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

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

#8239

Comparing

claude/confident-hamilton-mZIEo

(

48da899

) with

claude/confident-hamilton-mZIEo-benches

(

10939a6

)

+46%

Improvement: 240

Regression: 26

Untouched: 1241

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.

Benchmarks

1507 total

pushdown_compare[(1000, 16, 4)]

encodings/fsst/benches/fsst_compress.rs

-59%

141.6 µs345.8 µs

pushdown_compare[(1000, 4, 4)]

encodings/fsst/benches/fsst_compress.rs

-59%

142.5 µs345.3 µs

pushdown_compare[(1000, 64, 4)]

encodings/fsst/benches/fsst_compress.rs

-59%

142.7 µs345 µs

pushdown_compare[(1000, 4, 8)]

encodings/fsst/benches/fsst_compress.rs

-58%

145.9 µs349.6 µs

pushdown_compare[(1000, 64, 8)]

encodings/fsst/benches/fsst_compress.rs

-58%

148 µs351.3 µs

pushdown_compare[(1000, 16, 8)]

encodings/fsst/benches/fsst_compress.rs

-57%

154.4 µs357.3 µs

pushdown_compare[(10000, 64, 4)]

encodings/fsst/benches/fsst_compress.rs

-49%

214.2 µs417 µs

pushdown_compare[(10000, 64, 8)]

encodings/fsst/benches/fsst_compress.rs

-48%

221.6 µs424.2 µs

pushdown_compare[(10000, 4, 4)]

encodings/fsst/benches/fsst_compress.rs

-47%

221.2 µs418.1 µs

pushdown_compare[(10000, 16, 4)]

encodings/fsst/benches/fsst_compress.rs

-47%

221.4 µs418.3 µs

pushdown_compare[(10000, 4, 8)]

encodings/fsst/benches/fsst_compress.rs

-46%

227.1 µs423.6 µs

pushdown_compare[(10000, 16, 8)]

encodings/fsst/benches/fsst_compress.rs

-43%

263.8 µs459.6 µs

eq_pushdown_low_match

encodings/fsst/benches/fsst_url_compare.rs

-17%

955.2 µs1,152.4 µs

eq_pushdown_high_match

encodings/fsst/benches/fsst_url_compare.rs

-16%

1.1 ms1.2 ms

cuda/bitpacked_u8/unpack/3bw[100M]

vortex-cuda/benches/bitpacked_cuda.rs::benches::benchmark_bitunpack::cuda

-15%

298.8 µs350.9 µs

decompress_fsst[(10000, 16, 4)]

encodings/fsst/benches/fsst_compress.rs

-12%

509.3 µs579.9 µs

fsst_decompress_string

vortex/benches/single_encoding_throughput.rs

-12%

3.1 ms3.5 ms

chunked_into_canonical[(10, 10000, 16, 4)]

encodings/fsst/benches/fsst_compress.rs

-12%

5.2 ms5.9 ms

chunked_canonicalize_into[(10, 10000, 16, 4)]

encodings/fsst/benches/fsst_compress.rs

-12%

5.2 ms5.9 ms

decompress_fsst[(10000, 16, 8)]

encodings/fsst/benches/fsst_compress.rs

-11%

561.9 µs631.9 µs

decompress_fsst[(10000, 4, 4)]

encodings/fsst/benches/fsst_compress.rs

-11%

347.5 µs389.1 µs

chunked_canonicalize_into[(10, 10000, 4, 4)]

encodings/fsst/benches/fsst_compress.rs

-11%

3.6 ms4 ms

bitwise_not_vortex_buffer_mut[128]

vortex-buffer/benches/vortex_bitbuffer.rs

-11%

246.1 ns275.3 ns

decompress_fsst[(10000, 4, 8)]

encodings/fsst/benches/fsst_compress.rs

-10%

356.9 µs398.4 µs

chunked_into_canonical[(10, 10000, 4, 4)]

encodings/fsst/benches/fsst_compress.rs

-10%

3.6 ms4 ms

Commits

Click on a commit to change the comparison range

Base

claude/confident-hamilton-mZIEo-benches

10939a6

+46%

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

48da899

7 hours ago

by claude

Home Terms Privacy Docs