks, short-circuits when any bit is unset (with padding bits masked to 1)
With nulls: Iterates paired (null, value) chunks, checking null & value != 0 (has_true) or null & !value != 0
(has_false)

Alternatives considered

Fully vectorized (no early stopping): Would process the entire bitmap like true_count() but with simpler bitwise ops
instead of popcount. Marginally faster than true_count() but misses the main optimization opportunity.
Per-element iteration with early stopping: self.iter().any(|v| v == Some(true)). Simple but processes one bit at a
time, missing SIMD vectorization of the inner loop. Our approach processes 64 bits at a time while still supporting early
exit.

The chosen approach balances SIMD-friendly bulk processing (64 bits per iteration) with early termination, giving the best of
both worlds.

Test Plan

Unit tests covering: all-true, all-false, mixed, empty, nullable (all-valid-true, all-valid-false, all-null), non-aligned
lengths (65 elements, 64+1 with trailing false)
Criterion benchmarks comparing has_true/has_false vs true_count across sizes (64, 1024, 65536) and data distributions

🤖 Generated with [Claude Code](https://claude.com/claude-code

Short-circuiting methods that return early on the first matching chunk, avoiding full popcount scans. Useful for replacing common patterns like `true_count() == 0` or `true_count() == len`. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Switch the no-nulls paths from BitChunks::iter_padded() (opaque iterator, prevents auto-vectorization) to UnalignedBitChunk (aligned &[u64] slice that LLVM can vectorize). Process chunks in blocks of 64 u64s with a fold + short-circuit between blocks. Worst-case full-scan at 65536 elements drops from ~255ns to ~50ns (5x), now 2x faster than true_count() (~100ns) thanks to simpler per-element ops (OR/AND fold vs popcount). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

adriangb · 2026-03-05T10:32:30Z

run benchmark boolean_array

alamb-ghbot · 2026-03-05T10:32:39Z

🤖 Hi @adriangb, thanks for the request (#9511 (comment)).

scrape_comments.py only supports whitelisted benchmarks.

Standard: (none)
Criterion: array_from, array_iter, arrow_reader, arrow_reader_clickbench, arrow_reader_row_filter, arrow_statistics, arrow_writer, bitwise_kernel, boolean_kernels, buffer_bit_ops, builder, cast_kernels, coalesce_kernels, comparison_kernels, concatenate_kernel, csv_writer, encoding, filter_kernels, interleave_kernels, ipc_reader, json-reader, metadata, row_format, sort_kernel, take_kernels, union_array, variant_builder, variant_kernels, variant_validation, view_types, zip_kernels

Please choose one or more of these with run benchmark <name> or run benchmark <name1> <name2>...

You can also set environment variables on subsequent lines:

run benchmark tpch_mem
DATAFUSION_RUNTIME_MEMORY_LIMIT=1G

Unsupported benchmarks: boolean_array.

adriangb · 2026-03-05T11:07:59Z

cc @Dandandan

adriangb · 2026-03-12T22:06:23Z

run benchmark boolean_array

alamb-ghbot · 2026-03-12T22:06:31Z

🤖 Hi @adriangb, thanks for the request (#9511 (comment)).

scrape_comments.py only supports whitelisted benchmarks.

Standard: (none)
Criterion: array_from, array_iter, arrow_reader, arrow_reader_clickbench, arrow_reader_row_filter, arrow_statistics, arrow_writer, bitwise_kernel, boolean_kernels, buffer_bit_ops, builder, cast_kernels, coalesce_kernels, comparison_kernels, concatenate_kernel, csv_writer, encoding, filter_kernels, interleave_kernels, ipc_reader, json-reader, json_reader, metadata, row_format, sort_kernel, take_kernels, union_array, variant_builder, variant_kernels, variant_validation, view_types, zip_kernels

Please choose one or more of these with run benchmark <name> or run benchmark <name1> <name2>...

You can also set environment variables on subsequent lines:

run benchmark tpch_mem
DATAFUSION_RUNTIME_MEMORY_LIMIT=1G

Unsupported benchmarks: boolean_array.

adriangb · 2026-03-12T22:09:59Z

run benchmark boolean_array

alamb-ghbot · 2026-03-12T22:10:07Z

🤖 Hi @adriangb, thanks for the request (#9511 (comment)).

scrape_comments.py only supports whitelisted benchmarks.

