claude/confident-hamilton-mZIEo - Branch - vortex-data/vortex

perf(fastlanes): fuse bit-packed compare into a transposed mask + untranspose

#8239Merged

Comparing

claude/confident-hamilton-mZIEo

(

bd3fbaa

) with

develop

(

583b003

)

+60%

Improvement: 239

Untouched: 1268

Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data. For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments, which may affect the accuracy of the results. Learn more about regression causes.

Benchmarks

1507 total

fsst_decompress_string

vortex/benches/single_encoding_throughput.rs

3.1 ms3.1 ms

take_search_chunked[(0.1, 0.1)]

vortex-array/benches/take_patches.rs

441.6 µs441.6 µs

canonicalize_sparse_list[(1024, 17, 8)]

encodings/sparse/benches/sparse_canonical.rs

463.7 µs463.7 µs

take_indices[(100000, 4)]

encodings/runend/benches/run_end_compress.rs

1.4 ms1.4 ms

compress[(100000, 4)]

encodings/runend/benches/run_end_compress.rs

2 ms2 ms

null_count_run_end[(100000, 4, 0.5)]

encodings/runend/benches/run_end_null_count.rs

668.5 µs668.5 µs

take_search_chunked[(0.1, 1.0)]

vortex-array/benches/take_patches.rs

4.1 ms4.1 ms

encode_primitives[i64, (10000, 2)]

vortex-array/benches/dict_compress.rs

404.8 µs404.8 µs

decompress[dict_fsst_varbin_bp_string]

vortex/benches/common_encoding_tree_throughput.rs

8.5 ms8.5 ms

cuda/dict/u64_values_u8_codes[100M]

vortex-cuda/benches/dict_cuda.rs::benches::benchmark_dict::cuda

2 ms2 ms

chunked_canonicalize_into[(1000, 50, 8, 64)]

encodings/fsst/benches/fsst_compress.rs

6.3 ms6.3 ms

take_search_chunked[(0.1, 0.5)]

vortex-array/benches/take_patches.rs

2.1 ms2.1 ms

density_sweep_dense_runs[0.01]

vortex-array/benches/filter_bool.rs

44.9 µs44.9 µs

density_sweep_dense_runs[0.005]

vortex-array/benches/filter_bool.rs

44.9 µs44.9 µs

bench_compare_sliced_dict_varbinview[(10000, 10000)]

vortex-array/benches/dict_compare.rs

254.7 µs254.7 µs

density_sweep_dense_runs[0.9999]

vortex-array/benches/filter_bool.rs

46.2 µs46.2 µs

compact[(4096, 10)]

vortex-array/benches/varbinview_compact.rs

512.5 µs512.6 µs

case_when_nary_early_dominant[10000]

vortex-array/benches/expr/case_when_bench.rs

186.8 µs186.8 µs

train_compressor[(10000, 4, 8)]

encodings/fsst/benches/fsst_compress.rs

3.7 ms3.7 ms

chunked_constant_i32_append_to_builder[(100, 100)]

vortex-array/benches/chunk_array_builder.rs

47.5 µs47.5 µs

compress_fsst[(1000, 4, 8)]

encodings/fsst/benches/fsst_compress.rs

190.9 µs190.9 µs

compress_fsst[(1000, 16, 4)]

encodings/fsst/benches/fsst_compress.rs

242.7 µs242.7 µs

very_sparse_mask_cached[(0.5, 0.005, "self_dense_mask_0p5pct")]

vortex-mask/benches/intersect_by_rank.rs

48.7 µs48.7 µs

very_sparse_mask_uncached[(0.1, 0.01, "self_10pct_mask_1pct")]

vortex-mask/benches/intersect_by_rank.rs

48.9 µs48.9 µs

i32_large

vortex-array/benches/listview_rebuild.rs

284.1 µs284.1 µs

Commits

Click on a commit to change the comparison range

Base

develop

583b003

+46%

perf(fastlanes): fuse bit-packed compare into a transposed mask + untranspose

48da899

10 days ago

by claude

+17.53%

Merge branch 'develop' into claude/confident-hamilton-mZIEo

ab9c8d6

9 days ago

by joseph-isaacs

-4.72%

ci(wasm): pin in-development FastLanes in the excluded wasm-test workspace

08ed4a4

9 days ago

by claude

+5.42%

wip

816032b

9 days ago

by joseph-isaacs

-0.47%

wip

e4dd660

9 days ago

by joseph-isaacs

-65.08%

wip

211903c

9 days ago

by joseph-isaacs

+60.04%

wip

0649f03

9 days ago

by joseph-isaacs

+1.04%

wip

83c90b9

9 days ago

by joseph-isaacs

-1.41%

wip

4d50ff1

9 days ago

by joseph-isaacs

-58.4%

wip

c117649

9 days ago

by joseph-isaacs

+59.68%

wip

10e904c

9 days ago

by joseph-isaacs

+0.41%

Fix sliced_with_patches compare test to obtain a BitPacked via SliceKernel

c495055

9 days ago

by joseph-isaacs

-0.41%

Import SliceKernel instead of absolute path to satisfy clippy::absolute-paths

bd3fbaa

9 days ago

by joseph-isaacs

Home Terms Privacy Docs