Closed beckermr closed 1 month ago
pre-commit.ci autofix
pre-commit.ci autofix
pre-commit.ci autofix
Comparing beckermr-patch-1
(de0aec2) with main
(d755064)
⚡ 1
improvements
❌ 3 (👁 3)
regressions
✅ 7
untouched benchmarks
Benchmark | main |
beckermr-patch-1 |
Change | |
---|---|---|---|---|
👁 | test_benchmarks_lanczos_interp[kval-no_conserve_dc-run] |
443.8 µs | 645.8 µs | -31.28% |
⚡ | test_benchmarks_lanczos_interp[xval-conserve_dc-compile] |
142.9 ms | 129.8 ms | +10.11% |
👁 | test_benchmarks_lanczos_interp[xval-conserve_dc-run] |
877.9 µs | 977.5 µs | -10.19% |
👁 | test_benchmarks_lanczos_interp[xval-no_conserve_dc-run] |
731.8 µs | 829.8 µs | -11.81% |
@ismael-mendoza This PR is ready for review. As far as I can tell, the performance regressions are changes from jax+jaxlib that have exposed more of the underlying sin calls to codspeed. So I am unconcerned about them.
This PR moves us to ruff and ruff-format to avoid issues with flake8 pyproject support.
closes #108