From owner-chemistry@ccl.net Thu Jul 13 01:08:48 1995 Received: from jake.chem.unsw.EDU.AU for michaels@jake.chem.unsw.EDU.AU by www.ccl.net (8.6.10/930601.1506) id AAA17302; Thu, 13 Jul 1995 00:56:44 -0400 Received: by jake.chem.unsw.EDU.AU (931110.SGI/930416.SGI.AUTO) for chemistry@ccl.net id AA06424; Thu, 13 Jul 95 14:58:03 -0700 From: michaels@jake.chem.unsw.EDU.AU (Michael Shephard) Message-Id: <9507132158.AA06424@jake.chem.unsw.EDU.AU> Subject: Compiling Mopac93 on an SGI To: chemistry@ccl.net Date: Thu, 13 Jul 95 14:58:03 PDT X-Mailer: ELM [version 2.3 PL11] I am having difficulty in compiling Mopac93 (Release 2 obtained >from QCPE) for an SGI machine (R4400 Indigo Irix 5.2) Well, actually it compiles OK without error but the mopac.exe produced doesn't work very well at all. Is there am SGI version of the source, or is the sun source supplied the "generic unix version", or possibly there is an SGI version of the makefile? Any assistance appreciated Michael -- _________________________________________________________________________ | Michael Shephard. | M.Shephard@unsw.EDU.AU | | University of New South Wales. | michaels@jake.chem.unsw.EDU.AU | | Australia | | ,p ,p a I am having difficulty in compiling Mopac93 (Release 2 obtained >from QCPE) for an SGI machine (R4400 Indigo Irix 5.2) Well, actually it compiles OK without error but the mopac.exe produced doesn't work very well at all. Is there am SGI version of the source, or is the Sun source on the tape the "generic unix version", or possibly there is an SGI version of the makefile? Any assistance appreciated Michael -- _________________________________________________________________________ | Michael Shephard. | M.Shephard@unsw.EDU.AU | | University of New South Wales. | michaels@jake.chem.unsw.EDU.AU | | Australia | | From owner-chemistry@ccl.net Thu Jul 13 02:08:53 1995 Received: from li4d for tran@limod1.ludwig.edu.au by www.ccl.net (8.6.10/930601.1506) id BAA17902; Thu, 13 Jul 1995 01:57:51 -0400 Received: from limod1.ludwig.edu.au by li4d via ESMTP (940816.SGI.8.6.9/930416.SGI) for <@li4d.ludwig.edu.au:chemistry@ccl.net> id PAA19540; Thu, 13 Jul 1995 15:57:23 +1000 Received: by limod1.ludwig.edu.au (940816.SGI.8.6.9/920502.SGI.AUTO) id PAA02563; Thu, 13 Jul 1995 15:57:10 +1000 From: "Tran Trung Tran" Message-Id: <9507131557.ZM2561@limod1.ludwig.edu.au> Date: Thu, 13 Jul 1995 15:57:01 -0500 X-Mailer: Z-Mail (3.2.0 26oct94 MediaMail) To: chemistry@ccl.net Subject: Experimental Delta-Delta G? Cc: tran@limod1.ludwig.edu.au Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii To whom this may concern, Authors who perform free-energy perturbation calculations sometimes compare their calculated delta-delta-G values with experimental values. However, I have had problems in finding out how the experimental delta-delta-G values are determined. Could somebody please provide references or a brief explanation as to how this is done. With Thanks -- Tran Trung Tran Ludwig Inst. For Cancer Research P.O. Box Royal Melbourne Hospital Victoria, 3050 Australia Tel. (613)9347-3155 Fax. (613)9347-1938 email. tran@ludwig.edu.au From owner-chemistry@ccl.net Thu Jul 13 04:53:52 1995 Received: from rzusuntk.unizh.ch for toukie@zui.unizh.ch by www.ccl.net (8.6.10/930601.1506) id EAA19881; Thu, 13 Jul 1995 04:47:00 -0400 From: Received: from rzurs3.unizh.ch by rzusuntk.unizh.ch (4.1/SMI-4.1.9) id AA22728; Thu, 13 Jul 95 10:46:55 +0200 Received: by rzurs3.unizh.ch (AIX 3.2/UCB 5.64/4.03) id AA23564; Thu, 13 Jul 1995 10:46:55 +0200 Message-Id: <9507130846.AA23564@rzurs3.unizh.ch> Subject: Rotation time for radical To: chemistry@ccl.net Date: Thu, 13 Jul 1995 10:46:54 +0200 (MEST) X-Mailer: ELM [version 2.4 PL24 PGP2] Content-Type: text Content-Length: 528 Dear Colleagues; Does anyone have a reference to an experimental measurement of the 180- degree rotation time for a primary carbon radical of the type H | R-C. | H around the R-C bond? If so, I would be most grateful to hear from you. Thanks in advance to all responders. Sincerely, (Dr.) S. Shapiro Zuerich toukie@zui.unizh.ch From owner-chemistry@ccl.net Thu Jul 13 06:38:54 1995 Received: from vax2.sara.nl for mirko@SARA.NL by www.ccl.net (8.6.10/930601.1506) id GAA20827; Thu, 13 Jul 1995 06:26:25 -0400 Received: from horus.sara.nl by SARA.NL (PMDF V4.2-15 #2498) id <01HSTL6E6ZBKAM37MH@SARA.NL>; Thu, 13 Jul 1995 12:29:34 +0200 (MET-DST) Received: from hk-pc1.chem.uva.nl by horus.sara.nl (AIX 3.2/UCB 5.64/4.03) id AA92326; Thu, 13 Jul 1995 12:25:58 +0200 Date: Thu, 13 Jul 1995 12:25:58 +0200 From: mirko@SARA.NL (Mirko Kranenburg) Subject: pi-stacking in MOPAC? To: chemistry@ccl.net Message-id: X-Envelope-to: chemistry@ccl.net MIME-version: 1.0 Content-type: text/plain; charset="us-ascii" Content-transfer-encoding: 7BIT X-Sender: a412knsm@horus.sara.nl Dear Netters, Currently I am performing geometry optimizations on compounds in which pi-stacking has an important contribution to the overall geometry. The results from MOPAC 6 calculations (AM1, PM3 and MNDO) show no pi- stacking whatsoever, even when a X-ray structure was used as input. My question: does MOPAC take pi-stacking into account, and are there any relevant references concerning this problem. Thanks in advance, Mirko Kranenburg Universiteit van Amsterdam J.H. van 't Hoff Research Institute Dept. of Inorganic Chemistry and Homogeneous Catalysis Nieuwe Achtergracht 166 1018 WV Amsterdam The Netherlands tel. +31-20-5256417 fax. +31-20-5256456 e-mail: mirko@sara.nl From owner-chemistry@ccl.net Thu Jul 13 08:54:09 1995 Received: from saturn.kent.edu for gardner@saturn.kent.edu by www.ccl.net (8.6.10/930601.1506) id IAA22440; Thu, 13 Jul 1995 08:51:11 -0400 Received: by saturn.kent.edu (AIX 3.2/UCB 5.64/4.03) id AA12910; Thu, 13 Jul 1995 08:40:10 -0400 Date: Thu, 13 Jul 1995 08:40:10 -0400 (EDT) From: Leon Gardner To: chemistry@ccl.net Subject: coordinates of atoms in carbohydrates. Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII I am wondering if anyone knows where I can find detailed information on the Cartesian coordinates of atoms in carbohydrates, or in the monosaccharides that compose them. If such information is not available, is there a computer program than can calculate these coordinates.? Thanks for your help. Leon Gardner gardner@saturn.kent.edu From owner-chemistry@ccl.net Thu Jul 13 09:38:56 1995 Received: from ns.trans.net for kmay@trans.net by www.ccl.net (8.6.10/930601.1506) id JAA23137; Thu, 13 Jul 1995 09:27:17 -0400 Received: from by ns.trans.net (4.1/SMI-4.1) id AB12426; Thu, 13 Jul 95 15:26:54 BST Message-Id: <9507131426.AB12426@ns.trans.net> X-Sender: kmay@mailhost Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 13 Jul 1995 15:28:53 +0200 To: CHEMISTRY@ccl.net From: kmay@trans.net (Klaus J.W.May) Subject: French atomic tests. Let's disagree ( who disagrees) > #### ##### #### ##### > # # # # # # > #### # # # # # > # # # # ##### > # # # # # # > #### # #### # > > # # # # #### # ###### ## ##### > ## # # # # # # # # # # # > # # # # # # # ##### # # # # > # # # # # # # # ###### ##### > # ## # # # # # # # # # # > # # #### #### ###### ###### # # # # > > ##### ###### #### ##### #### # > # # # # # # > # ##### #### # #### # > # # # # # > # # # # # # # # > # ###### #### # #### # > > > > > 1 SHIMIZU Seishi Physics,University of Tokyo,Japan > 2 Yuichi Nishihara Physics,University of Tokyo,Japan > 3 Hirohisa TANIGUCHI Physics,University of Tokyo,Japan > 4 Takashi Tomoeda Physics,University of Tokyo,Japan > 5 Tomoki KOBAYASHI Physics,University of Tokyo,Japan > 6 Munehito ARAI Physics,University of Tokyo,Japan > 7 Akira Okazaki Physics,University of Tokyo,Japan > 8 Atsushi Matsumura Physics, Tohoku University, Japan > 9 Kouta Yamamoto Chemistry,Tohoku University,Japan > 10 Yasushi UJIOKA Degremont S.A., France > 11 Toru Hara Universite de Paris Sud, France > 12 Rene Bakker CEA - Sacley, France > 13 David Garzella Universite de Paris Sud, France > 14 Henk Blok Vrije Universiteit/NIKHEF, Amsterdam > 15 Igor Passchier NIKHEF, Amsterdam > 16 Ard