fsst decompress optimize try 2
#7001Comparing
claude/fsst-decompress-optimize-TylL2
(7fd4e7c
) withdevelop
(50868db
)-12%
- Improvement
- 48
- Regression
- 2
- Untouched
- 959
- New
- 40
- Skipped
- 1515
Benchmarks
2564 total
chunked_canonicalize_into[(1000, 50, 8, 64)]
encodings/fsst/benches/fsst_compress.rs-12%
10.5 ms12 ms
chunked_into_canonical[(1000, 50, 8, 64)]
encodings/fsst/benches/fsst_compress.rs-12%
10.8 ms12.3 ms
decompress_fsst[(10000, 4, 4)]
encodings/fsst/benches/fsst_compress.rs+33%
767.5 µs575 µs
chunked_canonicalize_into[(10, 10000, 4, 4)]
encodings/fsst/benches/fsst_compress.rs+33%
7.8 ms5.9 ms
chunked_into_canonical[(10, 10000, 4, 4)]
encodings/fsst/benches/fsst_compress.rs+33%
7.8 ms5.9 ms
decompress_fsst[(10000, 4, 8)]
encodings/fsst/benches/fsst_compress.rs+33%
787.6 µs593.6 µs
decompress_fsst[(10000, 16, 4)]
encodings/fsst/benches/fsst_compress.rs+22%
1,142.4 µs934.5 µs
chunked_canonicalize_into[(10, 10000, 16, 4)]
encodings/fsst/benches/fsst_compress.rs+22%
11.6 ms9.5 ms
chunked_into_canonical[(10, 10000, 16, 4)]
encodings/fsst/benches/fsst_compress.rs+22%
11.6 ms9.5 ms
decompress_fsst[(10000, 16, 8)]
encodings/fsst/benches/fsst_compress.rs+22%
1.3 ms1 ms
canonicalize_compare[(10000, 4, 4)]
encodings/fsst/benches/fsst_compress.rs+20%
1,153.8 µs960.9 µs
canonicalize_compare[(10000, 4, 8)]
encodings/fsst/benches/fsst_compress.rs+20%
1,173.5 µs979.3 µs
fsst_decompress_string
vortex/benches/single_encoding_throughput.rs+19%
10.8 ms9.1 ms
decompress_fsst[(1000, 4, 4)]
encodings/fsst/benches/fsst_compress.rs+18%
102.5 µs86.5 µs
decompress_fsst[(1000, 4, 8)]
encodings/fsst/benches/fsst_compress.rs+18%
105.8 µs89.4 µs
decompress_fsst[(10000, 64, 8)]
encodings/fsst/benches/fsst_compress.rs+18%
3.1 ms2.6 ms
decompress_fsst[(10000, 64, 4)]
encodings/fsst/benches/fsst_compress.rs+18%
2.6 ms2.2 ms
canonicalize_compare[(10000, 16, 4)]
encodings/fsst/benches/fsst_compress.rs+17%
1.4 ms1.2 ms
canonicalize_compare[(10000, 16, 8)]
encodings/fsst/benches/fsst_compress.rs+17%
1.5 ms1.3 ms
chunked_into_canonical[(10, 10000, 64, 4)]
encodings/fsst/benches/fsst_compress.rs+17%
28.1 ms24 ms
chunked_canonicalize_into[(10, 10000, 64, 4)]
encodings/fsst/benches/fsst_compress.rs+16%
28.1 ms24.2 ms
canonicalize_compare[(10000, 64, 8)]
encodings/fsst/benches/fsst_compress.rs+16%
3.3 ms2.9 ms
canonicalize_compare[(10000, 64, 4)]
encodings/fsst/benches/fsst_compress.rs+16%
2.9 ms2.5 ms
decompress_fsst[(1000, 64, 4)]
encodings/fsst/benches/fsst_compress.rs+14%
287.7 µs251.7 µs
decompress_fsst[(1000, 64, 8)]
encodings/fsst/benches/fsst_compress.rs+14%
331.1 µs289.9 µs
Commits
Click on a commit to change the comparison rangeBase
develop
50868db
-21.05%
Simplify emit_before_escape macro and remove optimization doc
ecdc86f
2 days ago
by claude+8.82%
Optimize FSST inline view construction and escape handling
7fd4e7c
2 days ago
by claude