Sylwester Arabas
Sylwester Arabas
@Sfonxu, I'm looking around and cannot find any trace of a record of which is the actual Numba version where the incompatibility appeared, do you have any record of this...
but on CI we're using newer releases, and our sanity test passes, and there are no timeouts... https://github.com/open-atmos/PyMPDATA/blob/ed8d8c1c7dc6f0d921cb6253dd4e2d4be3be4e2a/setup.py#L34
however, for Linux, we also only test with OpenMP! https://github.com/open-atmos/PyMPDATA/blob/ed8d8c1c7dc6f0d921cb6253dd4e2d4be3be4e2a/.github/workflows/tests.yml#L137-L138
while Colab's default is [Intel TBB](https://www.google.com/url?sa=t&source=web&rct=j&opi=89978449&url=https://pl.wikipedia.org/wiki/Intel_Threading_Building_Blocks&ved=2ahUKEwjr4sGH1IaRAxUlkcMKHbSkC8sQFnoECBwQAQ&usg=AOvVaw3ADJw93QsG2rD_Jc2IUAdt):
and Numba docs quite explicitly mention that TBB might use fewer threads than other backends (https://numba.pydata.org/numba-doc/dev/developer/threading_implementation.html#caveats):
so it seems, that we should "simply" require that anyone using PyMPDATA picks `omp` as Numba Threading layer
well. the warning message that we already print actually does suggest it!
> It is possible to change the backend on Colab via adding this before any Numba-dependent code is run: > > import numba numba.config.THREADING_LAYER="omp" We should then add it to...
a check for setting NUMBA_THREADING_LAYER in all notebooks is being worked on here: https://github.com/open-atmos/devops_tests/pull/52 Thanks @AgnieszkaZaba !
@AgnieszkaZaba, I don't understand the issue title, could you please elaborate :)