From owner-chemistry@ccl.net Tue Sep 21 13:14:00 2010 From: "Peter Burger burger],[chemie.uni-hamburg.de" To: CCL Subject: CCL: 7.2K SATA vs 15K SAS|a|CCSD(T) Message-Id: <-42796-100921130922-4286-uWXXwRITywkpw4AelX/OCw|a|server.ccl.net> X-Original-From: "Peter Burger" Date: Tue, 21 Sep 2010 13:09:20 -0400 Sent to CCL by: "Peter Burger" [burger]^[chemie.uni-hamburg.de] Dear CCLers. We are in the process to purchase new nodes and have to decide wether to buy machines with more memory >> 48 GB or faster hard disks: I would therefore like to learn from your experience with the dependence of the performance of coupled cluster calculations on the file I/O system. The actual question is, how well fare the latest 7.2K rpm SATA disks with large areal density in comparison with 15K rpm SAS disks for CCSD(T) calculations. I guess one important question I am asking is, how scratch files are accessed. If it is dominated by linear transfer from the disk, the latest SATA disks should perform pretty well. If random access plays a major role SAS disks should do way better. Thanks alot for your input in advance. Peter From owner-chemistry@ccl.net Tue Sep 21 14:23:00 2010 From: "Christopher Cramer cramer]_[umn.edu" To: CCL Subject: CCL: New pKa review Message-Id: <-42797-100921120134-16326-IKE4k+9FV/l1Yp0zDj91Qw() server.ccl.net> X-Original-From: Christopher Cramer Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset=us-ascii Date: Tue, 21 Sep 2010 11:01:26 -0500 Mime-Version: 1.0 (Apple Message framework v1081) Sent to CCL by: Christopher Cramer [cramer-x-umn.edu] Listmembers, Apropos the recent posts on pKa prediction (a subject of perennial interest for CCL), my mail this morning brought me Volume 6 of Annual Reports in Computational Chemistry, where Alongi and Shields have a very thorough review article entitled "Theoretical Calculation of Acid Dissociation Constants: A Review Article", pp. 113-138. Those with interests in the area will no doubt wish to consult this review -- chock-full of details and references through 2010. No doubt you can request a reprint from George Shields if you don't have handy access to the volume itself, but, as a former Chair of the Computers in Chemistry Division of the ACS (lo these many years ago...), I feel obliged to point out that, for the paltry price of COMP membership, you can guarantee yourself a shelf full of Annu. Rep. Comp. Chem. volumes moving forward. I honestly have found these to be a wonderful contribution that the COMP Division makes to the field -- inevitably packed with timely and pedagogically oriented review articles -- I salute the current editor Ralph Wheeler and the COMP leadership that fostered and developed the initiative. Chris Cramer -- Christopher J. Cramer Elmore H. Northey Professor University of Minnesota Department of Chemistry 207 Pleasant St. SE Minneapolis, MN 55455-0431 -------------------------- Phone: (612) 624-0859 || FAX: (612) 626-7541 Mobile: (952) 297-2575 email: cramer]=[umn.edu jabber: cramer]=[jabber.umn.edu http://pollux.chem.umn.edu/~cramer (website includes information about the textbook "Essentials of Computational Chemistry: Theories and Models, 2nd Edition") From owner-chemistry@ccl.net Tue Sep 21 14:57:00 2010 From: "Gennady L Gutsev gennady.gutsev/a\famu.edu" To: CCL Subject: CCL:G: deleted file restoration Message-Id: <-42798-100921143141-30176-KVxv/chnu9XZDecI+fRKAw^server.ccl.net> X-Original-From: "Gennady L Gutsev" Date: Tue, 21 Sep 2010 14:31:40 -0400 Sent to CCL by: "Gennady L Gutsev" [gennady.gutsev() famu.edu] Is there a possibility to restore accidentally deleted Gaussian output and checkpoint files on a cluster with PBS ? Thank you, Gennady From owner-chemistry@ccl.net Tue Sep 21 15:33:01 2010 From: "Jamin Krinsky jamink!