Standard: (none)
Criterion: array_from, array_iter, arrow_reader, arrow_reader_clickbench, arrow_reader_row_filter, arrow_statistics, arrow_writer, bitwise_kernel, boolean_kernels, buffer_bit_ops, builder, cast_kernels, coalesce_kernels, comparison_kernels, concatenate_kernel, csv_writer, encoding, filter_kernels, interleave_kernels, ipc_reader, json-reader, json_reader, metadata, row_format, sort_kernel, take_kernels, union_array, variant_builder, variant_kernels, variant_validation, view_types, zip_kernels

Please choose one or more of these with run benchmark <name> or run benchmark <name1> <name2>...

You can also set environment variables on subsequent lines:

run benchmark tpch_mem
DATAFUSION_RUNTIME_MEMORY_LIMIT=1G

Unsupported benchmarks: boolean_array.

adriangbot · 2026-03-12T22:18:17Z

Benchmark job started for this request (job bench-c4050518673-174). Results will be posted here when complete.

adriangbot · 2026-03-12T22:18:18Z

Benchmark job started for this request (job bench-c4050498266-175). Results will be posted here when complete.

adriangbot · 2026-03-12T22:20:27Z

🤖 Arrow criterion benchmark running (GKE) | trigger
Linux bench-c4050498266-175-6hrsh 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux
Comparing add-has-true-has-false (1551756) to 5ba4515 (merge-base) diff
BENCH_NAME=boolean_array
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench boolean_array
BENCH_FILTER=
Results will be posted here when complete

adriangbot · 2026-03-12T22:20:47Z

Benchmark for this request failed.

Last 20 lines of output:

Click to expand

    project_record
    record_batch
    regexp_kernels
    row_format
    row_group_index_reader
    row_selection_cursor
    row_selector
    serde
    sort_kernel
    string_dictionary_builder
    string_run_builder
    string_run_iterator
    substring_kernels
    take_kernels
    union_array
    variant_builder
    variant_kernels
    variant_validation
    view_types
    zip_kernels

adriangbot · 2026-03-12T22:21:08Z

🤖 Arrow criterion benchmark running (GKE) | trigger
Linux bench-c4050518673-174-fxsf5 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux
Comparing add-has-true-has-false (1551756) to 5ba4515 (merge-base) diff
BENCH_NAME=boolean_array
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench boolean_array
BENCH_FILTER=
Results will be posted here when complete

adriangbot · 2026-03-12T22:21:15Z

Benchmark for this request failed.

Last 20 lines of output:

Click to expand

    project_record
    record_batch
    regexp_kernels
    row_format
    row_group_index_reader
    row_selection_cursor
    row_selector
    serde
    sort_kernel
    string_dictionary_builder
    string_run_builder
    string_run_iterator
    substring_kernels
    take_kernels
    union_array
    variant_builder
    variant_kernels
    variant_validation
    view_types
    zip_kernels

adriangbot · 2026-03-12T22:21:22Z

🤖 Arrow criterion benchmark running (GKE) | trigger
Linux bench-c4050498266-175-lwtcv 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux
Comparing add-has-true-has-false (1551756) to 5ba4515 (merge-base) diff
BENCH_NAME=boolean_array
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench boolean_array
BENCH_FILTER=
Results will be posted here when complete

adriangbot · 2026-03-12T22:21:28Z

Benchmark for this request failed.

Last 20 lines of output:

Click to expand

    project_record
    record_batch
    regexp_kernels
    row_format
    row_group_index_reader
    row_selection_cursor
    row_selector
    serde
    sort_kernel
    string_dictionary_builder
    string_run_builder
    string_run_iterator
    substring_kernels
    take_kernels
    union_array
    variant_builder
    variant_kernels
    variant_validation
    view_types
    zip_kernels

adriangbot · 2026-03-12T22:21:49Z

🤖 Arrow criterion benchmark running (GKE) | trigger
Linux bench-c4050518673-174-wkwzs 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux
Comparing add-has-true-has-false (1551756) to 5ba4515 (merge-base) diff
BENCH_NAME=boolean_array
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench boolean_array
BENCH_FILTER=
Results will be posted here when complete

adriangbot · 2026-03-12T22:21:55Z

Benchmark for this request failed.

