From chp1aa@surrey.ac.uk Fri Apr 23 06:06:53 1993 From: Mr Andrew D Allen Message-Id: <9304230859.AA12233@central.surrey.ac.uk> Subject: G92 and modelling Anti-cancer drugs To: CHEMISTRY@ccl.net Date: Fri, 23 Apr 93 9:59:24 BST Dear fellow modellers, I am currently modelling a series of anti-neoplastic anti-biotics which are currently in use as chemotherapy treatment. The series are based the parent compound Adrimycin (aka Doxorubicin), which have applications for a wide spectrum of cancer treatment. If anyone has any data they can share from their own modelling programs I would like to hear from you. Thanx in advance Andy. (chp1aa@surrey.ac.uk) From rbw@msc.edu Fri Apr 23 05:05:20 1993 Date: Fri, 23 Apr 93 10:05:20 -0500 From: rbw@msc.edu (Richard Walsh) Message-Id: <9304231505.AA28732@uh.msc.edu> To: chemistry@ccl.net, d3e102@ames.pnl.gov Subject: Re: Benchmark Info. Available David Feller writes ... > > People will have to judge the cost effectiveness of workstations vs super- > computers vs PCs for themselves. However, I can't help but note that the > quantity which is of primary interest to most scientists using these programs > is the "time-to-solution". Very fast, very expensive computers are nearly > always oversubscribed. This results in lengthy waits in long queues for a > chance at the limited resources. Our experience has been that it is often > possible for the unemcumbered tortoise to beat the overburdened hare. > On the other hand (or hare's foot) ... On the PC benchmarks that I ran on the C90 I failed to mention that they were run interactively (as is a substantial amount of the work on our machines). The expansion ratios were much less than 2 (but more than 1 ... no parallel). So with these 1 to 20 hour PC decks with expansion ratios of presumably 1 ... the hare easily wins the throughput race! I know your not surprised, but read on ... a fully parallel G92 on an 8 or 16 cpu C90 raises the prospect of multiple 'harespower' and fractional expansion ratios. And, of course, real industrial stength jobs might not have a ratio of 2 even at our (22 CRAY cpus) site, but would the tortoises do better ... a 16 cpu C90 provides at least a factor of four more capacity than a 8 cpu YMP. From the back of my envelope, I conclude, that even with a expansion ratio of 3, on these small problems I can do more computing in three days than the PC in question can do in a year. Richard Walsh Minnesota Supercomputer Center From Mike_Hopper.XRCC@xerox.com Fri Apr 23 02:04:09 1993 Date: Fri, 23 Apr 1993 09:04:09 PDT From: hopper.xrcc@xerox.com Subject: Developments in Chemical Engineering Journal? To: chemistry@ccl.net Message-Id: <"23-Apr-93 12:04:09 EDT".*.Mike_Hopper.XRCC@Xerox.com> I have a reference to an article titled:- "Chemical Process Simulation Packages" authored by M.A. Hussain that appeared in the Journal DEVELOPMENTS in CHEMICAL ENGINEERING vol 15, #3, pp 163-178, 1992. Can anyone identify the publisher and/or ISBN of this publication. Our Technical Information Centre has not been able to find out anything about this publication. Thanks Mike Hopper Scale Up Engineering Laboratory Xerox Research Centre of Canada From DSMITH@uoft02.utoledo.edu Fri Apr 23 08:32:56 1993 Date: Fri, 23 Apr 1993 13:32:56 -0500 (EST) From: "DR. DOUGLAS A. SMITH, UNIVERSITY OF TOLEDO" Subject: direct vs. in-core calculations To: chemistry@ccl.net Message-Id: <01GXCOZA7C88005EON@UOFT02.UTOLEDO.EDU> When using a program such as Gaussian 92, when should one use the DIRECT option and when should one use the INCORE option. Recent postings to this mail exploder (Cray timings, in comparison to 486 wall clock times) showed tremendous differences in some cases between these two options. Thanks for the comments. Doug Douglas A. Smith Assistant Professor of Chemistry The University of Toledo Toledo, OH 43606-3390 voice 419-537-2116 fax 419-537-4033 email dsmith@uoft02.utoledo.edu From feng@sgi.chem.temple.edu Fri Apr 23 08:03:39 1993 Date: Fri, 23 Apr 93 12:03:39 -0400 From: feng@sgi.chem.temple.edu (Feng Chen) Message-Id: <9304231603.AA00799@sgi.chem.temple.edu> To: chemistry@ccl.net Subject: PCModel ? Hi: Can somebody provide me some infomation about PCMODEL ?, please. Is