Siu Kwan Lam
Siu Kwan Lam
I need to check this against https://github.com/numba/numba/pull/9259 which might help
@gmarkall did https://github.com/conda-forge/numba-feedstock/pull/131 for numba 0.59.0rc1
TODO: - [x] Add a config to disable the fail cache.
TODOs: - [ ] check https://github.com/numba/numba/issues/9375#issuecomment-1893955887
Potential solution: https://github.com/orgs/community/discussions/39062#discussioncomment-4138177
Right, the two variants are semantically identical. It is possible to implement such rewrite in Numba, however, the current Numba infrastructure is not designed to be friendly for such transformations....
From triage meeting: The two variants used to have similar performance until a patch for iterating over ND-array was merged. ~@stuartarchibald may remember the PR number.~ Probably is https://github.com/numba/numba/issues/7753#issuecomment-1015309956
@gmarkall, I've updated the chart and added explanations. The numpy2.0 work is also moved to a more accurate timeframe around late Q4 and early Q1 as that is roughly when...
superseded by https://github.com/numba/numba/issues/9540