Re: [OMPI users] Help with Binding in 1.8.8: Use only second socket

I tried the following with OpenMPI 1.8.1 and 1.10.1. The both worked. In my case a node has 2 sockets like yours, but each socket has 12 cores and lstopo showed core numbers for the second socket are from 12 to 23. * mpirun --report-bindings --bind-to core --cpu-set 12,13,14,15,16,17,18,19 -np 8 j

Re: [OMPI users] Help with Binding in 1.8.8: Use only second socket

Ralph, Huh. That isn't in the Open MPI 1.8.8 mpirun man page. It is in Open MPI 1.10, so I'm guessing someone noticed it wasn't there. Explains why I didn't try it out. I'm assuming this option is respected on all nodes? Note: a SmarterManThanI™ here at Goddard thought up this: #!/bin/bash rank=

Re: [OMPI users] Help with Binding in 1.8.8: Use only second socket

Try adding —cpu-set a,b,c,… where the a,b,c… are the core id’s of your second socket. I’m working on a cleaner option as this has come up before. > On Dec 21, 2015, at 5:29 AM, Matt Thompson > wrote: > > Dear Open MPI Gurus, > > I'm currently trying to do something

[OMPI users] Help with Binding in 1.8.8: Use only second socket

Dear Open MPI Gurus, I'm currently trying to do something with Open MPI 1.8.8 that I'm pretty sure is possible, but I'm just not smart enough to figure out. Namely, I'm seeing some odd GPU timings and I think it's because I was dumb and assumed the GPU was on the PCI bus next to Socket #0 as some