van Sighem NIKHEF, Amsterdam > 17 Johan Noordhoek KOL Leiden > 18 C.M.C.M. van Woerkens Kamerlingh Onnes Laboratory, Leiden > 19 Annemarie Borst, Vrije Universiteit Amsterdam > 20 Gijs Nelemans Universiteit Utrecht > 21 Susanne Buiter Universiteit Utrecht > 22 Yvo Kok Paleomagnetic Lab., Utrecht > 23 Thom Pick Paleomagnetic Lab., Utrecht University > 24 Dagmar Olbertz Universiteit Utrecht > 25 Eleonore Stutzmann Institut de Physique du Globe de Paris, France > 26 Nicole Girardin Institut de Physique du Globe de Paris, France > 27 Francois Girardin Ecole Nat. Sup. des Telecommunications,France > 28 Axel Manthey Comnets RWTHJ Aachen, Germany > 29 Frank Brockners LfBS RWTH Aachen, Germany > 30 Fridtjof Lange-Rehberg LfBS RWTH Aachen, Germany > 31 Roman Breuer RZ, RWTH Aachen, Germany > 32 Bernd Greuel Universitaet zu Koeln, Germany > 33 Christoph Gaukel KFA Juelich, Germany > 34 Benedikt Engels KFA Juelich, Germany > 35 Peter Richard KFA Juelich, Germany > 36 Stefan Bluegel Forschungszentrum Juelich, Germany > 37 Doris Vogtenhuber Universitaet Wien, Austria > 38 Hellfried Schreiber Inst.Theoret.Chemie, Austria > 39 Bianca Habermann IMP Vienna, Austria > ********** > > Dear Sirs, > > This is a chain letter to urge the french > government to stop nuclear tests. > If you agree with us, please add your name to the list above, > and send copies to your freinds. > We will add up the lists that had come back to us, and send it > to the French Government. > > If you happen to be the hundredth,two hundredth, three hundredth, > and so on, on the list, please send a copy of the mail back to the > addresses below, so that we can keep track of this project. If you have > any comment please send mails to us. And also, > if you are multi-lingual and have friends who may not understand > English, please translate this message and add it to the end of the mail. > Thank you very much. > > > ******* addresses of the organizers > shimizu@femto.phys.s.u-tokyo.ac.jp > keshi@uticeaix1.icepp.s.u-tokyo.ac.jp <- please use this adress > ******* > > > Klaus J.W. May e-mail: kmay@trans.net May & Partner SciTech Consult Phone/Fax: xx49-(89) 333 569 Mandlstr. 15 Mobile national: 0171-80 80 900 international: xx49-1712-80 80 900 D - 80802 Munich GERMANY From owner-chemistry@ccl.net Thu Jul 13 09:53:56 1995 Received: from goodyear.com for eofstead@goodyear.com by www.ccl.net (8.6.10/930601.1506) id JAA23737; Thu, 13 Jul 1995 09:49:41 -0400 Received: from rdsrv2 (rdsrv2.goodyear.com) by goodyear.com (4.1/SMI-4.1) id AA10731; Thu, 13 Jul 95 09:49:40 EDT Received: from [163.243.12.103] (rds991) by rdsrv2 (4.1/SMI-4.1) id AA10844; Thu, 13 Jul 95 09:49:37 EDT Message-Id: <9507131349.AA10844@rdsrv2> X-Sender: t410b16@rdsrv2 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 13 Jul 1995 09:49:17 -0500 To: chemistry@ccl.net From: eofstead@goodyear.com (Eilert Ofstead) Subject: Summary: criteria for agostic interactions Cc: eofstead@goodyear.com Dear netters; Recently I posted the following question: >Are there criteria which define/verify the presence of an agostic H---metal >interaction? > >I am mainly interested in organometallics where the H is not on C-alpha. I have enclosed the replies to this question at the end of this note. My thanks to all who responded! I was mainly interested in looking at this issue for structures generated in modeling studies, where experimental data are not available. Well-defined agostic interactions appear to be easily recognized from a simple examination of the geometry of the structure in most cases. Unfortunately, I suspect that I am dealing with systems where this effect is very marginal, yet may be important. There's the rub. How do you quantify these weak interactions? I also came across a method which is based on second-order perturbation energy estimates (Kaufmann et al, Organometallics, v7, 1597, 1988). This method seems to allow one to obtain quantitative estimates of these non-bonded interactions. Any comments on this technique? More importantly, can this method be invoked using Spartan software, for example? (forgive my novice state of ignorance) REPLIES ===================================== > >there are several diagnostic spectroscopic criteria: > >1) 13C-NMR: Low C-H coupling constant may be for instance lowered to >70 Hz for a sp2-C-H carbon atom > >2) IR: C-H stretching frequency in the range of 2600-2800 cm-1 > >also: selective Deuteration of the particular C-H bond helps to establish >2) and also allows to distinguish between an agostic interaction and an >equilibrium between the C-H elimimation product (gives a metal hydride and >an olefin in the case of a beta-hydrogen atom) and the metal alkyl complex >(equilibrium isotope pertubation). For details of the latter check >the following review by M. Brookhart, M.L.H. Green, Progress on Inorganic >Chemistry (1988?) - I don't have the exact reference handy but I can find >it for you if necessary. > >Further methods are: > >X-ray and Neutron diffraction: Grounds: M-H distance and M-C-H (alpha) and >C-C-H (beta) angle arguments. Care should be taken however in the >interpretation of X-ray crystal structures due to the inherent problems of >this method to locate the "true" proton position. > >Alpha vs Beta vs .. There are quite a number of examples for both - lately >even also Gamma-H agostic interactions have been discussed in the context >of early transition metal Ziegler-Natta polymerizations with metallocene >based catalysts. Also alpha (evidenced) and beta agostic interactions may >play a role in this catalytic system. >You'll find a pretty good discussion on agostic interactions in >Bob Crabtree's monograph: The Organometallic Chemistry of the Transition >Metals, 2. ed, Wiley Interscience, (1994/3). Hope this helps, let me know >if I can be of further assistance > >Peter >_________________________________ >Peter Burger >Anorg.-Chem. Institut >Universitaet Zuerich >Winterthurerstr. 190 >8057 Zuerich >Switzerland >PHONE:+41 1 257 4692 and 4677 (work) > +41 1 252 1793 (home) >FAX: +41 1 364 0191 >E-mail:chburger@aci.unizh.ch ********************************************************* > >A few years ago, we described (Mealli and Proserpio, Comments Inorganic >Chemistry, 9, 37, 1989) how the agostic interactions can play a role for the >stability of certain Re3 clusters. The arguments were based on some careful >considerations of the stereochemistry of the non-hydrogen substituents >at the alkyl groups suspected to give rise to the agostic interactions. >Further support to the hypothesis came from the simplest MO theory arguments >which allowed us to see some clear-cut differences between models with and >without the agostic interactions at work. >You may wish to take a look at that paper. >Best regards, >Dr. Carlo Mealli >ISSECC, CNR >Via J.NARDI 39, 50132 Firenze, ITALY >e-mail: mealli@cacao.issecc.fi.cnr.it >Tel. 0039-55-2346653 (243990) - Fax 2478366 - Home 4492709 ********************************************************* > > We have looked at agostic metal complexes and have tried several >approaches to 'quantifying' agostic interactions. > One simple approach is to look at the CH bond length versus >non-agostic models. This is often a tough comparison since the CH bonds >lengthen very little when participating in an agostic interaction, which seems >to be consistent with exptl evidence from neutron diffraction. Likewise, if >you can calc the IR spectrum you should see a lowering of the CH stretching >freq for an agostic CH bond. With GAMESS we can decompose the IR spectrum into >normal mode contributions so you can look at the force constant of the agostic >CH bond and compare it to non-agostic CH bonds. > Other things to look for are bond angles involving agostic CH bonds >which deviate from 'normal' (i.e., VSEPPR) values. It is not clear if this >will be as evident when looking at carbons other than C-alpha. > You should also be able to plot the MOs and see if there is a >multi-centered interaction among the agostic MHC unit. Similarly, plotting >the total density in the MHC plane may yield some clues. > Finally, in our work we have used a metric parameter defined by >Crabtree >and co-workers in the following paper. > > RH Crabtree et al. Inorg. Chem. 1985, 24, 1986. > >Essentially, you need the calcd MH, MC, and CH distances, which you can plug >into the equation given for rbp. As rbp decreases, agostic intrxn should >increase. There are some exptlly determined systems to provide calibration. > None of these approaches are unambiguous, but together they should >provide some insight into whether or not a substantial agostic interaction >is present. > >Sincerely, > >Tom Cundari >+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+== >= >Thomas R. Cundari Address from 3/15/95 -> ca. August: >Asst. Professor of Chemistry Prof. Tom Cundari >Computational Inorganic Chemistry Lab Visiting Scientist >University of Memphis Cornell Theory Center >Memphis, TN 38152 520 Engineering & Theory Center >Bldg >phone: 901-678-2629 Cornell University >fax: 901-678-3447 Ithaca, New York 14853-3801 >e-mail: cundarit@cc.memphis.edu phone: (607)254-8686 > office: (607)254-8621 >http://www.memst.edu/chemistry/umchem.html fax: (607)254-8888 > e-mail: same as always >+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+== ********************************************************* > Reply to: RE>CCL:verification of agostic inter >This would be an obvious application of the "Theory of >Atoms in Molecules" (AIM), by Richard Bader and coworkers. > >See the book, or the large literature on the subject. You >need an 'ab initio' waverfunction and the AIM programs. > >Richard Bone >__________________________________________________________ > >Richard G. A. Bone, PhD. >Computational Chemist >Terrapin Technologies, Inc. >South San Francisco >USA > >E-mail rgab@trpntech.com ********************************************************* >1/r is probably pretty good but still bad compared with explicit >water. > >arne > >-------------------------------------------------------- > From: Arne Elofsson > Email: arne@hodgkin.mbi.ucla.edu > WWW: http://www.doe-mbi.ucla.edu/arne/main.html ********************************************************* Once again, thanks all. - --------------------------------------------------------------------- Eilert Ofstead, Sr R&D Assoc Research Division e-mail: eofstead@goodyear.com The Goodyear Tire & Rubber Co voice: 216-796-3983 142 Goodyear Blvd fax: 216-796-3304 Akron, OH 44305 USA _____________________________________________________________________ From owner-chemistry@ccl.net Thu Jul 13 12:24:00 1995 Received: from alchemy.chem.utoronto.ca for elewars@alchemy.chem.utoronto.ca by www.ccl.net (8.6.10/930601.1506) id MAA27829; Thu, 13 Jul 1995 12:18:16 -0400 Received: (from elewars@localhost) by alchemy.chem.utoronto.ca (8.6.10/8.6.10) id MAA12031 for chemistry@ccl.net; Thu, 13 Jul 1995 12:18:15 -0400 Date: Thu, 13 Jul 1995 12:18:15 -0400 From: "E. Lewars" Message-Id: <199507131618.MAA12031@alchemy.chem.utoronto.ca> To: chemistry@ccl.net Subject: 3-21G(*) BASIS Re the recent query about the 3-21G(*) basis set: This is discussed in the ab initio bible (Hehre, Radom, Schleyer, Pople) pp 84-86; see also several of the Tables in this book. On p 86: "...the 3-21G(*) basis set should find considerable use as an alternative to 6-31G* as a means of providing structures and normal-mode frequencies that closely approximate those of the larger basis representation." In Hehre's *Practical Strategies for Electronic Structure Calculations*, on e.g. pp 54-69, it is shown that the 3-21G(*) basis gives relative energies extraordinarily close tho those of the 6-31G* basis (look at compounds with the Na--Cl elements, for which the * is relevant). maybe the point is not whether it's "unbalanced", but whether it works (as far as d-orbital participation goes, Nature is unbalanced). Errol Lewars ==== From owner-chemistry@ccl.net Thu Jul 13 12:39:11 1995 Received: from ns.trans.net for kmay@trans.net by www.ccl.net (8.6.10/930601.1506) id MAA28013; Thu, 13 Jul 1995 12:28:54 -0400 Received: from [194.59.178.14] by ns.trans.net (4.1/SMI-4.1) id AA12936; Thu, 13 Jul 95 18:28:38 BST Message-Id: <9507131728.AA12936@ns.trans.net> X-Sender: kmay@mailhost Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 13 Jul 1995 18:30:37 +0200 To: CHEMISTRY@ccl.net From: kmay@trans.net (Klaus J.W.May) Subject: Chainletter about French Atom Testing Dear netters, my sincerest apologies to everybody on the list. I was not aware to commit such a violation of Netetiquette. And the least thing I intended was to hurt the intentions of this letter. This was the first chain letter I ever came across . I am personally very moved by the French decision. I worked in research of enrichment of fallout (Strontium 90) in human bones several years ago. We still collect debris from the 50ies! Todays science knows better than then. And I thought chemists have quite some insight into the matter. Therefore CCL. However, I've got overwhelming support from the list so far - and justifyed criticism. I'll never do it again! And please, help me! Do not post a string of messages discussing this issue! Let this be the last bandwith-waster on this list. Again PLEASE do not discuss this issue on the list - do not punish all CCLers ! It's been my fault - send flames to me personally!! Cheers Klaus :,-( From owner-chemistry@ccl.net Thu Jul 13 13:08:59 1995 Received: from danpost.uni-c.dk for peon@medchem.dfh.dk by www.ccl.net (8.6.10/930601.1506) id NAA29079; Thu, 13 Jul 1995 13:05:10 -0400 Received: from medchem.dfh.dk (medchem.dfh.dk [130.225.177.15]) by danpost.uni-c.dk (8.6.4/8.6) with ESMTP id TAA01573 for <@danpost.uni-c.dk:chemistry@ccl.net>; Thu, 13 Jul 1995 19:05:07 +0200 Received: from [130.225.177.59] by medchem.dfh.dk via SMTP (940816.SGI.8.6.9/940406.SGI) for id TAA19832; Thu, 13 Jul 1995 19:26:10 +0200 Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 13 Jul 1995 19:05:07 +0200 To: chemistry@ccl.net From: peon@medchem.dfh.dk (Per-Ola Norrby) Subject: Re: CCL:3-21G(*) BASIS >Re the recent query about the 3-21G(*) basis set: I just want to add my 5 cents worth about a very common source of confusion: The original question specified 3-21G*, but it was clear from the discussion that 3-21G(*) was implied. Note that 3-21G(*) only adds d-orbitals to atoms that are believed to "need" them for regular bonds in easily observed compounds, that is, elements heavier than Ne. In most programs it's also possible to specify the less economical set 3-21G*, which uses polarization on all non-hydrogen atoms. Per-Ola Norrby * Per-Ola Norrby * The Royal Danish School of Pharmacy, Dept. of Med. Chem. * Universitetsparken 2, DK 2100 Copenhagen, Denmark * tel. +45-35376777-506, +45-35370850 fax +45-35372209 * Internet: peon@medchem.dfh.dk, peo@compchem.dfh.dk From owner-chemistry@ccl.net Thu Jul 13 16:39:02 1995 Received: from synthesis.chem.arizona.edu for matt@synthesis.chem.arizona.edu by www.ccl.net (8.6.10/930601.1506) id QAA03693; Thu, 13 Jul 1995 16:27:22 -0400 Received: by synthesis.chem.arizona.edu (5.65/DEC-Ultrix/4.3) id AA06976; Thu, 13 Jul 1995 13:24:12 -0700 Date: Thu, 13 Jul 1995 13:24:12 -0700 (MST) From: Matt Stahl To: chemistry@ccl.net Subject: validity of constrained calcs. Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Netters, I am working on a project which involves parameterizing molecular mechanics torsion potentials. We used gaussian to generate energy vs. torsion data by performing constrained optimizations. One torsion angle was held constant while the rest of the coordinates in the molecule were optimized. Someone recently made the comment that constrained optimizations with gaussian are not appropriate for such a study because the unconstrained coordinates do not fully optimize. This suggestion seems odd because the literature seems to be repleat with examples of isolating the energetics of a molecular motions (distances, angles, and torsions) through constrained optimization. How far off the mark am i in thinking that constrained opts. are appropriate for this type of work. The alternative method would be to perform an IRC (internal reaction coordinate) calculation to characterize the path between the two minima of interest, but as i understand it IRC calculations are dependant on Linear Synchronous Transit calculations which ESTIMATE the transition state. Can one perform an IRC from a calculated transition state (opt=ts)? If not, how much more "valid" is an IRC calculation than constrained optimization if at any point the IRC has to "guess" the path? thanks, matt ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Mr. Matthew Stahl | Department of Chemistry | I'd rather be right half of the University of Arizona | time than half right all the Tucson, AZ 85721 | time. matt@synthesis.chem.arizona.edu | -Malcom S. Forbes ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ From owner-chemistry@ccl.net Thu Jul 13 18:39:05 1995 Received: from hodgkin.mbi.ucla.edu for arne@hodgkin.mbi.ucla.edu by www.ccl.net (8.6.10/930601.1506) id SAA06002; Thu, 13 Jul 1995 18:36:04 -0400 Received: by hodgkin.mbi.ucla.edu; id AA00954; Thu, 13 Jul 1995 15:44:07 -0700 Date: Thu, 13 Jul 1995 15:44:07 -0700 From: Arne Elofsson Message-Id: <9507132244.AA00954@hodgkin.mbi.ucla.edu> To: CHEMISTRY@ccl.net Subject: Summary of NT vs UNIX question. thanxs for all the responses I'll include the summaries at the end and just summarize (and comment) the summaries (and comments) here: It seems like most people did not think it was feasible to run a comp.chem lab based on NT. Mainly because of the lack of all tools that you get for free with unix. I can support that, though I haven't seen NT in a while. But I have seen that there are commercial (and free) products that you can run on NT, bith csh and ksh is supported and most unix commands like awk,grep sed etc exists. And the compiler environment (especially for C/C++) are excellent. much better than on most unix boxes. Most people recomended me to bye PC:s instead and run linux (or Next). My fear for that is that even if the price performance of a PC is unbeatable it is still pretty far behind the best workstations in floating point capacity. Both for NEXT and linux is the lack of a god fortran compiler a problem too. I guess that solaris for intel has a fortran compiler, but I haven't seen any numbers for it. arne PS. These are the approximately prices that I'm talking about All configurations for a descent workstation (64 mb mem, 1mb cache 1gb harddrive, descent monitor, scsi ...) Very very approximative prices (partly my guesses) Name Price SpecFp Specint $/fp PC-linux-pentium-133 $6000 110 140 55 PC-P6-180Mhz ('96) $6000 230? 260? 27 Clone-Dec-alpha-21164 $19000 550 340 35 Clone-Dec-alpha-21064a $12500 290 200 43 IBM-PowerPc $8000 155 150 51 Dec-alpha-pc-21064 $14000 290 200 48 so the PC might look OK, but the price performance is not that staggering, I probably would go with it if I could get a good fortran compiler. (aybee I can talk Intel into port their compiler to linux (: arne -------------------------------------------------------- From: Arne Elofsson Email: arne@hodgkin.mbi.ucla.edu WWW: http://www.doe-mbi.ucla.edu/arne/main.html From: wallyr@netcom.netcom.com (Walter E. Reiher III) Subject: Re: CCL:Windows NT for computational chemistry ? Date: Tue, 11 Jul 1995 10:09:38 -0700 Return-Path: wallyr@netcom.netcom.com Received: from p90 by netcomsv.netcom.com with SMTP (8.6.12/SMI-4.1) id KAA04839; Tue, 11 Jul 1995 10:09:38 -0700 Message-Id: <199507111709.KAA04839@netcomsv.netcom.com> X-Sender: wallyr@netcom.com X-Mailer: Windows Eudora Version 2.0.3 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: Arne Elofsson I've been using Windows NT as my primary platform for over a year now, but on an Intel platform, not an Alpha. While I'm an enthusiastic supporter of NT over DOS/Windows, I'm not convinced that it's worthwhile for the application you described. In my opinion, it's not yet a replacement for UNIX in our world (computational chemistry). >1. I'm working on protein folding and some molecular > dynamics. Mainly using my own programs and Charmm. > And the normal tools people use on a workstation Your mileage may vary, but have you actually looked at a FORTRAN compiler for an Alpha NT workstation? CHARMM is in FORTRAN. Is it supported under NT? When I last worked with CHARMM, it had a lot of memory allocation stuff in it that MAY prove difficult to port to a different OS. My primary gripe with NT is with FORTRAN compilers: while excellent development environments exist under NT for C/C++, FORTRAN is a really poor cousin. I've had a great deal of difficulty porting very standard, portable comp chem apps from UNIX to NT and it's because of the compilers. Things might be better on an Alpha if you can get a FORTRAN from DEC. >2. I'm starting a new lab in the fall, and will have > to buy computers for me (and eventual gradstudents) > I will keep on developing software as well as running > programs. (CPU speed is my main goal) Don't forget that time spent running programs is generally much less than time spent writing them, and raw CPU speed alone isn't a good metric. >3. I've been looking around and at the moment it seems as > if the best price/performance ratio is for some clone > makers building machines based on dec-alpha chips, > and maybee some other chips too. The problem is that > most of these systems do not run any unix but only > windows NT. For other systems the UNIX is much more > expensive than Win-NT (for instance IBM:s PowerPC). > Therefore I'm considering using NT. (Even if I am > a UNIX freak.) Look at the apps available under NT versus UNIX: just be aware that you're going to have to "blaze a trail" with NT. I sure wish I could take advantage of neat UNIX tools like PVM. >4. Linux would be an alternative but the version for the > alpha is not out yet. (Last time I looked it was running > but networking and X wasn't up yet) > >My questions: > >1. Is any comp-chemistry groups using NT at all ? > If so what are your experiences ? >2. Does anyone know about any molecular graphic > package that runs under NT. (Other than rasmol) The only "serious" package I'm aware of under NT is HyperChem. The group I'm working with has developed its own graphics package, which can't be distributed. >3. Anyone have had any ecperience with NT for serious > computations, (i.e. just running jobs that take a > long time and a lot of memory). Is there any benchmarks > that compare NT vs Unix. (Probably mainly effectiveness > of compilers). >4. What tools do people use under NT ? I recently started using Microsoft Visual C++ 2.0 (runs under NT only). It's a very nice development environment for C. >5. Am I completely stupid to consider this ? >6. Will I really save money (when I sum up the cost of > compilers and tools compared to unixlicenses etc) I know grad student and postdoc time is free :-)! Seriously, time should be a factor here. Be careful about the "economy" of clone computers. Again, things might be better in Alpha land, but I believe much more money is wasted on the time taken to try to make "cheap" hardware work than was saved on buying it in the first place. Good luck! Please summarize the responses. Wally ======================================================================== Walter E. Reiher III, Ph.D. WallyR@netcom.com Consultant in Computational Chemistry P.O. Box 61056 voice 408-720-0240 Sunnyvale, CA 94088 fax 408-720-0378 From: jaw@camsci.com (Joel A. Wolff) Subject: Re: CCL:Windows NT for computational chemistry ? Date: Tue, 11 Jul 1995 09:03:07 -0400 Return-Path: jaw@camsci.com Received: from [149.53.1.43] (technetium.camsci.com) by camsci.com (4.1/3.1.090690-Cambridge Scientific Computing) id AA25323; Tue, 11 Jul 95 09:02:19 EDT X-Sender: jaw@helium.camsci.com Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: Arne Elofsson > >My questions: > >2. Does anyone know about any molecular graphic > package that runs under NT. (Other than