From wagenert@Mailer.Uni-Marburg.DE Wed Oct 30 08:30:43 1996 Received: from Mailer.Uni-Marburg.DE for wagenert@Mailer.Uni-Marburg.DE by www.ccl.net (8.8.2/950822.1) id HAA03878; Wed, 30 Oct 1996 07:17:10 -0500 (EST) Received: by Mailer.Uni-Marburg.DE (AIX 3.2/UCB 5.64/20.07.94) id AA88376; Wed, 30 Oct 1996 13:16:36 +0100 From: wagenert@Mailer.Uni-Marburg.DE (Wagener Thomas) Message-Id: <9610301216.AA88376@Mailer.Uni-Marburg.DE> Subject: Compiling Gaussian for R10k and IRIX 6.2 To: CHEMISTRY@www.ccl.net Date: Wed, 30 Oct 1996 13:16:35 +0100 (CET) X-Mailer: ELM [version 2.4 PL25] Content-Type: text Hi everyone, last week I made a request on compiling Gaussian94 for the SGI PC with R10000 processors. Many thanks to all who answered especially Roberto Gomperts from SGI. **************************************************************************** From: "Roberto Gomperts" Hi Thomas, Gaussian was certified for the r10k based machines only starting with rev. D.3 Please contact Gaussian Inc. to get a new source and makefile. -- -- Roberto Roberto Gomperts "Gracias a la vida, roberto@boston.sgi.com que me ha dado tanto ..." phone: (508) 567 2251 Violeta Parra Fax: (508) 562 4755 **************************************************************************** However, the mails I received may be of general interest for the people using Gaussian on SGI computers so I am including all the responses I have received. Thank you very much, Thomas - "You are wrong! 2+2=5.8354! Please adjust your equipment accordingly." The Computer (your friend) ***************************************************************************** * Thomas Wagener * Tel: ++49-(0)6421-285561 * * FB Chemie * * * Phillips-Universitaet Marburg * Email: wagenert@mailer.uni-marburg.de * * * * * 35032 Marburg * * * Germany * * ***************************************************************************** From: Danne R Rasmussen Hello Thomas, We have G94 running here at the Australian National University Supercomputer Facility on an SGI PC with R10000 chips. I suggest you need to get a more current release of G94. From recollection we had troubles with versions prior to D.2 I believe the current version is D.4 I think there are also many valuable bug fixes between versions C.2 and D.4 Cheers, Danne From: Michael A McAllister Thomas, we also have a few R10K machines -- word from Gaussian directly is that they do NOT have a verified working version of G94 for the R10K processor (IP 28). That was as of about a month ago when I talkedto them. They were working on it, so i think there is hope that it will work eventually. I got around the problem by using R8K optimized binaries on the R10K. They seem to run fine and give correct test results. Interestingly, the R4400 based binaries do NOT give correst answers when run on the R10K. They do give correct answers when used on the R5K, however. Go figure. As I said, wedo have a few of these R10K's and would love to know if you do find someone who has a successful Makefile. Good luck. Mike McAllister ************************************************************************** Assistant Professor of Chemistry voice: 817-565-4584 University of North Texas fax: 817-565-4318 P.O. Box 5068 email: mam0008@jove.acs.unt.edu Denton, TX 76203 www: http://www.unt.edu/~mam0008 ************************************************************************** From: "Dr. Daniel L. Severance" Hi, I have done it. I picked up the latest version of G94 which had the updated Makefile. The people at Gaussian also said there were some program changes involved so you might want to call them. Dan From: "Brent H. Besler" I was only able to get G94 rev D4 working correctly by compiling it in R8000 mode. I just modified the g94 makefile so that the processor type was set to R8000. G94 gives correct results. In R10000 mode it dies usually in MP2 gradients or in HF gradients. From: Jeremy R Greenwood Dear Thomas, We had no end of trouble with G94 on R10000 until we ordered the D.3 revision, which worked first time. I highly recommend spending the $$$ on the update and not wasting more time. Hope this helps, Jeremy From: young@argus.cem.msu.edu Hi, We had problems getting g94 to run under 6.2 on 4400's and will be trying 10000's this week. We did finally get it to run under the 6.2. It required installing two patches to the BLAS libraries, one that came out a few months ago and one from a week or two ago. Hope this helps. Dave Young young@slater.cem.msu.edu -------------------------------------------------------------------------- Computational chemistry, teaching, visualization and programming. -------------------------------------------------------------------------- From: "C. S. Page" Dear Thomas, We have had a lot of fun trying to compile G94 on an R5000 under Irix 6.2, and, basically, the problem has not yet been resolved. The only fully working version that I have was compiled with the Mips 3 compiler option using a number of patches to eg. the BLAS routines. These will also help matters in the majority of cases (all of the test jobs will run, for example) for the Mips 4 version. The ones we have are: Patch SG0001262: libblas autoblock problem Patch SG0001378: Complex Arithmetic Broken in 6.2 libftn Patch SG0001436: fix for branch in branch delay slot in libfastm Patch SG0001496: libblas generates incorrect answers on R5000/R10000 An alternative is to use the Challenge Complib libraries instead BLAS: the BLAS patches are quite large (~50Mb in total) and you may wish to forget about them. If you have any trouble getting hold of the patches, please let me know and I will e-mail the distribution. I will also see if I can find the input deck that runs on an R4400, but gives nan's in l502 on an R5/10k using the Mips 4 libraries: if you can get this one to run, your problems are over! We have been in touch with SGI and they have been looking into the matter for some time. There will be a posting to CCL when there is a solution. I hope that this is of some help to you. Chris Page __________ Department of Chemistry Tel: +44 (0)171 594 5795 Imperial College GB-LONDON SW7 2AY From: grzesb@asp.biogeo.uw.edu.pl (Grzegorz Bakalarski) Dear Dr Weagner To compile g94 on R10000 you could probably use -mips4 -64 and -r10000 flags. Probably you also should specify 64 bit pointers: -DP64. As I know there are more specific options (e.g. rounding off, IEEE aritmethics ....) Moreover I am not sure if rev C.2 has been prepared to run on R10000. Even for Rev. D.1 I can see support for R8000 PC. So probably you will have more problems Best regards, G Bakalarski ( grzesb@icm.edu.pl) From: Johan Landin Hello Thomas, Maybe you need a later revision of Gaussian 94 in order to succeed? I believe that the first revision certified for IRIX 6.2 and R10000 is Gaussian 94 Revision D.3. However, since I have not checked this explicitly myself I suggest that you contact Gaussian Inc. directly in these matters. If your university already has a licence for Gaussian 94 the cost for a new tape, with the latest revision, will be only $200. Regards, /Johan Landin -- ___________________________________________________________________ Johan Landin Tel: +46 31 773 3767 Dept. of Medical Biochemistry Fax: +46 31 41 6108 Medicinaregatan 9 Home: +46 31 14 7554 S-413 90 Goteborg, Sweden Email: landin@mednet.gu.se From: "Jean Francois Marquis" hi Thomas and Dorothee, It seems like you have some problems to re-compile GAUSSIAN on our machine. Did you ever compile this code on SGI with IRIX 5.3 ? What cc command line did you use ? Could you verify what versions of compiler is installed on your SGI by the command "versions -b | grep compiler" Some suggestions described in few steps : First if you use Kernigham and Richie C put -cckr option in the cc invocation line, in other case ANSI C is the default. Second As you know IRIX 6.2 and R10000 are fully 64 bits stuff... so if this code is 32 bit (i mean addresses, pointers...) you could begun to re-compile it with these options "cc -32 ...." If it's OK, go ahead with "cc -n32 .." which translate "old 32 bits code" in "new 32 bits one" and will be more efficient on the R10K. And at last you can if needed take advantage of the 64 bits addressing (pointers,tabs, and files greater than 2Go in memory) but be careful on the code. Some cast like (int) p where p is a pointer (consquently 64 bits) will give a false or non sense result... Hope this helps, jef. -- ================================================================ = Jean-Francois MARQUIS = Pre-sales engineer = SGI France AIX-EN PROVENCE Office = Parc club du golf, bat 20 = Av. Guillibert de la Lauziere = BP 107000 Aix-en-Provence cedex 3 = FRANCE ( ) ____ ( ) _/__|__\___ ( -- (_ _|__ __] AIX -- () () TOULOUSE = Tel : (33) 04 42 39 77 79 = Fax : (33) 04 42 39 77 50 = E-Mail : jef@aix.sgi.com ================================================================ From ccl@www.ccl.net Wed Oct 30 12:17:59 1996 Received: from bedrock.ccl.net for ccl@www.ccl.net by www.ccl.net (8.8.2/950822.1) id MAA06032; Wed, 30 Oct 1996 12:13:01 -0500 (EST) Received: from hil-img-1.compuserve.com for 100317.2656@CompuServe.COM by bedrock.ccl.net (8.8.2/950822.1) id MAA06239; Wed, 30 Oct 1996 12:12:34 -0500 (EST) Received: by hil-img-1.compuserve.com (8.6.10/5.950515) id MAA18457; Wed, 30 Oct 1996 12:11:56 -0500 Date: 30 Oct 96 12:09:05 EST From: Herbert Koeppen <100317.2656@CompuServe.COM> To: "Comput. Chem. List CCL" Subject: RE: YAK Message-ID: <961030170904_100317.2656_JHF45-1@CompuServe.COM> Alexandros, contact Angelo Vedani who developed the program YAK: Dr. Angelo Vedani SIAT Biographics Laboratory Missionsstr. 