AMBER: SPARC benchmarks (was re Fujitsu PRIMEPOWER)

From: Atro Tossavainen <atro.tossavainen+amber.helsinki.fi>
Date: Tue, 11 Mar 2008 16:33:33 +0200 (EET)

Dear all,

Bob Duke suggested I post these, although I don't know whether it's of
any use to anyone. :-)

I got access to the Sun Fire 25K at csc.fi and ran these benchmarks
there. It's a 48+48 core (UltraSPARC IV . 1.2 GHz) box with 192+192
GB RAM. I used the Sun Studio 9 compilers there and Sun MPI. This
is on AMBER 8.

It seems that the difference between sander and pmemd is really quite
remarkable on the SPARC platform. I've been getting improvements of
up to 4.2x. Put another way, sander is really painful to use on SPARCs.
I don't know why this is the case, though.

JAC benchmark, single CPU, sander: 22.1 ps per day.

JAC benchmark, pmemd:

CPUs ps per day
1 74.0
2 137.0 (92.6%)
4 271.4 (91.7%)
8 524.7 (88.6%)
16 994.5 (84.0%)
32 1835.2 (77.5%)

The scaling is on par with the SDSC DataStar, loses a little bit
to the SGI Origin 3800, but seems to beat the Altixes.

-- 
Atro Tossavainen (Mr.)               / The Institute of Biotechnology at
Systems Analyst, Techno-Amish &     / the University of Helsinki, Finland,
+358-9-19158939  UNIX Dinosaur     / employs me, but my opinions are my own.
< URL : http : / / www . helsinki . fi / %7E atossava / > NO FILE ATTACHMENTS
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Wed Mar 12 2008 - 06:07:43 PDT
Custom Search