From chemistry-request@ccl.net Mon Mar 25 10:26:07 1991 Date: Mon, 25 Mar 91 08:41:20 CST From: Piotr Paneth Subject: Draw2 To: CHEMISTRY@ccl.net Status: R I am looking for draw2 program which can be run on any ibm (mainframe or pc), or other public domain program or commercial program which can read/ write mopac/ampac files and display structures. Piotr From chemistry-request@ccl.net Mon Mar 25 15:41:53 1991 Date: Mon, 25 Mar 91 15:07:11 EST From: m10!frisch@uunet.UU.NET (Michael Frisch) Subject: Gaussian 90 SCF times To: chemistry@ccl.net Status: R On the subject of workstations for ab initio calculations, one person commented: We have been having very good experience with the Silicon Graphics RISC servers, running both Gaussian-90 and GAMESS (as well as a host of other numerical simulation, scattering theory, and other heavy computations). Running on one (of the 4) CPU of a year-old 4D/340 with 33Mhz CPU's, a large (ca. 850 Mb of scratch) SCF (G-90) calculation which took 100 minutes on a CRAY-XMP, ran in only 300 minutes on our SGI box. I understand that < $20K 1-CPU versions of this machine WITH graphics are now available, and from our experience, they would also be very robust multi-user servers; we are usually running with several background jobs, plus several interactive users using both dumb terminals and X-terminals, and notice no real problems. These machines can also be readily goosed up with cheap memory (e.g., $982.Cdn/8Mb from Kingston) and SCSI disks (ca. $3300.Cdn/1.2Gb) from 3-rd parties. The SGI is a good machine, however, I'd like to point out that an SCF job which takes 100 minutes on a Cray in Gaussian 90 using conventional SCF typically takes half that much CPU time when run in direct mode. That doesn't change the above conclusions, but we would like people to get the most out of their Crays as well! Michael Frisch Gaussian, Inc. ------- From chemistry-request@ccl.net Mon Mar 25 16:33:35 1991 From: mwitten@cerberus.chpc.utexas.edu Subject: chemistry benchmark query To: chemistry@ccl.net Date: Mon, 25 Mar 91 14:49:27 CST Status: R I am curious about anybody's experience using the IBM Risc boxes to perform computational chemistry problems. This is a rather general query, but we are interested in the problem as we are about to acquire a couple of the IBM's. Direct responses to either of the addresses below please. Thanks, Matthew Witten, Ph.D. -- _____________________________________________________________________ Matthew Witten, Ph.D. Director, Applications Research & Development Associate Director, UT System Center For High Performance Computing Balcones Research Center, 1.154 CMS 10100 Burnet Road, Austin, TX 78758-4497 USA Phone: (512) 471-2472 FAX: (512) 471-2445/2449 E-MAIL MWITTEN@HERMES.CHPC.UTEXAS.EDU or MWITTEN@UTCHPC.BITNET "some intellectuals make their living by creating obscurities for the rest of us to puzzle over" _____________________________________________________________________ From chemistry-request@ccl.net Mon Mar 25 16:50:28 1991 Date: Mon, 25 Mar 91 14:49:27 CST From: mwitten@cerberus.chpc.utexas.EDU Subject: chemistry benchmark query To: chemistry@ccl.net Status: R I am curious about anybody's experience using the IBM Risc boxes to perform computational chemistry problems. This is a rather general query, but we are interested in the problem as we are about to acquire a couple of the IBM's. Direct responses to either of the addresses below please. Thanks, Matthew Witten, Ph.D. -- _____________________________________________________________________ Matthew Witten, Ph.D. Director, Applications Research & Development Associate Director, UT System Center For High Performance Computing Balcones Research Center, 1.154 CMS 10100 Burnet Road, Austin, TX 78758-4497 USA Phone: (512) 471-2472 FAX: (512) 471-2445/2449 E-MAIL MWITTEN@HERMES.CHPC.UTEXAS.EDU or MWITTEN@UTCHPC.BITNET "some intellectuals make their living by creating obscurities for the rest of us to puzzle over" _____________________________________________________________________ From chemistry-request@ccl.net Mon Mar 25 16:51:46 1991 Date: Mon, 25 Mar 91 16:21:52 EST From: G. Ravishanker To: mwitten@cerberus.chpc.utexas.edu Subject: chemistry benchmark query Status: R Hi We purchased three IBM RISC/6000 Powerstations last fall. Our primary research is in the area of molecular simulations, using Molecular Dynamics and Monte Carlo simulations. We also have another group doing electrostatic models for proteins. The time hog in all these are the molecular simulations. The molecules we study are DNA and protein and our systems range in size from 600 atoms (for vacuum simulations) to about 7000 (for aqueous solutions). We also run the MD simulations for about 200-300 picosecond range. Our experience so far has been extremely good. We use a program called WESDYN, which was derived from another MD program called GROMOS. Our typical benchmarks show that the speed we get is between 1/8 to 1/10th of the CRAY Y-MP time. Given that we have these boxes give 24 hrs of dedicated processor time, compared to getting 2-3 hrs of CRAY time a day, we are sort of breaking even in terms of throughput. We are able to run 10 picosecond of dynamics on a system of 6000 atoms in 24 hours (clock time). The machine is very easy to install. The service engineers are extremely helpful, though they are not fully knowledgeable. They try all sorts of channels to get the problems solved quickly. The OS is pretty stable and the on-line documentation called info-explorer takes a while to get used to, but is great. X-Windows works great for the most part. (It is only X11 Rev. 3 compliant) Silicon Graphics GL programs work great for the most part. The scheme of opening and resizing windows does not work exactly the way it does on Silicon Graphics machines, but most other operations work just fine. We have had Polygen Quanta, a molecular modeling program up and running on the IBM RISC for the past two months and are pretty happy with it. Overall, we are very pleased with the speed of the machine and its overall stability. We have had instances where the student runs the machine