Any news on Infiniband ?

Josip Loncaric josip at nianet.org
Tue Mar 11 12:10:54 PST 2003


Patrick Geoffray wrote:
> On Thu, 2003-02-27 at 00:46, Anthony Skjellum wrote:
>>
>>We had good luck with MPI/Pro over VI
> 
> [...]
> The good part of Giganet was, IMHO, the packet engine built on top of a
> ATM chip, with a very good medium message pipeline. [...]
> But let me remind you the current state
> of VI: dead in the water.

Sadly, Giganet cLAN is no longer supported on Linux clusters, but it worked 
well with 2.2 kernels and MPI/Pro using interrupt driven receive.  We could 
productively bring more nodes into the calculation, and the overall code 
performance often doubled.

Oh well, life moves on.

Sincerely,
Josip

P.S.  Dell performance comparisons of NAS Parallel Benchmarks using MPICH (and 
friends) over Fast Ethernet, Gigabit Ethernet, Giganet, and Myrinet can be 
found at:

  http://ftp.us.dell.com/app/3q01-hpc.pdf

This is similar to my own tests of Fast Ethernet vs. Giganet cLAN (w/various 
MPI libraries) reported at http://research.nianet.org/~josip/MPIonCoral.html

-- 
Dr. Josip Loncaric, Research Fellow              mailto:josip at nianet.org
National Institute of Aerospace       http://research.nianet.org/~josip/
144 Research Drive                       mailto:j.loncaric at larc.nasa.gov
Hampton, VA 23666, USA         Tel. +1-757-766-1395  Fax +1-757-766-1401




More information about the Beowulf mailing list