Last 20 lines of output:

Click to expand

    project_record
    record_batch
    regexp_kernels
    row_format
    row_group_index_reader
    row_selection_cursor
    row_selector
    serde
    sort_kernel
    string_dictionary_builder
    string_run_builder
    string_run_iterator
    substring_kernels
    take_kernels
    union_array
    variant_builder
    variant_kernels
    variant_validation
    view_types
    zip_kernels

adriangbot · 2026-03-12T22:22:12Z

🤖 Arrow criterion benchmark running (GKE) | trigger
Linux bench-c4050498266-175-n8f8t 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux
Comparing add-has-true-has-false (1551756) to 5ba4515 (merge-base) diff
BENCH_NAME=boolean_array
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench boolean_array
BENCH_FILTER=
Results will be posted here when complete

adriangbot · 2026-03-12T22:22:18Z

Benchmark for this request failed.

Last 20 lines of output:

Click to expand

    project_record
    record_batch
    regexp_kernels
    row_format
    row_group_index_reader
    row_selection_cursor
    row_selector
    serde
    sort_kernel
    string_dictionary_builder
    string_run_builder
    string_run_iterator
    substring_kernels
    take_kernels
    union_array
    variant_builder
    variant_kernels
    variant_validation
    view_types
    zip_kernels

adriangbot · 2026-03-12T22:22:40Z

🤖 Arrow criterion benchmark running (GKE) | trigger
Linux bench-c4050518673-174-wtbsf 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux
Comparing add-has-true-has-false (1551756) to 5ba4515 (merge-base) diff
BENCH_NAME=boolean_array
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench boolean_array
BENCH_FILTER=
Results will be posted here when complete

adriangbot · 2026-03-12T22:22:46Z

Benchmark for this request failed.

Last 20 lines of output:

Click to expand

    project_record
    record_batch
    regexp_kernels
    row_format
    row_group_index_reader
    row_selection_cursor
    row_selector
    serde
    sort_kernel
    string_dictionary_builder
    string_run_builder
    string_run_iterator
    substring_kernels
    take_kernels
    union_array
    variant_builder
    variant_kernels
    variant_validation
    view_types
    zip_kernels

adriangbot · 2026-03-12T22:23:21Z

🤖 Arrow criterion benchmark running (GKE) | trigger
Linux bench-c4050498266-175-zqtls 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux
Comparing add-has-true-has-false (1551756) to 5ba4515 (merge-base) diff
BENCH_NAME=boolean_array
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench boolean_array
BENCH_FILTER=
Results will be posted here when complete

adriangbot · 2026-03-12T22:23:27Z

Benchmark for this request failed.

Last 20 lines of output:

Click to expand

    project_record
    record_batch
    regexp_kernels
    row_format
    row_group_index_reader
    row_selection_cursor
    row_selector
    serde
    sort_kernel
    string_dictionary_builder
    string_run_builder
    string_run_iterator
    substring_kernels
    take_kernels
    union_array
    variant_builder
    variant_kernels
    variant_validation
    view_types
    zip_kernels

adriangbot · 2026-03-12T22:23:48Z

🤖 Arrow criterion benchmark running (GKE) | trigger
Linux bench-c4050518673-174-zvqnn 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux
Comparing add-has-true-has-false (1551756) to 5ba4515 (merge-base) diff
BENCH_NAME=boolean_array
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench boolean_array
BENCH_FILTER=
Results will be posted here when complete

adriangbot · 2026-03-12T22:23:54Z

Benchmark for this request failed.

Last 20 lines of output:

Click to expand

    project_record
    record_batch
    regexp_kernels
    row_format
    row_group_index_reader
    row_selection_cursor
    row_selector
    serde
    sort_kernel
    string_dictionary_builder
    string_run_builder
    string_run_iterator
    substring_kernels
    take_kernels
    union_array
    variant_builder
    variant_kernels
    variant_validation
    view_types
    zip_kernels

adriangb · 2026-03-12T23:59:22Z

run benchmark record_batch

adriangbot · 2026-03-12T23:59:24Z

Benchmark job started for this request (job bench-c4051194027-185). Results will be posted here when complete.

