bosilca
bosilca
Please rebase to current main to get rid of the `mpi4py` failure.
I saw your question on MPICH mailing list, but it was slightly different than what you asked on OMPI user mailing list or what you state on this issue. Because,...
Makes sense, we gain peanuts on the small send but we pay the branching/testing cost for all multiple test/wait requests. Let's put aside the discussion about performance, the code we...
that's our default. Literally everyone in the world is doing these checks.
A slightly different approach (#12691) was used to solve this.
A fair question is what is SchedMD seeing as a potential development path in which none of the parties constraint the development of the others ?
1. Then we should provide a similar patch for all instances of string storage in the entire OMPI code base, not fix just one that happen to have highlighted a...
Both processes initialize the BTLs, so it seems obvious they both selected the OB1 PML. But somehow that information is now properly exchaged between them. More precisely, it sounds as...
@rhc54 that is not correct, you seems to be confusing the "correct" and the "bad" output. If you take a look at the complete output posted you will notice the...
Do you need to run with the OFI? provider ? The problem with the workaround is that your local communications are now using sockets instead of shared memory, leading to...