bosilca

Results 318 comments of bosilca

I am certain no code currently in OMPI optimizes the collective communication pattern based on rank closeness, the code stays away from such assumptions because as you mentionned they are...

Not sure what I'm expected to provide here ?

I cannot reproduce with main (v2.x-dev-9807-g0d126c6405) compiled in release mode. I increased the number of threads (THREAD_COUNT) up to 16, and run each test 100 times. Everything went smooth.

Let's try something else before we go to such testing lengths. Assuming you are using the OB1 PML with the BTLs tcp and self (this will force all non-local communications...

You can call it in any context, gdb will execute it in it's own context. I think the segfault is because you are missing the second argument for the call...

Can't replicate on a setup with a single IP address. This, as well as the assert location reported by @devreal might indicate that the issue is in the code handling...

I tried to replicate on the same hardware as @devreal, but so far without success. I will talk with him tomorrow to see if he can share a setup where...

The assertion only triggers in a scenario with multiple IP network interfaces, as soon as you restrict to a single interface you should never see the assert. Back to the...

Too much smartness kill all smartness. The issue here is two folds: the MPI standard require all unknown keys to be removed, and our communicator creation and initialization is done...

These are bandaid solutions, tailored for a specific case. The real problem is the info keys are attached in the middle of the process, too late to be used during...