CCL:FYI: superlinear speedups in GROMACS (fwd)

Eugene Leitl Eugene.Leitl at lrz.uni-muenchen.de
Fri Mar 8 12:10:29 PST 2002


-- Eugen* Leitl <a href="http://leitl.org">leitl</a>
______________________________________________________________
ICBMTO: N48 04'14.8'' E11 36'41.2'' http://www.leitl.org
57F9CFD3: ED90 0433 EB74 E4A9 537F CFF5 86E7 629B 57F9 CFD3

---------- Forwarded message ----------
Date: Fri, 8 Mar 2002 13:28:37 -0600 (CST)
From: Fred P. Arnold <fparnold at chem.northwestern.edu>
To: Eugene Leitl <Eugene.Leitl at lrz.uni-muenchen.de>
Cc: chemistry at ccl.net
Subject: Re: CCL:FYI: superlinear speedups in GROMACS

We saw benchmarks of GAMESS-US on the IBM SP2 years ago that did this.
Same issue; eventually the pieces are small enough to fit into cache, and
it takes off like a shot, as long as communication overhead doesn't begin
to dominate the calculation.

						-Fred

					Frederick P. Arnold, Jr.
					NUIT, Northwestern U.
					f-arnold at northwestern.edu






More information about the Beowulf mailing list