Improved performance and solved performance regressions of v2 against v1#405
Improved performance and solved performance regressions of v2 against v1#405alejandro-vaz wants to merge 4 commits intoservo:v2from
Conversation
|
this closes #404 |
|
The issue should be now fixed, I just accidentally removed some |
|
hey, the gh actions workflow is waiting for maintainer approval to start. just leaving this comment here so whenever someone sees it can trigger the rerun (it should now be fine, see above) |
|
it should be now fixed
it does seem that some benchmarks got a bit slower but it's within ranges so it might just be noise (anyway, it'd be 5% at most) |
|
it's ready to be reviewed. also, it closes #404 I'm open to feedback and suggestions |
Improvements
This PR solves the performance regressions v2 introduced. Its most meaningful improvements are (in nanoseconds):
bench_extend: 80.48bench_extend_filtered: 82.81bench_extend_filtered_small: 7.26bench_extend_from_slice: 82.10bench_extend_from_slice_small: 9.80bench_insert_push: 307.51bench_insert_push_small: 39.03bench_insert_small: 116.55bench_push: 277.76bench_push_small: 41.78bench_push_vec: 213.06bench_push_vec_small: 27.75bench_pushpop: 303.43bench_pushpop_vec: 143.72bench_remove_small: 77.25Some benchmarks score faster than on the v1 implementation.
Added a test for safe unwind.
Performance