adriangbot · 2026-03-13T00:02:16Z

🤖 Arrow criterion benchmark running (GKE) | trigger
Linux bench-c4051194027-185-9kzpt 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux
Comparing add-has-true-has-false (1551756) to 5ba4515 (merge-base) diff
BENCH_NAME=record_batch
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench record_batch
BENCH_FILTER=
Results will be posted here when complete

adriangbot · 2026-03-13T00:05:28Z

🤖 Arrow criterion benchmark completed (GKE) | trigger

Details

group                            add-has-true-has-false                 main
-----                            ----------------------                 ----
project/1000x8192 -> 1x8192      1.00    155.6±1.37ns        ? ?/sec    1.00    155.0±1.39ns        ? ?/sec
project/1000x8192 -> 500x8192    1.00     13.7±0.04µs        ? ?/sec    1.00     13.6±0.04µs        ? ?/sec
project/1000x8192 -> 999x8192    1.00     27.0±0.09µs        ? ?/sec    1.00     27.0±0.09µs        ? ?/sec
project/100x8192 -> 1x8192       1.01    156.0±1.40ns        ? ?/sec    1.00    154.9±1.41ns        ? ?/sec
project/100x8192 -> 50x8192      1.00  1754.6±11.12ns        ? ?/sec    1.00  1757.2±13.68ns        ? ?/sec
project/100x8192 -> 99x8192      1.00      3.1±0.01µs        ? ?/sec    1.01      3.1±0.02µs        ? ?/sec
project/10x8192 -> 1x8192        1.03    166.4±5.97ns        ? ?/sec    1.00    160.8±1.20ns        ? ?/sec
project/10x8192 -> 5x8192        1.02    344.8±4.25ns        ? ?/sec    1.00    338.9±3.90ns        ? ?/sec
project/10x8192 -> 9x8192        1.01    501.4±5.25ns        ? ?/sec    1.00    497.0±2.77ns        ? ?/sec

Resource Usage

base (merge-base)

Metric	Value
Wall time	88.5s
Peak memory	2.0 GiB
Avg memory	1.9 GiB
CPU user	87.5s
CPU sys	0.8s
Disk read	0 B
Disk write	1.2 GiB

branch

Metric	Value
Wall time	87.4s
Peak memory	2.0 GiB
Avg memory	1.9 GiB
CPU user	87.2s
CPU sys	0.2s
Disk read	0 B
Disk write	29.5 MiB

adriangb · 2026-03-13T06:07:07Z

run benchmark boolean_array

adriangbot · 2026-03-13T06:07:12Z

Benchmark job started for this request (job bench-c4052991164-186). Results will be posted here when complete.

adriangbot · 2026-03-13T06:10:25Z

🤖 Arrow criterion benchmark running (GKE) | trigger
Linux bench-c4052991164-186-7k56q 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux
Comparing add-has-true-has-false (1551756) to 5ba4515 (merge-base) diff
BENCH_NAME=boolean_array
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench boolean_array
BENCH_FILTER=
Results will be posted here when complete

adriangbot · 2026-03-13T06:15:49Z

🤖 Arrow criterion benchmark completed (GKE) | trigger

New benchmark — branch-only results (no baseline comparison)