rasmol) FYI, both CS ChemDraw (2D) and CS Chem3D (3D) run under Windows NT. At the moment Chem3D only supports MM2, but we are working on adding better force fields. Joel A. Wolff, Ph.D. Customer Service Manager CambridgeSoft Corporation 875 Massachusetts Avenue Cambridge, MA 02139 Phone: (800) 315-7300 or (617) 491-2200 Fax: (617) 491-7203 Compuserve: 76070,615 Internet: info@camsci.com http://www.camsci.com From: Jussi Eloranta Subject: Re: CCL:Windows NT for computational chemistry ? Date: Tue, 11 Jul 1995 08:04:14 +0300 (EET DST) Return-Path: eloranta@voimax.voima.jkl.fi Received: (from eloranta@localhost) by voimax.voima.jkl.fi (8.6.10/8.6.9) id IAA06954 for arne@hodgkin.mbi.ucla.edu; Tue, 11 Jul 1995 08:04:14 +0300 Message-Id: <199507110504.IAA06954@voimax.voima.jkl.fi> To: arne@hodgkin.mbi.ucla.edu (Arne Elofsson) In-Reply-To: <9507110027.AA05550@hodgkin.mbi.ucla.edu> from "Arne Elofsson" at Jul 10, 95 05:27:50 pm X-Mailer: ELM [version 2.4 PL23] Content-Type: text Content-Length: 674 >4. Linux would be an alternative but the version for the > alpha is not out yet. (Last time I looked it was running > but networking and X wasn't up yet) The distribution kit for aplha is out. I recall that you can download it from gatekeeper.dec.com (?). The name was something like *blade* . I don't know how stable system this is.... >My questions: > >1. Is any comp-chemistry groups using NT at all ? > If so what are your experiences ? I haven't even seen one machine running NT. Even on intel platform.... regards, Jussi ps. I run mostly linux on P100 systems. I hear that P150 will be out soon... I don't know then how it compares to the aplha systems. From: Thomas Nhan Subject: Re: CCL:Windows NT for computational chemistry ? Date: Mon, 10 Jul 1995 19:09:56 +48000 Return-Path: tom@cassandra.chem.washington.edu Received: by cassandra.chem.washington.edu (920110.SGI/910805.SGI) for arne@hodgkin.mbi.ucla.edu id AA07162; Mon, 10 Jul 95 19:09:59 -0700 Reply-To: Thomas Nhan To: Arne Elofsson In-Reply-To: <9507110027.AA05550@hodgkin.mbi.ucla.edu> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII On Mon, 10 Jul 1995, Arne Elofsson wrote: > Sorry for wasting the bandwidth but I'd really hear > about peoples experiences using windows NT for serious > computational chemistry work. Well, I just want to state in advance that I have 3 buddies working @ Microsoft WinNT project and they don't have windows installed on their system. That should get one start thinking why :) > So my story is: > 3. I've been looking around and at the moment it seems as > if the best price/performance ratio is for some clone > makers building machines based on dec-alpha chips, > and maybee some other chips too. The problem is that > most of these systems do not run any unix but only > windows NT. For other systems the UNIX is much more > expensive than Win-NT (for instance IBM:s PowerPC). > Therefore I'm considering using NT. (Even if I am > a UNIX freak.) Not at all. DEC stations may seem really fast to you because it has 2Mb internal cache with a RISC chip while Intel's chip has a measly 256 kb internal cache with 16 Mb of memory, and it's probably under 64 kb DOS constraint. So if you want a fast machine, try getting more memory (>64 Mb) and a large internal cache with EDO RAM. With the same amount of money, you can get 5-6 P5-120 (or P6-166 when you get it). You can run circle around that single DEC! > 4. Linux would be an alternative but the version for the > alpha is not out yet. (Last time I looked it was running > but networking and X wasn't up yet) Well, Linux has changed quite abit since you last look. It is more popular than ever. With _free_ accelerated graphics for X and _all_ the commercial unix programs are now available (e.g. Mathematica, Matlab, you choice of editors and emulators for windows/mac/...) Even Intel decided to help out by providing optimizations for their Pentium to the free GCC and there is a free fortran compiler too (not Fortran->C translator) Besides the simple X and compiler, there are free implementations of practically everything now; they may not be 100% complete, but they are useable (e.g. LessTif for Motif, GNUStep for NeXTStep, MESA for OpenGL). > My questions: > 1. Is any comp-chemistry groups using NT at all ? > If so what are your experiences ? Not that I know of. People I know either use the high-end SGI for graphics, or low end pentium for number crunching. > 2. Does anyone know about any molecular graphic > package that runs under NT. (Other than rasmol) Nope. But there are many for X and GL. > 3. Anyone have had any ecperience with NT for serious > computations, (i.e. just running jobs that take a > long time and a lot of memory). Is there any benchmarks > that compare NT vs Unix. (Probably mainly effectiveness > of compilers). Well, then you probably prefer your compiler to be tested and used by millions daily. GCC used to generate slow code for Pentium because Intel did not provide support for optimization codes. But now, that should not hold anyone back from getting a cheap Intel or other PC-compatible. > 4. What tools do people use under NT ? > 5. Am I completely stupid to consider this ? Well,... > 6. Will I really save money (when I sum up the cost of > compilers and tools compared to unixlicenses etc) Yes. Even if you think Linux is fluff and go with NeXTStep, which will cost you about an additional $200/machine for software. Then you will get the support, X-window for simple graphics, and a great development platform. With 5-6 Pentiums running in parallel, I doubt that even the top-of-the-line DEC can beat it. By the end of the year, we should have our Pentium farms set up and see if they can match with our colleague's single Alpha :) > Thanxs for any help You are quite welcome. Just look more carefully at other alternatives before jumping in. SGI and DEC are not the main source of graphics and speed anymore. With the advent of standard graphics protocols (Xwindow and OpenGL are free) and cheap/free unix systems, the cheaper alternative will be from machines that was meant for game machines for little kids :) PC-compatibles are much cheaper to maintain than the proprietary workstations. -- Thomas Nhan Department of Chemistry University of Washington From: hughc@extro.ucc.su.OZ.AU (Hugh Capper) Subject: Re: CCL:Windows NT for computational chemistry ? Date: Tue, 11 Jul 1995 11:12:09 +1100 Return-Path: hughc@extro.ucc.su.oz.au Received: from hugh (hugh.pharmacol.su.OZ.AU [129.78.41.50]) by extra.ucc.su.OZ.AU (8.6.10/8.6.10) with SMTP id LAA22204 for ; Tue, 11 Jul 1995 11:14:36 +1000 Message-Id: <199507110114.LAA22204@extra.ucc.su.OZ.AU> X-Sender: hughc@postbox.su.edu.au X-Mailer: Windows Eudora Version 1.4.4 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: Arne Elofsson Dear Arne, I will answer your questions at the bottom. I run a Molecular Graphics/ Computational Chemistry lab in the Dept of Pahrmacology at the Uni of Sydney. We have an NT based system. >Sorry for wasting the bandwidth but I'd really hear >about peoples experiences using windows NT for serious >computational chemistry work. > >So my story is: > >1. I'm working on protein folding and some molecular > dynamics. Mainly using my own programs and Charmm. > And the normal tools people use on a workstation > >2. I'm starting a new lab in the fall, and will have > to buy computers for me (and eventual gradstudents) > I will keep on developing software as well as running > programs. (CPU speed is my main goal) > >3. I've been looking around and at the moment it seems as > if the best price/performance ratio is for some clone > makers building machines based on dec-alpha chips, > and maybee some other chips too. The problem is that > most of these systems do not run any unix but only > windows NT. For other systems the UNIX is much more > expensive than Win-NT (for instance IBM:s PowerPC). > Therefore I'm considering using NT. (Even if I am > a UNIX freak.) > >4. Linux would be an alternative but the version for the > alpha