From chemistry-request@server.ccl.net Sat May 10 10:24:50 2003 Received: from exasp.biogeo.uw.edu.pl ([212.87.8.1]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id h4AEOot30292 for ; Sat, 10 May 2003 10:24:50 -0400 Received: from localhost (localhost.localdomain [127.0.0.1]) by exasp.biogeo.uw.edu.pl (Postfix) with ESMTP id 72D3C15E84 for ; Sat, 10 May 2003 16:24:57 +0200 (CEST) Received: by exasp.biogeo.uw.edu.pl (Postfix, from userid 526) id C213615E2F; Sat, 10 May 2003 16:24:55 +0200 (CEST) Date: Sat, 10 May 2003 16:24:55 +0200 From: Grzegorz Bakalarski To: chemistry@ccl.net Subject: PC clusters for CompChem - next survey Message-ID: <20030510142455.GB27001@exasp.biogeo.uw.edu.pl> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.4i X-Virus-Scanned: by AMaViS snapshot-20020300 Dear All, According to my search in CCL archives there was no discussion about clusters for computational chemistry since at least half a year, so I'd hope we can start next turnaround. I'd not like to start next war or any flames so I'd like to ask for concrete living examples or firm guidelines. At the moment there is also new opportunity because Itenium 2 (64-bit) processors become more popular and even more - the next generation AMD 64-bit Opteron procesors have just been become available. Still we can also consider very fast 32-bit Pentium IV or Xeon from Intel or Athlons from AMD (with speed from 2 to over 3GHz). Having this in mind please share your opinions on current trends in building PC clusters for use in computational chemistry. To be more specific one can think about most popular programs like gamess, gaussian, amber, NWchem, Tripos , MSI programs - note: this should be rather general purpose computational cluster and not only for running QM or only MD ... It's obvious that price/performance is main criterium but still functionality, reliability, stability is of importance. Particular aspect are the following * size of nodes: 1 or 2 or 4 or more processors per node (up to now it seems that 2 processor nodes had best price/performance but what about new 4 processor Opteron motherboard) * performance: Opteron vs Itenium 2 vs Xeon (or maybe PIV or Athlons should be considered) * functionality: aren't Opterons (or even Itanium 2) too new - problems with software ? * stability: Athlons were known because of overheating - what about Opterons??? Still new motherboards are not well tested * the size of memory (I know that more is better, but what is enough): 2GB, 4GB, 8GB * interconnect: is 1Gb enough or should we consider other solutions (which ones?) ... the size of cluster maybe from 32 up to 128 nodes (depending on funds and prices) Any other comments or issues are welcome! Thanks for essential responses. Kind regards, P.S. I guess more are interested in, so please post to the list if possible. I'll summarize answers of course .... From chemistry-request@server.ccl.net Sun May 11 00:35:08 2003 Received: from ccl.net ([192.148.249.4]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id h4B4Z8t07112 for ; Sun, 11 May 2003 00:35:08 -0400 Received: from krakow.ccl.net (krakow.ccl.net [192.148.249.195]) by ccl.net (8.11.6+Sun/8.11.6/OSC 2.1) with ESMTP id h4B4Z8t18070; Sun, 11 May 2003 00:35:08 -0400 (EDT) Date: Sun, 11 May 2003 00:35:06 -0400 (EDT) From: Jan Labanowski To: chemistry@ccl.net, mel@asdn.net Subject: 03.10.10 11th Foresight Conf on Molec Nanotech, Burlingame, CA, USA Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII 11th Foresight Conference on Molecular Nanotechnology Thursday, October 9 -- Sunday, October 12, 2003 San Francisco Airport Marriott http://www.foresight.org/Conferences/MNT11/index.html [Pieces taken from Web Site by Jan Labanowski] The 11th Foresight Conference on Molecular Nanotechnology will be held at the San Francisco Airport Marriott, 1800 Old Bayshore Highway, Burlingame, CA 94010. The Conference will begin with a reception the evening of Thursday, October 9, and end Sunday, October 12, 2003. An intensive Tutorial on Foundations of Nanotechnology will be held on October 9. This 2003 Conference home page will provide the information you will need about the conference and to register for the conference. After the conference is over, this page will provide the permanent archive for the conference. Rapid advances in our ability to image, manipulate, and probe the properties of matter at the atomic scale-together with emerging insights into structure, function and self-assembly in biological systems-is bringing to fruition the tremendous promise of nanotechnology first recognized by Richard Feynman over 40 years ago. In the next decade, current research into the science and technology of nanostructures is expected to have a major impact on fields ranging from consumer electronics to space exploration and medicine. The Foresight Institute's first Conference on Nanotechnology, which pre-dated the National Nanotechnology Initiative by a decade, was the first comprehensive conference on the topic of nanotechology. Foresight sponsored events continue to be the premiere venue for discussing new and innovative multidisciplinary research in nanotechnology. Last year's conference, the 10th in the series, attracted researchers from academic, government and industrial laboratories world-wide, and included papers > from the electronics, medical, computing, and biological communities. Foresight's 11th Conference will continue this level of excellence by providing a forum in which leaders from all disciplines delving into science and technology at the nanoscale can present and discuss their most recent results and ideas. Submission information: Abstracts Due by June 2, 2003. Acceptance notices July 2003 There will be oral presentations and a poster session during the conference. The poster session will be held on Friday afternoon. On Saturday there will be additional time for viewing and discussion of the posters. All those who wish either to speak or to present a poster must submit an abstract. The deadline for submitting abstracts is June 2, 2003. The abstracts should be no longer than 500 words including references and footnotes. Each submitted abstract may include only one graphic in jpg or gif format. The size should be no larger than 400 x 400 pixels (will be displayed at 72 dpi). Abstracts not in the appropriate format will be returned without being considered. All accepted abstracts will be permanently available on the WWW at http://www.foresight.org