From chemistry-request@server.ccl.net Fri Mar 8 14:34:55 2002 Received: from hermes.bmc.uu.se ([130.238.39.245]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id g28JYs519753 for ; Fri, 8 Mar 2002 14:34:54 -0500 Received: from XRAY.bmc.uu.se ([130.238.37.10]) by hermes.bmc.uu.se (Post.Office MTA v3.5.3 release 223 ID# 0-62649U1500L1000S0V35) with ESMTP id se for ; Fri, 8 Mar 2002 20:34:14 +0100 Received: from zorn.bmc.uu.se (IDENT:spoel@zorn.bmc.uu.se [130.238.37.95]) by XRAY.bmc.uu.se (8.11.6/8.8.7) with ESMTP id g28JYpb12891 for ; Fri, 8 Mar 2002 20:34:51 +0100 Date: Fri, 8 Mar 2002 20:34:03 +0100 (CET) From: David van der Spoel X-Sender: spoel@zorn.bmc.uu.se cc: chemistry@ccl.net Subject: Re: CCL:FYI: superlinear speedups in GROMACS In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII On Fri, 8 Mar 2002, Eugene Leitl wrote: >Ask David VanDerSpoel, he'll tell you that they see superlinear speedups >in Gromacs jobs quite often. Has to do with the cache, I'd guess something >like with one CPU the job cant fit in L2 cache by itself, so lots of >main memory accesses are required - when the job is split for 2 cpus, >more of the job is in L2 caches, and despite the shared memory that LAM >uses for communications for dual nodes, the overhead for that doesnt outweigh >the advantage of having the job divided into more cache ram... works >out well! :) It works on IBM Sp2 as well, and is very nice indeed on Scali networks: http://www.gromacs.org/benchmarks/scaling.php Our memory access pattern is "quite serial", but not entirely, that's why the cache improves performance so much. Groeten, David. ________________________________________________________________________ Dr. David van der Spoel, Biomedical center, Dept. of Biochemistry Husargatan 3, Box 576, 75123 Uppsala, Sweden phone: 46 18 471 4205 fax: 46 18 511 755 spoel@xray.bmc.uu.se spoel@gromacs.org http://zorn.bmc.uu.se/~spoel ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++