chore: clean cancelled read requests earlier in driver state transitions
#6382Comparing
caicancai:fix/read-early-cancel-cleanup
(57159f2
) withdevelop
(d1097c5
)+34%
- Improvement
- 1
- Untouched
- 1137
- Skipped
- 1265
Benchmarks
2403 total
chunked_bool_into_canonical[(1000, 10)]
vortex-array/benches/chunk_array_builder.rs+34%
88.4 µs65.9 µs
compress[(1000000, 4)]
encodings/runend/benches/run_end_compress.rs+6%
26.1 ms24.7 ms
chunked_opt_bool_canonical_into[(100, 100)]
vortex-array/benches/chunk_array_builder.rs+4%
372.1 µs359 µs
encode_varbin[(1000, 8)]
vortex-array/benches/dict_compress.rs+3%
234.2 µs227.6 µs
encode_varbin[(10000, 32)]
vortex-array/benches/dict_compress.rs+2%
1.5 ms1.5 ms
take_map[(0.1, 0.1)]
vortex-array/benches/take_patches.rs+2%
929 µs910.5 µs
map_new_output[128]
vortex-buffer/benches/vortex_buffer.rs+2%
1.5 µs1.5 µs
take_map[(0.005, 0.5)]
vortex-array/benches/take_patches.rs+2%
1.3 ms1.2 ms
take_10_stratified
encodings/fastlanes/benches/bitpacking_take.rs+2%
7.4 µs7.3 µs
take_10k_contiguous
encodings/fastlanes/benches/bitpacking_take.rs+2%
7.4 µs7.3 µs
take_200k_first_chunk_only
encodings/fastlanes/benches/bitpacking_take.rs+2%
7.5 µs7.3 µs
take_indices[(1000000, 16)]
encodings/runend/benches/run_end_compress.rs+1%
9 µs8.9 µs
patched_take_200k_first_chunk_only
encodings/fastlanes/benches/bitpacking_take.rs+1%
7.6 µs7.5 µs
take_struct_sequential_indices
vortex-array/benches/take_struct.rs+1%
8.1 µs8 µs
patched_take_10_stratified
encodings/fastlanes/benches/bitpacking_take.rs+1%
8.1 µs8 µs
take_indices[(100000, 256)]
encodings/runend/benches/run_end_compress.rs+1%
8.9 µs8.8 µs
take_indices[(1000000, 1024)]
encodings/runend/benches/run_end_compress.rs+1%
8.8 µs8.7 µs
ScalarBuffer<i32>>, 128]
vortex-buffer/benches/vortex_buffer.rs::from_iter[Arrow<arrow_buffer::buffer::scalar+1%
2.9 µs2.8 µs
ScalarBuffer<i32>>, 128]
vortex-buffer/benches/vortex_buffer.rs::map_each[Arrow<arrow_buffer::buffer::scalar+1%
3 µs3 µs
take_map[(0.05, 0.1)]
vortex-array/benches/take_patches.rs+1%
597.9 µs592.4 µs
from_iter[Buffer<i32>, 128]
vortex-buffer/benches/vortex_buffer.rs+1%
4.1 µs4.1 µs
decompress[u16, (10000, 256)]
encodings/runend/benches/run_end_compress.rs0%
29.6 µs29.5 µs
ScalarBuffer<i32>>, 1024]
vortex-buffer/benches/vortex_buffer.rs::map_each[Arrow<arrow_buffer::buffer::scalar0%
7.1 µs7.1 µs
take_indices[(1000, 4)]
encodings/runend/benches/run_end_compress.rs0%
8.7 µs8.6 µs
from_iter[Buffer<i32>, 1024]
vortex-buffer/benches/vortex_buffer.rs0%
8.3 µs8.3 µs
Commits
Click on a commit to change the comparison rangeBase
develop
d1097c5
+34.26%
chore: clean cancelled read requests earlier in driver state transitions
57159f2
1 day ago
by caicancai