is not out yet. (Last time I looked it was running > but networking and X wasn't up yet) > >My questions: > >1. Is any comp-chemistry groups using NT at all ? > If so what are your experiences ? Yes we are. >2. Does anyone know about any molecular graphic > package that runs under NT. (Other than rasmol) We are using Gaussian on the NT box. It is quite good and very stable. We have not had a crash in over 9 months. The calculations are done quite quickly. On a cost/performance basis they are unbeatable. >3. Anyone have had any ecperience with NT for serious > computations, (i.e. just running jobs that take a > long time and a lot of memory). Is there any benchmarks > that compare NT vs Unix. (Probably mainly effectiveness > of compilers). The machine runs 64 Mbytes RAM but is also running the network and printers. It has a 4 Gbyte HDD that it uses 3 Gbyte exclusively for scratch space for Gaussian. Gauusian needs HUGE scratch spaces no matter what operating system you run. >4. What tools do people use under NT ? What do you means by tools? We have all our presentation graphics, including POVRAY and powerpoint running under NT. >5. Am I completely stupid to consider this ? No, but we have found that there are not many other modelling packages that run under NT at the moment. We have Windows for Workgroups and RS-6000 running off the NT system. Lots of people are getting Windows 95 compliant packages ready at the moment. These will run on NT. >6. Will I really save money (when I sum up the cost of > compilers and tools compared to unixlicenses etc) I think you probably will save money but that is something you will have to workout. I you have any further questions please contact me. Hugh Capper Biomolecular Computer Graphics Unit Dept. of Pharmacology The University of Sydney NSW 2006 Australia Tel 61 2 351 4308 Fax 61 2 351 3868 From: case@scripps.edu (David Case) Subject: Re: CCL:Windows NT for computational chemistry ? Date: Mon, 10 Jul 95 17:49:19 PDT Return-Path: case@scripps.edu Received: from dirac.scripps.edu by scripps.edu (5.61/1.34) id AA20699; Mon, 10 Jul 95 17:52:32 -0700 Received: by dirac.Scripps.EDU (4.1/SMI-4.1) id AA05852; Mon, 10 Jul 95 17:49:19 PDT Message-Id: <9507110049.AA05852@dirac.Scripps.EDU> To: arne@hodgkin.mbi.ucla.edu > Therefore I'm considering using NT. (Even if I am > a UNIX freak.) I find implausible that any momentary advantage in "price/performance" you might find with a Win-NT would outweigh the disadvantages of leaving UNIX, given the above statement. Don't underestimate all of the free software you get with Unix (stuff that really works for computational chemistry, and for which source code is included to allow you to move it to new platforms as you buy them.) > > 4. Linux would be an alternative but the version for the > alpha is not out yet. (Last time I looked it was running > but networking and X wasn't up yet) You could try this: get a machine that has Linux now, get your lab up and running (maybe a little slower than you would like, but with all the stuff you need). Then, when the time comes to buy a second machine, see if an alpha or power-pc (or whatever) version is available at that time. ...just my $.02 ...dave case From: Arne Elofsson Subject: CCL:Windows NT for computational chemistry ? Date: Mon, 10 Jul 1995 17:27:50 -0700 Return-Path: chemistry-request@ccl.net Received: by escher.mbi.ucla.edu (5.57/Ultrix3.0-C) id AA25258; Mon, 10 Jul 95 17:30:12 -0700 Received: for chemistry-request@ccl.net by www.ccl.net (8.6.10/930601.1506) id UAA27648; Mon, 10 Jul 1995 20:25:15 -0400 Received: from hodgkin.mbi.ucla.edu for arne@hodgkin.mbi.ucla.edu by www.ccl.net (8.6.10/930601.1506) id UAA27555; Mon, 10 Jul 1995 20:19:33 -0400 Received: by hodgkin.mbi.ucla.edu; id AA05550; Mon, 10 Jul 1995 17:27:50 -0700 Message-Id: <9507110027.AA05550@hodgkin.mbi.ucla.edu> To: CHEMISTRY@ccl.net Sender: Computational Chemistry List Errors-To: ccl@ccl.net Precedence: bulk Sorry for wasting the bandwidth but I'd really hear about peoples experiences using windows NT for serious computational chemistry work. So my story is: 1. I'm working on protein folding and some molecular dynamics. Mainly using my own programs and Charmm. And the normal tools people use on a workstation 2. I'm starting a new lab in the fall, and will have to buy computers for me (and eventual gradstudents) I will keep on developing software as well as running programs. (CPU speed is my main goal) 3. I've been looking around and at the moment it seems as if the best price/performance ratio is for some clone makers building machines based on dec-alpha chips, and maybee some other chips too. The problem is that most of these systems do not run any unix but only windows NT. For other systems the UNIX is much more expensive than Win-NT (for instance IBM:s PowerPC). Therefore I'm considering using NT. (Even if I am a UNIX freak.) 4. Linux would be an alternative but the version for the alpha is not out yet. (Last time I looked it was running but networking and X wasn't up yet) My questions: 1. Is any comp-chemistry groups using NT at all ? If so what are your experiences ? 2. Does anyone know about any molecular graphic package that runs under NT. (Other than rasmol) 3. Anyone have had any ecperience with NT for serious computations, (i.e. just running jobs that take a long time and a lot of memory). Is there any benchmarks that compare NT vs Unix. (Probably mainly effectiveness of compilers). 4. What tools do people use under NT ? 5. Am I completely stupid to consider this ? 6. Will I really save money (when I sum up the cost of compilers and tools compared to unixlicenses etc) Thanxs for any help arne -------------------------------------------------------- From: Arne Elofsson Email: arne@hodgkin.mbi.ucla.edu WWW: http://www.doe-mbi.ucla.edu/arne/main.html -------This is added Automatically by the Software-------- -- Original Sender Envelope Address: owner-chemistry@ccl.net -- Original Sender From: Address: arne@hodgkin.mbi.ucla.edu 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 From: ars@ari.net (ARSoftware Corporation) Subject: Re: CCL:Windows NT for computational chemistry ? Date: Tue, 11 Jul 1995 09:28:42 -0400 Return-Path: ars@ari.net Received: from arsoftware.ari.net (arsoftware.ari.net [198.69.192.82]) by mtolympus.ari.net (8.6.5/8.6.5) with SMTP id JAA25197 for ; Tue, 11 Jul 1995 09:28:42 -0400 Message-Id: <199507111328.JAA25197@mtolympus.ari.net> X-Sender: ars@mtolympus.ari.net X-Mailer: Windows Eudora Version 1.4.4 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: Arne Elofsson > >Dear Arne Elofsson, May I suggest you give Dr. Kenneth Tupper a call at ARSoftware, he does some heavy duty computational chemistry work in Windows NT. He can be reached at 301-459-3773. John Parikh > ===================================================== ARSoftware Corporation 8201 Corporate Drive Landover, MD 20785 Phone: (301) 459-3773 FAX: (301) 459-3776 e-mail: ars@ari.net From: hinsenk@ERE.UMontreal.CA (Hinsen Konrad) Subject: Re: CCL:Windows NT for computational chemistry ? Date: Mon, 10 Jul 1995 21:51:53 -0400 Return-Path: hinsenk@ERE.UMontreal.CA Received: from eole.ERE.UMontreal.CA (eole.ERE.UMontreal.CA [132.204.10.20]) by condor.CC.UMontreal.CA with ESMTP id VAA22891 (8.6.11/IDA-1.6 for ); Mon, 10 Jul 1995 21:51:54 -0400 Received: from cyclone.ERE.UMontreal.CA by eole.ERE.UMontreal.CA (950221.405.SGI.8.6.10/5.17) id VAA19475; Mon, 10 Jul 1995 21:51:53 -0400 Received: by cyclone.ERE.UMontreal.CA (950221.405.SGI.8.6.10/5.17) id VAA11947; Mon, 10 Jul 1995 21:51:53 -0400 Message-Id: <199507110151.VAA11947@cyclone.ERE.UMontreal.CA> To: arne@hodgkin.mbi.ucla.edu In-Reply-To: <9507110027.AA05550@hodgkin.mbi.ucla.edu> (message from Arne Elofsson on Mon, 10 Jul 1995 17:27:50 -0700) 3. Anyone have had any ecperience with NT for serious computations, (i.e. just running jobs that take a long time and a lot of memory). Is there any benchmarks that compare NT vs Unix. (Probably mainly effectiveness of compilers). No first-hand experience, but I know some people (not in science) who have seriously considered NT for other computationally intensive tasks. They concluded that NT is in the same ballpark as most modern Unix versions as far as the OS is concerned. The choice of compilers is necessarily smaller, and their quality probably CPU-dependent. I wouldn't expect problems for C and C++, but Fortran could be a problem. 