perf: wip sub seg p3
#7392Comparing
claude/optimize-build-performance-NzwKr
(df23421
) withdevelop
(3bdb376
)0%
- Untouched
- 1122
- Skipped
- 1455
Benchmarks
2577 total
for_compress_i32
vortex/benches/single_encoding_throughput.rs+10%
778 µs708.4 µs
encode_varbinview[(10000, 4)]
vortex-array/benches/dict_compress.rs+8%
1.2 ms1.1 ms
encode_varbinview[(10000, 8)]
vortex-array/benches/dict_compress.rs+4%
1.1 ms1.1 ms
map_new_output[128]
vortex-buffer/benches/vortex_buffer.rs+4%
851.9 ns822.8 ns
ScalarBuffer<i32>>, 128]
vortex-buffer/benches/vortex_buffer.rs::map_each[Arrow<arrow_buffer::buffer::scalar+3%
1.9 µs1.8 µs
dict_compress_u32
vortex/benches/single_encoding_throughput.rs+3%
4.2 ms4 ms
runend_compress_u32
vortex/benches/single_encoding_throughput.rs+2%
5.6 ms5.5 ms
canonicalize_compare[(1000, 4, 4)]
encodings/fsst/benches/fsst_compress.rs+2%
130.9 µs128.1 µs
push_n_vortex_buffer[u16, 128]
vortex-buffer/benches/vortex_buffer.rs+2%
1.6 µs1.6 µs
chunked_into_canonical[(1000, 100, 16, 4)]
encodings/fsst/benches/fsst_compress.rs+2%
9.7 ms9.6 ms
canonicalize_compare[(1000, 4, 8)]
encodings/fsst/benches/fsst_compress.rs+2%
132.2 µs130 µs
ScalarBuffer<i32>>, 1024]
vortex-buffer/benches/vortex_buffer.rs::map_each[Arrow<arrow_buffer::buffer::scalar+2%
3.7 µs3.6 µs
push_n_vortex_buffer[u64, 128]
vortex-buffer/benches/vortex_buffer.rs+2%
1.9 µs1.9 µs
map_each[Buffer<i32>, 128]
vortex-buffer/benches/vortex_buffer.rs+1%
2.1 µs2.1 µs
case_when_fragmented[1000]
vortex-array/benches/expr/case_when_bench.rs+1%
519.6 µs512.8 µs
decode_bool[10000_1000_alternating]
encodings/runend/benches/run_end_decode.rs+1%
9.3 µs9.2 µs
decode_bool[10000_1000_all_false]
encodings/runend/benches/run_end_decode.rs+1%
9.5 µs9.4 µs
decode_bool[10000_1000_all_true]
encodings/runend/benches/run_end_decode.rs+1%
9.5 µs9.4 µs
decode_bool[10000_1000_mostly_false]
encodings/runend/benches/run_end_decode.rs+1%
9.5 µs9.4 µs
decompress_fsst[(1000, 4, 4)]
encodings/fsst/benches/fsst_compress.rs+1%
65.2 µs64.4 µs
decode_bool[10000_1000_mostly_true]
encodings/runend/benches/run_end_decode.rs+1%
9.6 µs9.4 µs
canonicalize_compare[(1000, 64, 4)]
encodings/fsst/benches/fsst_compress.rs+1%
234.3 µs231.5 µs
chunked_into_canonical[(10, 10000, 4, 4)]
encodings/fsst/benches/fsst_compress.rs+1%
4.2 ms4.2 ms
decompress_fsst[(1000, 4, 8)]
encodings/fsst/benches/fsst_compress.rs+1%
67 µs66.3 µs
decode_bool[10000_100_all_true]
encodings/runend/benches/run_end_decode.rs+1%
11.7 µs11.5 µs
Commits
Click on a commit to change the comparison rangeBase
develop
3bdb376
-0.24%
perf: optimize LazyBufferHandle to restore 1-read-per-segment behavior
df23421
9 hours ago
by claude