it a commercial software ? If it is, their phone number ? Is there a shareware version available on anonymous FTP site ? Thank you very much. -- TTTTTTTTTT Feng Chen -- feng@sgi.chem.temple.edu II Tel: (215)204-5899 FAX: (215)204-1532 II II Dept of Chem, Temple Univeristy, Phila, PA 19122 TTTT ---------------------------------------------------------------------- From mail Fri Apr 23 09:33:31 1993 Date: Fri, 23 Apr 93 09:26:17 EDT From: ranck@albert.etown.edu (John P. Ranck) Message-Id: <9304231326.AA05166@albert.etown.edu> To: chemistry@ccl.net Subject: Gaussian 92 on PC Folks, In comparing the performance of Gaussian 92 for Windows running on a PC with the performance of Gaussian running on a Cray, I think we are missing the point. We all know that one obtains super performance on a supercomputer. The important matter is that (for a very modest $600 software cost) it is now POSSIBLE to do Gaussian ab initio computations on the PC's that we all have on our desks or in our classrooms that sit turned off or idle much of the time. This represents CAPABILITY WE DID NOT PREVIOUSLY HAVE -- either because we did not have a machine that would run Gaussian or we could not afford the programs or the charges for computer time. Sure, programs run slower on smaller machines, but for many of us, the difference between two hours and overnight is not significant. It will be tomorrow until we have time to look at the results anyway. As to Gaussian performance on an Intel 486 based machine, I offer the following. Two years ago, I ported Gaussian 90 to a PC running under Interactive Unix. The configuration of the PC was: Processor: i486 DX/25 (NOTE: far slower than today's machines) Hard Drive: Maxtor 213 MB IDE RAM: 16 MB All 170+ tests in the Gaussian test suite (except for two or three requiring truly huge disk files) were run successfully on this machine. (Even these tests were run successfully using a drive on another machine mounted via nfs.) The performance in each case was within a factor of about 2 (faster or slower!) compared to the same test run on a Sun SPARC system: Processor: 4/110 Hard Drive: 760 MB RAM: 24 MB The results were not exactly comparable because some routines choose the algorithm for computation in response to the amount or RAM and/or disk space available. Thus some tests on the PC were faster than on the Sun while others were slower -- BECAUSE THE MACHINES HAD DIFFERENT RAM/DISK CONFIGURATIONS. Because the majority of the time in a Gaussian run is spent crunching numbers rather than interacting with the user, the time for a run is more dependent on the processor than on the operating system or the GUI. Thus, I would expect performance of Gaussian 92 for Windows/DOS to be comparable with the results I obtained under Interactive Unix ON THE SAME MACHINE and the few tests I have run confirm this. I had one additonal problem. With my PC running Interactive Unix, what else could I do? Essentially nothing except software development projects because my everyday apps run under DOS. So -- I kept a second DOS/Windows PC on my desk to do my everyday work. The folks at Gaussian deserve a great big cheer for porting the entire Gaussian package to run IN MY MULTIPROCESSING ENVIRONMENT IN THE MACHINE ON MY DESK. If my office task requires lots of machine resources so that the MS/Windows performance is sluggish with Gaussian running in the background, I can suspend Gaussian processing until I get up for coffee or have to go to a meeting. What about the "under $3000" claim? An ad in the April 19 PC Week lists the following PC from Dell Computers for $2399. Processor: i486 DX2/66 Hard Drive: 230 MB IDE RAM: 4 MB Video: 14" SVGA (1024 x 768) Dual HD Diskettes DOS and Windows and Mouse Also in the same magazine, 12 MB additional RAM can be obtained for $348. Total cost: $2747 for a PC with substantially more processor speed than mine. Contrary to one recent posting, one does not NEED one gigabyte of disk space nor 32 MB RAM. (This estimate was made by a vendor -- not a user.) Hope this helps. John P. Ranck Internet: ranck@vax.etown.edu Department of Chemistry Voice: 717-361-1315 Elizabethtown College FAX: 717-361-1207 Elizabethtown, PA 17022-2298