60 CH-4055 Basel Schweiz Tel: ++41-61-321-0533 Fax: ++41-61-321-0540 e-mail: siat@dial.eunet.ch regards Herbert -- Dr. Herbert Koeppen, Boehringer Ingelheim KG, Medicinal Chemistry Department -- -- D-55216 Ingelheim, Germany -- phone: ++49-6132-776845 -- fax: ++49-6132-773418 -- e-mail: 100317.2656@compuserve.com From ccl@www.ccl.net Wed Oct 30 13:17:38 1996 Received: from bedrock.ccl.net for ccl@www.ccl.net by www.ccl.net (8.8.2/950822.1) id MAA06157; Wed, 30 Oct 1996 12:29:41 -0500 (EST) Received: from sapphire.tau.ac.il for becker@sapphire.tau.ac.il by bedrock.ccl.net (8.8.2/950822.1) id MAA06515; Wed, 30 Oct 1996 12:29:35 -0500 (EST) Received: by sapphire.tau.ac.il; id AA19894; Wed, 30 Oct 1996 19:29:19 +0200 Date: Wed, 30 Oct 1996 19:29:19 +0200 (IST) From: "Oren M. Becker" To: chemistry@ccl.net, drug.dna@sci.prot.ac.uk, pdb-l@pdb.pdb.bnl.gov, bioforum@net.bio.net, biophys@net.bio.net, microbio@net.bio.net, molmodel@net.bio.net, comp-bio@net.bio.net Cc: Yechiel Becker , Oren Becker Subject: Computation in Molecular Biology Workshop, 2-11 Feb 1997 Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII --------------------------------------------------------------------- UNESCO - Hebrew University of Jerusalem International School for Molecular Biology and Microbiology (ISMBM) --------------------------------------------------------------------- International Workshop on Computation in Molecular Biology 2-11 February, 1997 at The Silicon Graphics Laboratory The Hebrew University of Jerusalem Sponsored by UNESCO Paris and The Hebrew University of Jerusalem --------------------------------------------------------------------- Aims: Computation in Molecular Biology at the Silicon Graphics Lab. The course is designed for graduate students and researchers in molecular biology who are interested in studying and practicing computational methods. Number of Participants: 20 Scientific Program: Sequence analysis, modeling of biomolecules and molecular graphics, advanced topics in computational molecular biology. Teaching Staff: H. Margalit (HUJ, Jerusalem) O. M. Becker (Tel Aviv U) M. Berman (Silicon Graphics Biomedical, Ltd) C. Sander (EMBL, Heidelberg) R. Elber (HUJ, Jerusalem) A. Goldblum (HUJ, Jerusalem) Housing: All participants will be housed in a hotel (at a reduced rate) near the University Givat Ram Campus (walking distance), two participants per room. Single room should be requested. Fellowships: A limited number of fellowships to cover part of the expenses will be available. (Participants are advised to request travel fellowships from national or international scientific organizations). Registration: Please send your application by fax: name, address, fax number, telephone number, e-mail address, information on your research studies (PhD students should attach a letter of recommendation from instructor) to: FAX: 972-2-6784010 ("Computation in Molecular Biology") Registration deadline: 1 December, 1996 Acceptance letters by: 1 January, 1997 Prof. Yechiel Becker Director UNESCO-HUJ ISMBM e-mail: becker@md2.huji.ac.il --------------------------------------------------------------------- From ccl@www.ccl.net Wed Oct 30 15:17:39 1996 Received: from bedrock.ccl.net for ccl@www.ccl.net by www.ccl.net (8.8.2/950822.1) id OAA06756; Wed, 30 Oct 1996 14:25:11 -0500 (EST) Received: from cliff.bms.com for david@viper.bms.com by bedrock.ccl.net (8.8.2/950822.1) id OAA08875; Wed, 30 Oct 1996 14:24:55 -0500 (EST) Received: from viper (viper.wf.bms.com) by cliff.bms.com (PMDF V5.0-7 #15142) id <01IB9BPEBEGG00O7R7@cliff.bms.com> for chemistry@ccl.net; Wed, 30 Oct 1996 15:24:33 -0500 (EST) Received: by viper (950413.SGI.8.6.12/930416.SGI) for chemistry@ccl.net id OAA28579; Wed, 30 Oct 1996 14:23:36 -0500 Date: Wed, 30 Oct 1996 14:23:36 -0500 From: "David R. Langley" Subject: Counterion Distribution To: chemistry@ccl.net Message-id: <9610301423.ZM28577@viper.wf.bms.com> MIME-version: 1.0 X-Mailer: Z-Mail (3.2.1 6apr95 MediaMail) Content-type: text/plain; charset=us-ascii Content-transfer-encoding: 7BIT Dear All, I am looking for a program or CHARMM script to analize the most populated counterion positions around a biomolecule from a MD simulation. In (M.A.L. Eriksson, T. Hard and L. Nilsson, Biophysical Journal, 1995, 68, 402-426) a 3-d histogram density map was constructed from an analysis of the MD run. Is this program avalible? David R. Langley -- David R. Langley -- David R. Langley