bosilca

More fences was never a solid way to fix conceptual bugs. We care about scale, just adding more fences on the least performance network is not desirable, especially at scale....

Awesome! OMPI has support for a mechanism allowing to start a component (ompi/mca/hook) at different stages of the life cycle of an MPI process, such as the beginning of MPI_Init...

I agree with @ggouaillardet why would that matter you only care about local addresses and remote displacements when dealing with windows. Why is the naming scheme important ?

We're talking about a shortcut that would require one process to know how a different process named it's windows. Basically, a delegation mechanism, where instead of applying some operation directly...

I love the comments in that file `# Write out a host file because OMPI's mpiexec is dumb` !! Fascinating

Exactly what I though, they use a non-MPI feature to implement support for delegation. They identified a few solutions, including the correct solution, aka. compute a unique ID in CAF,...

Here is a reproducer highlighting that CAF assumption that MPICH Window ID are unique is generally incorrect, and it only works if all windows are created globally. In that case,...

@jsquyres the CAF programming model is tailored for the world model. Images can fail (capability exists but I am not sure about support) but there is no way to add...

There was no response because there is little we can do on the MPI side here as we do not control the daemons. Setting aside the question related to resilience,...

For parsec this is not entirely accurate. The size of the data (to be precise the count) to be received was always part of the header, and once receive stored...