Brad just curious.

Did you tweak any other values for starting and running a job on such a large system?  You say unmodified,  but OpenMPI lets you tweak many values at runtime.  

I would be curious to expand what I know from what you discovered.

Brock Palen
Center for Advanced Computing

On Jun 16, 2008, at 10:12 PM, Brad Benton wrote:
Greetings Open MPI users; we thought you'd be interested in the 
following announcement...

A new supercomputer, powered by Open MPI, has broken the petaflop 
barrier to become the world's fastest supercomputer.  The 
"Roadrunner" system was jointly developed by Los Alamos National 
Laboratories and IBM.  Roadrunner's design uses a cluster of AMD 
dual-core processors coupled with computational accelerators based 
on the IBM Cell Broadband Engine.  The cluster consists of 3,060 
nodes, each of which has 2 dual-core AMD processors and associated 
Cell accelerators.  The AMD nodes are connected with 4x DDR 
InfiniBand links.

Open MPI was used as the communications library for the 12,240 
processes comprising the Linpack run which broke the Petaflop 
barrier at 1.026 Petaflop/s.  The version of Open MPI used in the 
run-for-record was a pre-release version of the upcoming 1.3 
release.  Enhancements in this release include modifications for 
efficient, scalable process launch.  As such, Open MPI was run 
unmodified from a snapshot of the pre-1.3 source base (meaning: 
there are no Roadrunner-specific enhancements that are unportable to 
other environments -- all Open MPI users benefit from the 
scalability and performance improvements contributed by the 
Roadrunner project).

--Brad Benton
Open MPI/Roadrunner Team
users mailing list