^!berkeley.edu" To: CCL Subject: CCL:G: Guess=Read Message-Id: <-42799-100921145716-15961-gMbwaR/kkwxAvQSw39PSgw:server.ccl.net> X-Original-From: Jamin Krinsky Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset=ISO-8859-1 Date: Tue, 21 Sep 2010 11:57:05 -0700 MIME-Version: 1.0 Sent to CCL by: Jamin Krinsky [jamink~~berkeley.edu] Hi Shirin, If you mean that you want to make a copy of the checkpoint file with a new name for use in the second part of a two-part calculation (using the Link1 option), copying from the checkpoint file for the first part, this is not possible. The second part overwrites the the first part. If you just want to retain the chk file from the first part, you can specify not to save the second part to checkpoint like so: %chk=yourcheckpointname.chk # route section here Title Mol spec --Link1-- nosave %chk=yourcheckpointname.chk # guess=read geom=allcheck New Title The other option is, instead of using Link1, run two Gaussian jobs using a batch script, something like this (in a linux C shell): !/bin/csh g03 < firstjob.com > firstjob.out cp firstjob.chk secondjob.chk g03 < secondjob.com > secondjob.out Where you set up secondjob.com before running the script, which uses guess=read. Then you would just execute your script in the directory containing your files. Hope this helps. Jamin On Sat, Sep 18, 2010 at 11:36 PM, Shirin Seifert shirin.seifert!=!gmail.com wrote: > Thank you Samuel.However, I need to do several jobs on the same chkpoint > file using  --Link1-- option. > > > On Sat, Sep 18, 2010 at 3:32 PM, fatso fatsi > wrote: >> >>  Hi Shirin, >> >>  I don't think or know if what you say is possible. Follow this steps >> however: >> >> a) copy your original chkpoint file to a new name. >>     you can use something like: cp original.chk real-original.chk >> >> b) do you guess=read run on the original.chk file >> >> c) rename the chk files after your calculation is done. >> >> I hope that works, >> Samuel Odoh >> >> On Sat, Sep 18, 2010 at 1:58 AM, Shirin Seifert shirin.seifert a gmail.com >> wrote: >>> >>> Dear All, >>> When I use guess= read in the route section, can I save the generated >>> checkpoint file into a new one (with different name)? how? >>> Thanks in advance >> > > From owner-chemistry@ccl.net Tue Sep 21 16:34:00 2010 From: "Jerome Kieffer jerome.Kieffer.~!~.terre-adelie.org" To: CCL Subject: CCL: 7.2K SATA vs 15K SAS~!~CCSD(T) Message-Id: <-42800-100921144821-9363-/wBuOQLm8k9jZPH8HyBrwQ~!~server.ccl.net> X-Original-From: Jerome Kieffer Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset=ISO-8859-1 Date: Tue, 21 Sep 2010 20:48:12 +0200 Mime-Version: 1.0 Sent to CCL by: Jerome Kieffer [jerome.Kieffer{}terre-adelie.org] On Tue, 21 Sep 2010 13:09:20 -0400 "Peter Burger burger],[chemie.uni-hamburg.de" wrote: > The actual question is, how well fare the latest 7.2K rpm > SATA disks with large areal density in comparison with 15K > rpm SAS disks for CCSD(T) calculations. I have a workstation with 2 15krpm in Raid1 (mirroring) and a large (1.5TB) 7200rpm disk for data. I am not doing CCSD(T) calculation with it, but the large SATA disk is even faster (with large data chunks) than the SAS ones. root!^!lintaillefer:~# hdparm -tT /dev/sda /dev/sda: Timing cached reads: 13204 MB in 2.00 seconds = 6608.17 MB/sec Timing buffered disk reads: 360 MB in 3.02 seconds = 119.38 MB/sec root!^!lintaillefer:~# hdparm -tT /dev/sdb /dev/sdb: Timing cached reads: 12932 MB in 2.00 seconds = 6471.45 MB/sec Timing buffered disk reads: 372 MB in 3.00 seconds = 123.99 MB/sec -- Jérôme KIEFFER http://www.terre-adelie.org