Faster true count using AVX2 and AVX512 instructions
#6931Comparing
rk/truecount
(d3c062d
) withdevelop
(fc4d111
)+88%
- Improvement
- 8
- Untouched
- 1001
- Skipped
- 1515
Benchmarks
2524 total
true_count_vortex_buffer[128]
vortex-buffer/benches/vortex_bitbuffer.rs+88%
1,013.9 ns538.6 ns
true_count_vortex_buffer[65536]
vortex-buffer/benches/vortex_bitbuffer.rs+80%
11.8 µs6.6 µs
true_count_vortex_buffer[16384]
vortex-buffer/benches/vortex_bitbuffer.rs+52%
3.6 µs2.4 µs
null_count_run_end[(100000, 1024, 0.1)]
encodings/runend/benches/run_end_null_count.rs+13%
65.6 µs57.8 µs
null_count_run_end[(100000, 1024, 0.5)]
encodings/runend/benches/run_end_null_count.rs+13%
65.9 µs58.2 µs
null_count_run_end[(100000, 256, 0.01)]
encodings/runend/benches/run_end_null_count.rs+12%
72.8 µs64.9 µs
null_count_run_end[(100000, 256, 0.1)]
encodings/runend/benches/run_end_null_count.rs+12%
73.1 µs65.3 µs
null_count_run_end[(100000, 256, 0.5)]
encodings/runend/benches/run_end_null_count.rs+12%
75.2 µs67.4 µs
bench_many_nulls[0.01]
vortex-array/benches/dict_unreferenced_mask.rs+9%
59.2 µs54.1 µs
bitwise_not_vortex_buffer_mut[128]
vortex-buffer/benches/vortex_bitbuffer.rs+8%
376.1 ns346.9 ns
bitwise_not_vortex_buffer_mut[1024]
vortex-buffer/benches/vortex_bitbuffer.rs+6%
535.6 ns506.4 ns
decode_bool_nullable[10000_1000_alternating_mostly_valid]
encodings/runend/benches/run_end_decode.rs+5%
20.9 µs19.9 µs
true_count_vortex_buffer[2048]
vortex-buffer/benches/vortex_bitbuffer.rs+5%
1.3 µs1.2 µs
bitwise_not_vortex_buffer_mut[2048]
vortex-buffer/benches/vortex_bitbuffer.rs+4%
709.4 ns680.3 ns
append_buffer_vortex_buffer[65536]
vortex-buffer/benches/vortex_bitbuffer.rs+4%
176 µs168.8 µs
null_count_run_end[(100000, 16, 0.01)]
encodings/runend/benches/run_end_null_count.rs+4%
223.9 µs215.2 µs
append_buffer_vortex_buffer[16384]
vortex-buffer/benches/vortex_bitbuffer.rs+4%
57.4 µs55.2 µs
null_count_run_end[(100000, 16, 0.1)]
encodings/runend/benches/run_end_null_count.rs+4%
234.4 µs225.7 µs
append_buffer_vortex_buffer[1024]
vortex-buffer/benches/vortex_bitbuffer.rs+4%
26.4 µs25.5 µs
append_buffer_vortex_buffer[2048]
vortex-buffer/benches/vortex_bitbuffer.rs+3%
28.5 µs27.6 µs
null_count_run_end[(100000, 16, 0.5)]
encodings/runend/benches/run_end_null_count.rs+3%
281.9 µs273.2 µs
chunked_opt_bool_canonical_into[(1000, 10)]
vortex-array/benches/chunk_array_builder.rs+3%
95.8 µs92.9 µs
decode_bool_nullable[10000_100_alternating_mostly_valid]
encodings/runend/benches/run_end_decode.rs+3%
26.1 µs25.3 µs
bench_many_nulls[0.1]
vortex-array/benches/dict_unreferenced_mask.rs+3%
176.3 µs171.2 µs
append_buffer_vortex_buffer[128]
vortex-buffer/benches/vortex_bitbuffer.rs+3%
21.2 µs20.6 µs
Commits
Click on a commit to change the comparison rangeBase
develop
fc4d111
+88.24%
trythis
d3c062d
12 hours ago
by robert3005