Details

group                                add-has-true-has-false
-----                                ----------------------
has_false(all_false, 1024)           1.00      8.4±0.04ns        ? ?/sec
has_false(all_false, 64)             1.00      8.1±0.09ns        ? ?/sec
has_false(all_false, 65536)          1.00     13.6±0.03ns        ? ?/sec
has_false(all_true, 1024)            1.00      8.4±0.04ns        ? ?/sec
has_false(all_true, 64)              1.00      8.3±0.10ns        ? ?/sec
has_false(all_true, 65536)           1.00    122.9±0.66ns        ? ?/sec
has_false(mixed_early, 1024)         1.00      8.4±0.04ns        ? ?/sec
has_false(mixed_early, 64)           1.00      8.1±0.07ns        ? ?/sec
has_false(mixed_early, 65536)        1.00     13.6±0.01ns        ? ?/sec
has_false(nulls_all_true, 1024)      1.00     24.9±0.46ns        ? ?/sec
has_false(nulls_all_true, 64)        1.00      6.8±0.07ns        ? ?/sec
has_false(nulls_all_true, 65536)     1.00    946.1±3.04ns        ? ?/sec
has_true(all_false, 1024)            1.00      8.4±0.02ns        ? ?/sec
has_true(all_false, 64)              1.00      8.2±0.07ns        ? ?/sec
has_true(all_false, 65536)           1.00    121.2±0.48ns        ? ?/sec
has_true(all_true, 1024)             1.00      8.3±0.02ns        ? ?/sec
has_true(all_true, 64)               1.00      7.8±0.37ns        ? ?/sec
has_true(all_true, 65536)            1.00     13.5±0.03ns        ? ?/sec
has_true(nulls_all_true, 1024)       1.00      6.5±0.09ns        ? ?/sec
has_true(nulls_all_true, 64)         1.00      6.5±0.08ns        ? ?/sec
has_true(nulls_all_true, 65536)      1.00      6.5±0.09ns        ? ?/sec
true_count(all_false, 1024)          1.00     10.0±0.02ns        ? ?/sec
true_count(all_false, 64)            1.00      8.7±0.31ns        ? ?/sec
true_count(all_false, 65536)         1.00    227.7±0.05ns        ? ?/sec
true_count(all_true, 1024)           1.00     10.1±0.44ns        ? ?/sec
true_count(all_true, 64)             1.00      8.7±0.39ns        ? ?/sec
true_count(all_true, 65536)          1.00    227.7±0.06ns        ? ?/sec
true_count(mixed_early, 1024)        1.00     10.1±0.02ns        ? ?/sec
true_count(mixed_early, 64)          1.00      8.7±0.32ns        ? ?/sec
true_count(mixed_early, 65536)       1.00    227.7±0.06ns        ? ?/sec
true_count(nulls_all_true, 1024)     1.00     18.9±0.04ns        ? ?/sec
true_count(nulls_all_true, 64)       1.00      7.3±0.04ns        ? ?/sec
true_count(nulls_all_true, 65536)    1.00    717.5±0.74ns        ? ?/sec

Resource Usage

branch

Metric	Value
Wall time	315.8s
Peak memory	1.7 GiB
Avg memory	1.6 GiB
CPU user	314.8s
CPU sys	0.9s
Disk read	4.0 KiB
Disk write	1.0 GiB

alamb · 2026-03-13T09:55:48Z

run benchmark boolean_array

adriangbot · 2026-03-13T09:55:53Z

Benchmark job started for this request (job bench-c4053957332-207). Results will be posted here when complete.

adriangbot · 2026-03-13T10:05:07Z

🤖 Arrow criterion benchmark running (GKE) | trigger
Linux bench-c4053957332-207-22g2d 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux
Comparing add-has-true-has-false (1551756) to 5ba4515 (merge-base) diff
BENCH_NAME=boolean_array
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench boolean_array
BENCH_FILTER=
Results will be posted here when complete

adriangbot · 2026-03-13T10:10:27Z

🤖 Arrow criterion benchmark completed (GKE) | trigger

New benchmark — branch-only results (no baseline comparison)

Details

group                                add-has-true-has-false
-----                                ----------------------
has_false(all_false, 1024)           1.00      8.4±0.04ns        ? ?/sec
has_false(all_false, 64)             1.00      8.1±0.13ns        ? ?/sec
has_false(all_false, 65536)          1.00     13.6±0.03ns        ? ?/sec
has_false(all_true, 1024)            1.00      8.6±0.02ns        ? ?/sec
has_false(all_true, 64)              1.00      8.5±0.10ns        ? ?/sec
has_false(all_true, 65536)           1.00    123.1±0.63ns        ? ?/sec
has_false(mixed_early, 1024)         1.00      8.4±0.03ns        ? ?/sec
has_false(mixed_early, 64)           1.00      8.1±0.10ns        ? ?/sec
has_false(mixed_early, 65536)        1.00     13.6±0.02ns        ? ?/sec
has_false(nulls_all_true, 1024)      1.00     24.7±0.51ns        ? ?/sec
has_false(nulls_all_true, 64)        1.00      6.8±0.07ns        ? ?/sec
has_false(nulls_all_true, 65536)     1.00    943.3±4.40ns        ? ?/sec
has_true(all_false, 1024)            1.00      8.4±0.04ns        ? ?/sec
has_true(all_false, 64)              1.00      8.2±0.08ns        ? ?/sec
has_true(all_false, 65536)           1.00    121.3±0.42ns        ? ?/sec
has_true(all_true, 1024)             1.00      8.6±0.03ns        ? ?/sec
has_true(all_true, 64)               1.00      8.0±0.09ns        ? ?/sec
has_true(all_true, 65536)            1.00     13.7±0.03ns        ? ?/sec
has_true(nulls_all_true, 1024)       1.00      6.3±0.07ns        ? ?/sec
has_true(nulls_all_true, 64)         1.00      6.3±0.09ns        ? ?/sec
has_true(nulls_all_true, 65536)      1.00      6.3±0.09ns        ? ?/sec
true_count(all_false, 1024)          1.00      9.9±0.01ns        ? ?/sec
true_count(all_false, 64)            1.00      8.7±0.27ns        ? ?/sec
true_count(all_false, 65536)         1.00    227.8±0.19ns        ? ?/sec
true_count(all_true, 1024)           1.00     10.2±0.06ns        ? ?/sec
true_count(all_true, 64)             1.00      9.1±0.26ns        ? ?/sec
true_count(all_true, 65536)          1.00    228.0±0.13ns        ? ?/sec
true_count(mixed_early, 1024)        1.00     10.0±0.14ns        ? ?/sec
true_count(mixed_early, 64)          1.00      8.7±0.28ns        ? ?/sec
true_count(mixed_early, 65536)       1.00    227.7±0.07ns        ? ?/sec
true_count(nulls_all_true, 1024)     1.00     18.4±0.05ns        ? ?/sec
true_count(nulls_all_true, 64)       1.00      7.2±0.13ns        ? ?/sec
true_count(nulls_all_true, 65536)    1.00    715.9±0.80ns        ? ?/sec

