Re: [OMPI devel] default mapping on master vs v2.x

2016-05-18 Thread Ralph Castain
I don’t think your stated analysis is quite correct. First, the topology is -always- retrieved. The only question is whether or not we set the #slots equal to the number of detected cpus. If the user specifies the #slots, then we respect that designation. If the user does not specify #slots, th

Re: [OMPI devel] default mapping on master vs v2.x

2016-05-18 Thread Gilles Gouaillardet
Folks, i ran some more test and found this with both master and v2.x : mpirun --host n0:16,n1:16 -np 4 --tag-output hostname | sort [1,0]:n0 [1,1]:n0 [1,2]:n0 [1,3]:n0 and same output with the --map-by socket option. now, without specifying the number of slots per hosts, and the --oversub