eth-cscs / DLA-Future

DLA-Future
https://eth-cscs.github.io/DLA-Future/master/
BSD 3-Clause "New" or "Revised" License
64 stars 14 forks source link

Add test stage for ROCm configuration #982

Closed msimberg closed 2 weeks ago

msimberg commented 1 year ago

This won't work yet...

msimberg commented 1 year ago

cscs-ci run

msimberg commented 1 year ago

cscs-ci run

msimberg commented 1 year ago

cscs-ci run

msimberg commented 1 year ago

cscs-ci run

msimberg commented 1 year ago

cscs-ci run

msimberg commented 1 year ago

cscs-ci run

msimberg commented 1 year ago

This is currently failing due to a buggy Cray MPICH used by sarus. The buggy MPICH leads to the all but the first MPI executable failing to initialize MPI in a SLURM job. Only the CUDA-aware MPICH seems to be buggy. This is waiting for a change to sarus which will use a non-CUDA-aware MPICH by default, and allow choosing the CUDA-aware one explicitly if needed.

msimberg commented 2 weeks ago

I think by now this is so out of date that it's easier to start over once we actually have access to AMD GPUs for CI again, so I'll close this.