5. Am I completely stupid to consider this ? Considering alternatives is never stupid. But think of the indirect costs: you won't be able to use most of the PD software written for Unix, and you and your students will need time to get used to a new system. And who is going to do system administration? I don't know how big the difference in OS license costs is, but I would be surprised if it were relevant compared to the costs (personnel time) of using a machine and keeping it running. 6. Will I really save money (when I sum up the cost of compilers and tools compared to unixlicenses etc) No idea, but it's not going to be an easy comparison... ------------------------------------------------------------------------------- Konrad Hinsen | E-Mail: hinsenk@ere.umontreal.ca Departement de chimie | Tel.: +1-514-343-6111 ext. 3953 Universite de Montreal | Fax: +1-514-343-7586 C.P. 6128, succ. A | Deutsch/Esperanto/English/Nederlands/ Montreal (QC) H3C 3J7 | Francais (phase experimentale) ------------------------------------------------------------------------------- From: polowin@hyper.hyper.com (Joel Polowin) Subject: Re: CCL:Windows NT for computational chemistry ? Date: Tue, 11 Jul 95 12:39:52 -0400 Return-Path: polowin@hyper.hyper.com Received: from hyper.hyper.com (hyper.hyper.com [204.50.3.217]) by www.hyper.com (8.6.9/8.6.9) with SMTP id MAA23694 for <@www.hyper.com:arne@hodgkin.mbi.ucla.edu>; Tue, 11 Jul 1995 12:09:41 -0400 Received: by hyper.hyper.com (920330.SGI/920502.SGI) for @www.hyper.com:arne@hodgkin.mbi.ucla.edu id AA19878; Tue, 11 Jul 95 12:39:52 -0400 Message-Id: <9507111639.AA19878@hyper.hyper.com> To: Arne Elofsson Hi, Arne. > 1. Is any comp-chemistry groups using NT at all ? > If so what are your experiences ? > 2. Does anyone know about any molecular graphic > package that runs under NT. (Other than rasmol) > 3. Anyone have had any ecperience with NT for serious > computations, (i.e. just running jobs that take a > long time and a lot of memory). Is there any benchmarks > that compare NT vs Unix. (Probably mainly effectiveness > of compilers). Well, for what it's worth, we've got a version of HyperChem that runs under NT on the DEC Alpha; its cost seems to be reasonable compared with other software (academic price for Release 4 is US$1250). As for power, we get about the same kind of performance with our stuff on a Pentium as with the DEC Alpha and with a variety of UNIX boxes -- regardless of platform, I get a more-or-less linear plot of computational speed (1/calculation time) vs. processor speed. Obviously there are a lot of variables, what with different hardware and operating system setups, but working on an Alpha certainly doesn't appear to be unreasonable. I'll append our benchmarking info below. Regards, Joel ------------ Joel Polowin, Ph.D. Manager, Scientific Support Email to: polowin@hyper.com Hypercube Inc, 419 Phillip St, Waterloo, Ont, Canada N2L 3X2 (519)725-4040 Info requests to: info@hyper.com Support questions to: support@hyper.com Email group: Send "subscribe hyperchem" to hyperchem-request@hyper.com WWW: http://www.hyper.com/ MNDO energy and gradient for C60, converged to 1e-2, without DIIS: Computer Time (s) Rel. speed 486DX 50 MHz 375 1 Pentium 90 MHz 87.6 4.28 DECpc AXP 150 MHz 59.8 6.27 SGI Indy R4000SC 100 Mhz 77.8 4.82 IBM RS/6000-250 66 MHz 51.8 7.24 (? - this is way off the line) DEC 3000 200 MHz (I think) 42.4 8.84 The first two numbers are for full HyperChem (user interface and calculation module) running on the same processor under Windows 3.11. The third is for full HyperChem running under Windows NT. The last three are for the computation module running on the unix computer with the HyperChem user interface running on a 486. All times are wall clock times for start to finish observed on the PC. All times are for HyperChem R4. The 486DX 50 MHz is a clone with 128k secondary cache and 20Mb RAM. The Pentium 90 MHz is a Dell Dimension XPS 90 with 256k secondary cache and 16Mb RAM. The SGI Indy 100MHz is an R4000SC with 1Mb secondary cache and 64Mb RAM. I don't know the cache/memory configurations of the IBM or DEC. All times were measured with a hand held stopwatch when machines were "quiet" (no other users) but not in single user mode. The maximal optimization switches that produced working code (-O2 for the RISCs I believe) were set for all compilations, and the source code was identical. From: "Michael D. Bartberger" Subject: Re: CCL:Windows NT for computational chemistry ? Date: Tue, 11 Jul 1995 00:38:39 -0400 (EDT) Return-Path: bartberg@server.chem.ufl.edu Received: (from bartberg@localhost) by server.chem.ufl.edu (8.6.10/8.6.9) id AAA06244; Tue, 11 Jul 1995 00:38:41 -0400 To: Arne Elofsson In-Reply-To: <9507110027.AA05550@hodgkin.mbi.ucla.edu> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Dear Arne: You may wish to take a serious look at Solaris x86. The folks at Sun have released this, and for a short time a while ago they were offering this for a special academic price of $99, although I think this is no longer the case. As far as benchmarks, I am not really certain, although I'm sure there are people at Sun who can help you out with this. For starters, you may wish to have a look at http://www.sun.com. Our sysadmin here did buy some copies of this (he is starting his own business (internet access provider) and from what I understand, he has been extremely pleased with its performance on some P-90s. I can give you his e-mail address if you would like to speak with him directly (let me get his permission first) :-) The nice thing about Solaris x86 is that (1) it is "officially" supported, and (2) really anything that "makes" on Sparc architeture should, in theory, "make" on Solaris x86. Unfortunately, I (1) did not respond in time to the $99 offer, (2) am too busy (and too cheap) ;) to buy it at the moment and install it on one of my machines and (3) do not yet know of anyone who has tried out some of the common computational codes under Solaris x86, although in principle they should work if they do so on the Sun Sparcs. This might be an effective low-cost alternative to pricy UNIX workstations, and you can still remain within your beloved UNIX environment. :) Furthermore, the "bang for the buck" in terms of speed / performance of the various flavors of UNIX on PC architechture is quite respectable. If I recall correctly, someone posted here that a P-100 running Linux was competitive with an RS6000/550. With Solaris x86 on a past Pentium you'd get the performance of a "real" OS (not to mention support) at a decent price. I would be extremely interested in any information you receive regarding benchmarks. I hope this helps. Regards, Michael ________________________________________________________________________________ Michael D. Bartberger bartberg@chem.ufl.edu TEL: (904) 392-3580 Department of Chemistry bartberg@qtp.ufl.edu FAX: (904) 846-0296 University of Florida Gainesville, FL 32611 USA ________________________________________________________________________________ From: olavt@ksedb.no Subject: Re: Help with (possible) migration from Unix to NT Date: 12 Jul 1995 11:36:13 GMT Return-Path: olavt@ksedb.no Received: by unix.digital.no (5.65/DEC-Ultrix/4.3) id AA15477; Wed, 12 Jul 1995 13:37:15 -0500 Message-Id: <9507121837.AA15477@unix.digital.no> To: arne@hodgkin.mbi.ucla.edu In article <6gu48ujibc.fsf@hodgkin.mbi.ucla.edu> arne@hodgkin.mbi.ucla.edu (Arne Elofsson) (Arne Elofsson) wrote: > I'd like to know if it is a feasible solution to try > to build a lab on NT machines ? Sure, you are correct that you get a lot more bang for the buck on