continuously for a week to 10 days. The only complaint we have is the lack of manuals. One needs to purchase it for a price. IBM wants everyone to use the info-explorer. This is fine when the machine is up and running. When one has only one machine and the machine goes down, it creates a big problem. The price is very attractive for the MFLOPS/MIPS ratio of the Powerstation 530. There are reasonable public domain software already available, such as TeX and emacs. Hope this is useful and if you need further info, please feel free to let me know. Ravi From chemistry-request@ccl.net Mon Mar 25 19:28:36 1991 Date: Mon, 25 Mar 91 17:58:48 CST From: shepard@dirac.tcg.anl.gov (Ron Shepard) To: chemistry@ccl.net Subject: Comments on Workstation Evaluation Criteria Status: R I have noticed a few questions about workstation selection for computational chemistry applications. Most responses I have read so far discuss specific applications. This type of response is fine, but a more general characterization might also be useful. For example, the answers to the following questions might be useful in a more general context. (1) Does the workstation support SIMM memory expansion? (2) Does the workstation support SCSI disk expansion? (3) What are the MFLOPS/Dollar and MIPS/Dollar ratios for some of the popular benchmarks? (4) Is the operating system unix-based? If so, then what flavor (SVR2, SVR3, SVR4, BSD2.2, BSD2.3, POSIX.1, POSIX.2, etc)? Which popular "nonstandard" utilities are supported? (e.g. m4, GNU emacs, GNU make, SCCS, RCS, tar, dump, restore, compress, etc.) (5) Does it support standard remote-shell commands and remote procedure calls? Does it support X-windows? (6) Does it both export and mount NFS file structures? (7) Does the fortran compiler support VMS extensions? Does the fortran compiler support any 8x extensions (e.g. array syntax)? Does the C compiler support both K&R and ANSI syntax? Comments: 1: The support of SIMM memory expansion is important if you want to avoid being soaked for memory upgrades. SIMM memory is a commodity selling currently for $35-$45/MB for 80-ns chips. If you are locked into a single vendor selling special memory boards, you can expect to pay up to 10X or more for memory expansion. 2: Similar comments apply to cheap SCSI disk expansion. SCSI disks now list for about $2,500/GB for disks with 10-15 ms access times and 2-4MB/s transfer rates. I don't know what the real "street" price of these devices are (probably 10-40% less than list), but even the list price is much less than many vendor-specific disk-expansion options. Some of these devices are rated at 100,000-200,000 hr. mean time between failure (MTBF). I don't know exactly how these ratings are determined, since I'm fairly sure that none of these disks have really been tested nonstop for 20 years, but such a rating must imply some kind of favorable reliability. Other SCSI devices that might be useful are optical disks (both CD-ROM and writeable removable media), and 8mm or DAT tape devices for backup and archival. 3: The MFLOPS/Dollar and MIPS/Dollar ratios are important, but are unfortunately rather controversial. The numerators are somewhat uncertain since benchmarks, by their very nature, do not measure "real-world" application throughput. However, like MPG ratings for cars, they provide at least a relative measure of performance. The denominator is uncertain since no computer vendor really sells their machines for list price. Perhaps specific units such as "LINPACK_MFLOPS/$STREET_PRICE" should be specified. 4: The type of OS and the utilities have implications about how easy it is to port codes to the machine, how well scripts from your friends and coworkers will work on the macine, and how well the machine will fit into a (presumably heterogeneous) network. These issues may be less important if you are not a programmer, if you know your application runs on the machine, if you work alone, you don't care if you talk to other machines or not, and if no one else is interested in your work. 5-6: These issues affect how well the machine fits in with the idea of distributed computing. Even if your application is intended to run on only a single machine, much software development is now directed towards a distributed computing model. Future versions of your favorite application may depend on some of these features, even if it is running on a single machine. One specific example of this is X-windows software. X-windows was originally designed to be a network-based windowing protocol, but vendors are now using X-windows also as a base upon which to build a graphical user interface (GUI). That is, a client application on a workstation uses the underlying network protocol to talk to the display server application running on the SAME machine. By the way, it is also useful to know which version of X is supported (11.3, 11.4, etc.) and if this software is bundled or sold as a separate product. 7: Does your application use NAMELIST input? If so, then it may be difficult to find a version running on a machine which does not support this common f77 extension. Even if you do, you cannot exchange input decks with your friends. Support of VMS fortran extensions is important because these have become a de-facto standard. Too many vendors in the kitchen spoiled the ANSI standardization effort last decade, so f77+VMS is about all there is to fall back on. Are you using or writing codes that you want to be compatible with the Connection Machine? If so, you will need a fortran that supports the 8x array syntax. My guess is that this extension will become even more common among "supercomputer(t)" compatible codes. [The "(t)" implies a time-dependent definition (:-).] Although I am less experienced with programming in C, I have run into similar problems with the various versions that are available. It is a hassel for someone to give you an ANSI C code when you only have a K&R compiler. Ron Shepard shepard@tcg.anl.gov [The above views are my own and do not necessarily reflect the opinions of any other person, agency, or life-form, intelligent or otherwise, in the universe.]