From jkl@ccl.net Sun Aug 23 04:48:29 1992 Date: Sun, 23 Aug 1992 04:48:23 -0400 From: jkl@ccl.net To: chemistry@ccl.net Subject: List suspended for a week Status: RO Dear Chemists (though some of us forgot how the lab smells...), These are your coordinators speaking. Both me (i.e., Jan Labanowski, jkl@ccl.net) and Dave Heisterberg (djh@ccl.net) are leaving for the American Chemical Society Meeting at Washington, D.C. for the most of next week. If you want to see us, drop by the IBM Booth in the Exposition Area. Since we are the only persons who can operate the list, the list would have to be left unattended for 5 days. We do not want to risk any troubles (i.e., like a month or so ago when some crazy computer started to return bounced messages to the list). Moreover, many people will also be at the ACS Meeting, and maiboxes can overfill. The list distribution of the list will be therefore suppressed starting Sunday afternoon (EST) Aug. 22. List will come back to life on Friday, Aug. 28. You can send messages to the list, but they will not be distributed, though they will be kept. We will appreciate your restraint however, since we will have to distribute all accumulated messages by Since I am reaching this age when you say to other people: "Better careful than sorry...", you need to excuse me. Jan jkl@ccl.net From m10!frisch@uunet.UU.NET Sun Aug 23 14:55:15 1992 Date: Sun, 23 Aug 92 14:53:04 EDT From: m10!frisch@uunet.UU.NET (Michael Frisch) Subject: Gaussian 92 disk requirements for UMP2 frequencies To: chemistry@ccl.net Status: R Subject: Disk space requirements for an MP2 G92 freq job Hello, I am trying to complete a MP2 full core, open shell, analytic frequency calculation with 188 basis functions (306 uncontracted ones) of a 10 atom system( 2 C, 1 O, and 7 H) with no symmetry using the Gaussian92 application. Unfortunately, the formula that predicts the disk space requirements of such a job (3*N^4/4 words of disk space, page 92, line 12 from the top of the Gaussian92 User's Guide when used with the keywords "verystingy" and "scf=direct") does not seem completely accurate, at least for the Cray X/MP. On a smaller job (130 basis functions) the ammount needed seemed closer to 4*N^4/5 words. Using the formula I find I need 7,495,190,016 bytes. I ran the job on a 8.2 gigabyte partition and the job failed after completing the MP2 gradient part because it run out of disk space (while writing on the *.scr file). Gaussian 92 revisions A and B take more disk for UMP2 frequencies than the documented amounts. The good news is that revision C, which is now shipping and which has been sent to your site automatically because you are on maintenance, has been greatly improved in this area, and now takes more like 1/2 N^4 words. The procedure for these big jobs should be to use MP2=(Stingy,MaxDisk=N) SCF=Direct, where N is the disk limit for the queue (or a bit less for safety). VeryStingy will be done automatically if necessary. There is still a fixed minimum amount of disk, but this will do the best it can. For RevC and the job above this should be under 7 GBytes, so if you run in your 8 GByte queue and specify MaxDisk=950000000 things should work fine. Mike Frisch ------- **** End of Forwarded Message **** -------