Windows NT than in the typical UNIX workstation world. > What compilers (mainly on an alpha) are good. > Are they compatible between C and fortran ? On Windows NT/Alpha you have Visual C++ 2.0. I have also heard that a port of the GNU C compiler is in progress. DEC has DEC Fortran for Windows NT/Alpha. You should be able to call DEC FORTRAN subroutines from VC++ 2.0. -- - Olav Tollefsen - Microsoft Norway - - Nothing I have written here is in any way, shape, or form official - Microsoft policy or opinion, and should not be represented as such - when quoted here or in any other forum. From: "Noah W. Allen" Subject: Windows NT Date: Tue, 11 Jul 1995 10:54:52 -0400 (EDT) Return-Path: nallen@cms.uncwil.edu Received: by cms.uncwil.edu id AA07765 (5.67a/IDA-1.5.cgs.2.13.95 for arne@hodgkin.mbi.ucla.edu); Tue, 11 Jul 1995 10:54:53 -0400 Message-Id: <199507111454.AA07765@cms.uncwil.edu> To: arne@hodgkin.mbi.ucla.edu X-Mailer: ELM [version 2.4 PL24alpha5] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1036 Arne: As a former IBM and AT&T OS Engineer I would like to encourage you to continue looking for alternatives to the Windows NT OS Platform. I'm presently doing work in computational chemistry using SGI workstations and would refuse explicitly to leave the UNIX environment. Linux for the DEC Alpha chip is due out near the 4Q of this year, maybe with a confirmation it would be worth the wait. I'm preferential to that version of UNIX lately. It's quite accomodating on many regards. Windows NT is a disaster in all regards. I have had minimal success with it handling major throughput, and excess CPU speeds with bandwidth requirements. It's simply not reliable, and I wouldn't consider it. I have a colleague using it who is a windows freak (which I am not) and he even relents to it's shortcomings involving networking and task management. It's just short of functional. Good luck with your decision and if I find more information about this for you, I'll let you know. Noah Allen UNC Wilmington Department of Chemistry From: eisenmen@orion.rz.mdc-berlin.de (Frank Eisenmenger) Subject: alpha & modelling, simulations Date: Tue, 11 Jul 1995 11:16:46 +0200 (MET) Return-Path: eisenmen@orion.rz.mdc-berlin.de Received: from orion.rz.mdc-berlin.de by neptun.edv.mdc-berlin.de with SMTP (16.8/KM-MDC/1.0) id AA09228; Tue, 11 Jul 95 12:21:15 +0200 Received: from localhost by orion.rz.mdc-berlin.de (8.6.4/10.1) id LAA10588; Tue, 11 Jul 1995 11:16:46 +0200 Message-Id: <199507110916.LAA10588@orion.rz.mdc-berlin.de> To: arne@hodgkin.mbi.ucla.edu X-Mailer: ELM [version 2.4 PL21] Content-Type: text Content-Length: 2279 Hi Arne, when I came into my group 1.5 years ago, they just bought 5 alpha-PC (150 MHz) running NT 3.1 without any compiler, just emulating DOS with a speed of a 286 After a while I got NT 3.5, which gives you the power of a 486, if you emulate DOS or Windows. Then I found a very good FORTRAN compiler from DEC, which makes my programs as fast as on our R4000 SGI. And finally, I put OSF/1 (now DEC Unix) on one of those PC which makes the same programs (molecular mechanics with real*8 work) 1.4 times faster. These days you can buy 233 MHz alpha-Stations which are told to be 233/150 times faster. So far, we could not find any decent C-compiler for NT (there is one adapted by Microsoft which seems not to work under NT 3.5). Therefore you cannot expect many things to be ported under NT. Better try DEC unix in any case with the alpha chip. If you want to develop programs and test them, then Linux on Pentium (with correct processor) is a very good idea to have. I tried the g77 FORTRAN compiler and found it to be ok, for portable code at least. Concerning graphics: now one can beat SGI with their GL in terms of the number of (public domain) programs written/implemented for. There might be a GL-library which can be purchased for DEC unix (?). N.B. I am not a memeber of any of these firms mentioned above. I think, a the guy who can really tell you most about alphas is Mrigank form India (he is also on the CCL-list and will answer you, I reckon). Good luck, Frank Eisenmenger +------------------------------------------------------------------------+ | Institute of Biochemistry | | Medical Faculty of the Humboldt-University (Charite) | | | | Hessische Str. 3-4 | | 10115 Berlin, Germany | | | | Phone: +49-30-28468-612 or -239 FAX: +49-30-28468-600 | | E-mail: eisenmen@orion.rz.mdc-berlin.de | +------------------------------------------------------------------------+ From: "Carl J. Appellof" Subject: Re: Help with (possible) migration from Unix to NT Date: Thu, 13 Jul 95 09:21:21 -0700 Return-Path: cja@zso.dec.com Received: from oleum.zso.dec.com by inet-gw-1.pa.dec.com (5.65/24Feb95) id AA07776; Thu, 13 Jul 95 09:12:25 -0700 Received: from chmist.zso.dec.com by oleum.zso.dec.com; (5.65/DECwest-OSF-Nov-11-93) Message-Id: <9507131612.AA16505@oleum.zso.dec.com> Organization: Digital Equipment Corporation X-Mailer: Mozilla 1.1 (Windows; I; 32bit) Mime-Version: 1.0 To: arne@hodgkin.mbi.ucla.edu References: <6gu48ujibc.fsf@hodgkin.mbi.ucla.edu> Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=us-ascii arne@hodgkin.mbi.ucla.edu (Arne Elofsson) (Arne Elofsson) wrote: > >What compilers (mainly on an alpha) are good. >Are they compatible between C and fortran ? > >What tools are people using ? The C compiler for Alpha NT is Microsoft Visual C++ 2.0 for Alpha. It has the same front end as the Intel version, but the back end produces code for Alpha. I believe that Digital has a FORTRAN compiler for NT too. I use emacs 19.29 on my NT Alpha system. Many UNIX utilities like sed and tar have been ported to NT. See if you can get a copy of the free "Digital Roadmap for NT" CD >from your local Digital people. It has copies of many of these utilities. Carl Appellof From owner-chemistry@ccl.net Thu Jul 13 20:09:02 1995 Received: from volvo.wavefun.com for hcj@mazda.wavefun.com by www.ccl.net (8.6.10/930601.1506) id UAA06848; Thu, 13 Jul 1995 20:04:56 -0400 Received: by volvo.wavefun.com (931110.SGI/930416.SGI) for CHEMISTRY@ccl.net id AA17056; Thu, 13 Jul 95 17:04:31 -0700 Received: from mazda.wavefun.com(198.147.95.6) by volvo via smap (V1.3) id sma017054; Thu Jul 13 17:04:24 1995 Received: by mazda.wavefun.com (931110.SGI/930416.SGI) for @volvo.wavefun.com:CHEMISTRY@ccl.net id AA19392; Thu, 13 Jul 95 17:04:28 -0700 Date: Thu, 13 Jul 95 17:04:28 -0700 From: hcj@mazda.wavefun.com (Harry C. Johnson) Message-Id: <9507140004.AA19392@mazda.wavefun.com> To: arne@hodgkin.mbi.ucla.edu Cc: CHEMISTRY@ccl.net In-Reply-To: <9507132244.AA00954@hodgkin.mbi.ucla.edu> (message from Arne Elofsson on Thu, 13 Jul 1995 15:44:07 -0700) Subject: Re: CCL:Summary of NT vs UNIX question. Reply-To: hcj@wavefun.com >>>>> "Arne" == Arne Elofsson writes: Arne> Most people recomended me to bye PC:s instead and run linux Arne> (or Next). My fear for that is that even if the price Arne> performance of a PC is unbeatable it is still pretty far Arne> behind the best workstations in floating point Arne> capacity. Both for NEXT and linux is the lack of a god Arne> fortran compiler a problem too. I guess that solaris for Arne> intel has a fortran compiler, but I haven't seen any numbers Arne> for it. You might check into g77, the GNU project's Fortran 77 compiler. I don't have any experience with it, but their C/C++ compilers are well respected. It is very young, but was in alpha testing for quite some time. Note: This is NOT the same as the f2c compiler/translator. g77 is supposed to be a full fortran compiler. Just a thought. -Harry +-----------------------+------------------+-------------------------+ |Harry C. Johnson IV | ______________ | E-Mail: hcj@wavefun.com | |Computational Chemist | \ _________ / | Phone: (714)955-2120 | |Wavefunction Inc. | \ \\\\\\\\/ | Fax: (714)955-2118 | |18401 Von Karman | \ \\\\\/ | | |Suite 370 | \ \\/ | "Seems to me its all | |Irvine, CA 92715 | \/ | just chemistry" - RUSH | +-----------------------+------------------+-------------------------+