From owner-chemistry@ccl.net Tue Oct 31 10:31:12 1995 Received: from bedrock.ccl.net for owner-chemistry@ccl.net by www.ccl.net (8.6.10/950822.1) id KAA10429; Tue, 31 Oct 1995 10:29:37 -0500 Received: from genie.terra.co.il for rvn@genie.terra.co.il by bedrock.ccl.net (8.6.10/950822.1) id KAA15515; Tue, 31 Oct 1995 10:29:23 -0500 Received: from genie.terra.co.il (localhost [127.0.0.1]) by genie.terra.co.il (8.6.12/8.6.12) with SMTP id PAA01488 for ; Tue, 31 Oct 1995 15:27:14 GMT Message-Id: <199510311527.PAA01488@genie.terra.co.il> Date: Tue, 31 Oct 95 17:27:15 0200 Sender: rvn@genie.terra.co.il From: Reuven Bakalash X-Mailer: Mozilla 1.1N (X11; I; IRIX 5.3 IP22) MIME-Version: 1.0 To: chemistry@ccl.net Subject: New parallel platform for Computational Chemistry X-URL: file:/users/genie/rvn/rvn/tr2000ron Content-Type: multipart/mixed; boundary="-------------------------------182321470316392" This is a multi-part message in MIME format. ---------------------------------182321470316392 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=us-ascii > CCL contact information > > This is the contact information which is currently appended to every message > sent to the CCL mailing list (to reduce resource use, it has been removed from > the searchable archive): > > CHEMISTRY@ccl.net -- everyone | CHEMISTRY-REQUEST@ccl.net -- coordinator > MAILSERV@ccl.net: HELP CHEMISTRY | Gopher: www.ccl.net 73 > Anon. ftp www.ccl.net | CHEMISTRY-SEARCH@ccl.net -- archive search > http://www.ccl.net/chemistry.html | for info send: HELP SEARCH to MAILSERV > > Here is the same information reformatted, with active WWW links: > > Post a message to CCL > chemistry@ccl.net > Send email to CCL coordinator > chemistry-request@ccl.net > CCL WWW home page > http://www.ccl.net/chemistry.html > CCL gopher interface > gopher://www.ccl.net:73/ > CCL anonymous FTP site > ftp://www.ccl.net > CCL email server. Use the subject "HELP CHEMISTRY" to obtain information. > mailserv@ccl.net > CCL archive search (for help send email to mailserv@ccl.net with subject "HELP > SEARCH") > chemistry-search@ccl.net > > For information about subscribing to the CCL, see the document Information > About Computational Chemistry List at the CCL WWW site. > > ------------------------------------------------------------------------------- > [Return to MedChem home page] > > David Hoekman > webmaster@medchem.claremont.edu ---------------------------------182321470316392 Content-Transfer-Encoding: 7bit Content-Type: text/plain TR2000 - A new powerful computing resource for CC ------------------------------------------------- ___________________________________________________________ | * Significant time reduction in long computational tasks| | * A fraction of cost of comparable systems | | * Full range of Software tools | | * A personal computing resource | ----------------------------------------------------------- TR2000 is a desk side parallel computing resource with UNIX front end. 4 to 32 Alpha processors with distributed memory (up to 64MB each). The TR2000 architecture offers a peak performance of 2.5 GFLOPS for Alpha 24064-150 MHz, and 5 GFLOPS for Alpha 24064-275 MHz. The performance scales linearly with each additional processor. TR2000 provides parallel applications in computational chemistry, such as Molecular Dynamics, Lattice Field Theory, Monte Carlo etc. It also provides libraries for porting serial packages. For Molecular Modeling it runs MOIL and CHARMMm23 with exceptional results. Here are comparative figures of leading computers running MbCO + 3830 water molecules (total 14026 atoms), 12-14A shift using FAST option in CHARMm23. All indicated times are in hours (run times are provided by M. Hadocek, NIH): COMP. SYS. # PROC. RUN TIME ---------- ------- -------- SGI/Indy 1 8.93 SGI/Pwr.Indigo2 1 2.93 IBM/SP2 16 0.28 Cray T3D 16 0.77 SGI/PChallenge 16 0.21 TR2000 16 0.65 Status: Two beta-sites are installed at : * Hebrew University, Jerusalem, Ron Alber - since 9/95 * Mount Sinai Medical Center, NY, Harel Weinstein - since 10/95 Prof. Ron Elber recommends(29/10/195): " The Terra-2000 super-workstation gave a significant boost to our abillities to pursue lengthy, numerically intensive calculations. The Terra-2000 - a parallel computer with 16 alpha 150 MHz processors, is employed in our laboratory as a "real" parallel computer running a single task. It is also employed as a cluster of workstations running the same program with different inputs.In both modes, the new computer provides unprecent performance in the workstation area. It makes it possible for us to pursue new calculations with significantly more statistics or simulations of larger systems." Contact: rvn@mail.terra.co.il ---------------------------------182321470316392-- From rcfort@maine.maine.edu Tue Oct 31 16:01:15 1995 Received: from MAINE.maine.edu for rcfort@maine.maine.edu by www.ccl.net (8.6.10/950822.1) id PAA19508; Tue, 31 Oct 1995 15:50:22 -0500 Date: Tue, 31 Oct 1995 15:50:22 -0500 Message-Id: <199510312050.PAA19508@www.ccl.net> Received: from rfort3.umeche.maine.edu by MAINE.maine.edu (IBM VM SMTP V2R3) with TCP; Tue, 31 Oct 95 15:48:30 EST X-Sender: rcfort@maine.maine.edu (Unverified) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: chemistry@www.ccl.net From: rcfort@maine.maine.edu (Raymond C. Fort, Jr.) Subject: Summary of X-clients X-Mailer: Here is the promised summary of responses on the topic of X-windows clients. Last week I asked the question: >I have recently received an NSF-ILI grant for the purchase of SGI >workstations that will be used to introduce molecular modeling into >the undergraduate and graduate curricula. Our plan is to run SPARTAN, >PCModel, and MacroModel on the workstations; all are available as >X-applications. Then we will provide student access from a Department >PC cluster by running X-terminal emulators on the PCs. >I would be interested to hear from persons who have successfully >applied this kind of connectivity to modeling software, even if >it did not involve the specific programs cited. What X-term >software is best? (We currently are looking at PC-Xware from NCD >and XoftWare from Age). What problems did you encounter? The most common response was to put LINUX on the PCs and use Xfree86. Fred Arnold (fparnold@structures.uchicago.edu), Kris Boulez (kris@bionmrl.rug.ac.be), Carlo Nervi (nervi@chpc06.ch.unito.it), Susan Jackels (sjackels@ac.wfu.edu), Carl David (DAVID@UConnVM.UConn.edu), and Konrad Hinsen (hinsenk@ere.umontreal.ca) suggested this option, and Konrad was good enough to summarize the best points of LINUX: - Cost. Linux is free and doesn't demand huge resources. - Flexibility. A Linux system is an independent (if not very powerful) workstation, which is sufficient for the needs of many students. So we could reduce the number of users on the "big" stations, and offer accounts to more students. - Administration. There is no shortage of graduate students who Linux very well (from running it on their own machines athome). So system administration was not a problem, after the initial installation. - Independence. A Linux system contains its own complete set of X software with fonts etc., whereas an X terminal may need an external font server, thereby increasing network load and creating a dependence on a specific machine (we often had network problem that could cut off some machine from the local Ethernet). This is the solution I will employ for access from my own PC, where I have a half gig LINUX partition installed. However, the PC cluster is also used by large numbers of people running Windows apps, and so this is not a solution for student access. Also many of our faculty are Mac users. The next most frequent suggestion was eXceed, from Hummingbird software (http://www.hcl.com/HOWTO.HTML). Positive experiences were reported by Fred Arnold (address above); Tom Pierce (rs0thp@rohmhaas.com), who used it specifically to access SPARTAN and MacroModel; and Ned Martin (martinn@uncwil.edu), accessing SPARTAN. eXceed also was mentioned in a very helpful summary from the CUR list provided by Mark Timken (timken@science.widener.edu). I have ordered a trial copy. MacX was the only software mentioned for Macintoshes, with positive experiences reported by Stephen Bowlus (BOWLUS@sandoz.com), Per Arvidsson (arvid@oc.chalmers.se), Dave Young (young@slater.cem.msu.edu), and Craig Burkhart (cburkhart@goodyear.com). We have some small experience with MacX ourselves, and find it usable, although a bit clunky. Other clients receiving some comment were XVision 6.0, used by Andrey Bliznyuk (Andrey.Bliznyuk@anu.edu.au) to access SPARTAN, but found poor by Tom Pierce (address above); ReflectionX (Stephen Bowlus; address above); the X-win32 shareware client suggested by Carlo Nervi, which Jim Macmillan (macmillan@cobra.uni.edu) found of poor quality. From an ad in BYTE, I had learned of XoftWare for Windows, from AGE Logic, Inc. (sales@age.com). Katie Breen was particularly helpful in faxing me a copy of the sales brochure. Xoftware comes in versions for both PC and Mac; Craig Martens (cmartens@uci.edu) has a positive experience with the Mac version. We will order trial copies. Similarly, we learned of PC Xware, from NCD, Inc. (info@pcx.ncd.com; http://www.ncd.com); no one seems to have had experience with this software, but from its literature it appears to be an especially robust application. It does not come in a Mac version, and is pricier than Xoftware. Several individuals suggested that accessing SPARTAN and other programs from more than 3-4 clients produced an unacceptable slowdown; Jim Macmillan suggested that the servers should be on their own branch of the local network to avoid untowrd effects on other users. Several respondants also reminded me that Wavefunction, supplier of SPARTAN, requires a $200 license for each client that accesses SPARTAN over a network. This charge is unfortunate; we will probably have to acquire licenses over several years to reach our full intended capacity. My thanks to all who responded, and apologies if I have inadvertantly omitted anyone's contribution. I will file a report later on our experiences with examination copies. Professor Ray Fort, Jr. Voice: (207)-581-1180 Department of Chemistry FAX: (207)-581-1191 University of Maine E-mail: rcfort@maine.maine.edu Orono ME 04469 rfortjr@fort.umeche.maine.edu Computational organic chemistry, especially on wood- and paper-related problems