Resource Usage

branch

Metric	Value
Wall time	311.8s
Peak memory	1.7 GiB
Avg memory	1.6 GiB
CPU user	310.7s
CPU sys	0.9s
Disk read	0 B
Disk write	1012.2 MiB

adriangb and others added 3 commits March 5, 2026 11:07

Add benchmark for has_true/has_false vs true_count

1f261ad

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

github-actions bot added the arrow Changes to the arrow crate label Mar 5, 2026

adriangb added 2 commits March 5, 2026 11:32

fix

768cba2

update docstring

1551756

Conversation

adriangb commented Mar 5, 2026

Motivation

Callsites in DataFusion

Benchmark Results

Implementation

Alternatives considered

Test Plan

Uh oh!

adriangb commented Mar 5, 2026

Uh oh!

alamb-ghbot commented Mar 5, 2026

Uh oh!

adriangb commented Mar 5, 2026

Uh oh!

adriangb commented Mar 12, 2026

Uh oh!

alamb-ghbot commented Mar 12, 2026

Uh oh!

adriangb commented Mar 12, 2026

Uh oh!

alamb-ghbot commented Mar 12, 2026

Uh oh!

adriangbot commented Mar 12, 2026

Uh oh!

adriangbot commented Mar 12, 2026

Uh oh!

adriangbot commented Mar 12, 2026

Uh oh!

adriangbot commented Mar 12, 2026

Uh oh!

adriangbot commented Mar 12, 2026

Uh oh!

adriangbot commented Mar 12, 2026

Uh oh!

adriangbot commented Mar 12, 2026

Uh oh!

adriangbot commented Mar 12, 2026

Uh oh!

adriangbot commented Mar 12, 2026

Uh oh!

adriangbot commented Mar 12, 2026

Uh oh!

adriangbot commented Mar 12, 2026

Uh oh!

adriangbot commented Mar 12, 2026

Uh oh!

adriangbot commented Mar 12, 2026

Uh oh!

adriangbot commented Mar 12, 2026

Uh oh!

adriangbot commented Mar 12, 2026

Uh oh!

adriangbot commented Mar 12, 2026

Uh oh!

adriangbot commented Mar 12, 2026

Uh oh!

adriangbot commented Mar 12, 2026

Uh oh!

adriangb commented Mar 12, 2026

Uh oh!

adriangbot commented Mar 12, 2026

Uh oh!

adriangbot commented Mar 13, 2026

Uh oh!

adriangbot commented Mar 13, 2026

Uh oh!

adriangb commented Mar 13, 2026

Uh oh!

adriangbot commented Mar 13, 2026

Uh oh!

adriangbot commented Mar 13, 2026

Uh oh!

adriangbot commented Mar 13, 2026

Uh oh!

alamb commented Mar 13, 2026

Uh oh!

adriangbot commented Mar 13, 2026

Uh oh!

adriangbot commented Mar 13, 2026