[Beowulf] mpich vs hp mpi performance

[Håkon Bugge]

At 19:47 26.09.2008, Joshua Baker-LePain wrote:

>DYNA has 2 main solver types -- explicit 
>(iteration based) and implicit (matrix 
>inversion, essentially).  If you're only using 
>one system and the explicit solver, then there's 
>no need for any MPI, as ls-dyna is multi-threaded.

I ran both SMP version and MPP version of ls-dyna 
on a dual-socket quad-core CloverTown. On the 
Neon refined revised data set, the SMP version 
achieved 25 jobs/24h. The MPP version achieved 36 
jobs/24h. Hence, if you're able to use the 
implicit solver, I advice you to use an MPI.


Thanks, Håkon


--
Håkon Bugge
Chief Technologist
mob. +47 92 48 45 14
off. +47 21 37 93 19
fax. +47 22 23 36 66
hbugge at platform.com
Skype: hakon_bugge

Platform Computing, Inc.