Cisco’s Jeff Squyres writes that Open MPI’s revamped v1.7 affinity system is actually quite flexible. Open MPI v1.7 uses hwloc to probe the topologies of the nodes on which it is launching processes.
Cisco’s Jeff Squyres writes that the new K supercomputer in Japan used OpenMPI to achieve 93 percent efficiency on 8 petaflop runs: “Several extensions have been made because of low latency ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results