Comparing
dani/optimize-simd
(852c862
) withmaster
(023646a
)-30%
- Improvement
- 12
- Regression
- 4
- Untouched
- 20
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
36 total
bench1_32b
benches/bench/main.rs::check::const_hex-30%
321.9 ns460 ns
bench1_32b
benches/bench/main.rs::encode_to_slice::const_hex-13%
429.4 ns493.1 ns
bench2_256b
benches/bench/main.rs::encode_to_slice::const_hex-11%
812.5 ns915 ns
bench1_32b
benches/bench/main.rs::decode::const_hex-10%
1.3 µs1.5 µs
bench6_1m
benches/bench/main.rs::check::const_hex+36%
1.6 ms1.2 ms
bench5_128k
benches/bench/main.rs::check::const_hex+36%
206.2 µs151.7 µs
bench4_16k
benches/bench/main.rs::check::const_hex+35%
26 µs19.3 µs
bench3_2k
benches/bench/main.rs::check::const_hex+27%
3.5 µs2.8 µs
bench6_1m
benches/bench/main.rs::decode_to_slice::const_hex+21%
2.5 ms2.1 ms
bench6_1m
benches/bench/main.rs::decode::const_hex+21%
2.5 ms2.1 ms
bench5_128k
benches/bench/main.rs::decode_to_slice::const_hex+21%
312 µs257.5 µs
bench5_128k
benches/bench/main.rs::decode::const_hex+21%
314.6 µs260.1 µs
bench4_16k
benches/bench/main.rs::decode_to_slice::const_hex+20%
39.7 µs32.9 µs
bench4_16k
benches/bench/main.rs::decode::const_hex+19%
41.6 µs34.9 µs
bench3_2k
benches/bench/main.rs::decode_to_slice::const_hex+16%
5.6 µs4.8 µs
bench3_2k
benches/bench/main.rs::decode::const_hex+10%
7.8 µs7.1 µs
bench6_1m
benches/bench/main.rs::encode_to_slice::const_hex+6%
1.7 ms1.6 ms
bench6_1m
benches/bench/main.rs::encode::const_hex+6%
1.7 ms1.6 ms
bench5_128k
benches/bench/main.rs::encode_to_slice::const_hex+5%
215.9 µs204.8 µs
bench5_128k
benches/bench/main.rs::encode::const_hex+5%
218.9 µs207.6 µs
bench4_16k
benches/bench/main.rs::encode::const_hex+5%
28.6 µs27.3 µs
bench4_16k
benches/bench/main.rs::encode_to_slice::const_hex+5%
27.3 µs26 µs
bench2_256b
benches/bench/main.rs::decode_to_slice::const_hex+3%
1.5 µs1.4 µs
bench6_1m
benches/bench/main.rs::format::const_hex+2%
4 ms3.9 ms
bench3_2k
benches/bench/main.rs::encode::const_hex+2%
5.7 µs5.6 µs
Commits
Click on a commit to change the comparison rangeBase
master
023646a
-14.3%
perf: optimize SIMD check and encode implementations
7bb2f10
2 months ago
by DaniPopes-5.71%
fix: use signed overflow trick for SSE2 check, clean up tests
612c134
2 months ago
by DaniPopes-10.21%
clean
3682bd0
2 months ago
by DaniPopes0%
perf: avoid SSSE3 loop in AVX2 encode remainder
556ea36
2 months ago
by DaniPopes0%
unoutline
655c087
2 months ago
by DaniPopes+4.95%
perf: skip remainder check for exact chunk multiples
e27abcd
2 months ago
by DaniPopes-4.74%
more empty fast paths
cbc85f7
2 months ago
by DaniPopes0%
refactor: reorder one_chunk fns below with_ fns, add debug assertions
c8e42cf
2 months ago
by DaniPopes0%
refactor: remove unnecessary closure type hints
852c862
2 months ago
by DaniPopes