On 2/26/19 5:13 AM, Daniel Letai wrote:
I couldn't find any documentation regarding which api from pmix or ucx
Slurm is using, and how stable those api are.
There is information about PMIx at least on the SchedMD website:
https://slurm.schedmd.com/mpi_guide.html#pmix
For UCX I'd suggest testing the latest 1.5 release, I need to do that
myself when I can find some time as I found a SEGV issue in earlier
releases where there are multiple IB adapters in a node which is
apparently fixed in 1.5.
All the best,
Chris