From owner-nwchem-users@emsl.pnl.gov Mon Jul 31 02:57:02 2006 Received: from odyssey.emsl.pnl.gov (localhost [127.0.0.1]) by odyssey.emsl.pnl.gov (8.13.6/8.13.6) with ESMTP id k6V9v2nH008947 for ; Mon, 31 Jul 2006 02:57:02 -0700 (PDT) Received: (from majordom@localhost) by odyssey.emsl.pnl.gov (8.13.6/8.13.6/Submit) id k6V9v2T3008946 for nwchem-users-outgoing-0915; Mon, 31 Jul 2006 02:57:02 -0700 (PDT) X-Ironport-SG: OK_Domains X-Ironport-SBRS: 3.5 X-IronPort-Anti-Spam-Filtered: true X-IronPort-Anti-Spam-Result: AQAAAC9wzUSKWQIJDg0d X-IronPort-AV: i="4.07,197,1151910000"; d="scan'208"; a="2909666:sNHT22444772" X-DL-MFrom: X-DL-Connect: Importance: normal X-MimeOLE: Produced By Microsoft MimeOLE V6.00.3790.504 Content-Class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Subject: RE: [NWCHEM] IBM Power 5 compilation error Date: Mon, 31 Jul 2006 10:56:51 +0100 Message-ID: X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [NWCHEM] IBM Power 5 compilation error thread-index: AcayZseGoILHg0MeTgW2kJyzQ8liHQCFI9/A From: "Van Dam, HJJ \(Huub\)" To: "Dr. Jose Luis Rodriguez-Lopez" Cc: X-OriginalArrivalTime: 31 Jul 2006 09:56:52.0863 (UTC) FILETIME=[A633B8F0:01C6B487] X-CCLRC-SPAM-report: -4.9 : BAYES_00 X-Scanned-By: MIMEDefang 2.37 Content-Transfer-Encoding: 8bit X-MIME-Autoconverted: from quoted-printable to 8bit by mserv6.dl.ac.uk id k6V9urCB032272 Sender: owner-nwchem-users@emsl.pnl.gov Precedence: bulk Dear Jose-Luis, Wonderful that means that your machine set up should be similar to the one I use. So with a bit of luck the following stuff I use should do the trick. I am assuming the use of ksh which seems to be the default on IBM systems. export NWCHEM_TARGET=LAPI64 export USE_MPI=yes export NOAIO=yes export BLASOPT=-lessl export NWCHEM_TOP=/nwchem-4.7 export NWCHEM_MODULES="qm" The BLASOPT variable should only be set if you have the essl library which is IBM's optimised BLAS/LAPACK library. The NOAIO variable relates to a long saga about asynchronous I/O on IBM platforms. There may well be a solution for this but I have never bothered with it. The simplest way out of the mess is to suppress the stuff by setting this variable. If anybody knows what the last and final answer is on this issue then they are kindly invited to let you know. Now change directory to $NWCHEM_TOP/src and run (make sure you are using GNU make, the IBM make command won't work with the NWChem makefiles): make realclean make nwchem_config make When you come to run some calculations you will need to set export RT_GRQ=ON export LAPI_USE_SHM=yes export MP_SHARED_MEMORY=yes export MP_CSS_INTERRUPT=yes You may want to set as well (this depends a bit whether your system software allows you to use one network adapter for both LAPI and MPI communications. At some point that was not possible but it should be now): export MP_MSG_API=mpi_lapi Also in your job script you will need to declare the right network adapters: #@ network.LAPI = csss,not_shared,us #@ network.MPI = csss,not_shared,us And then it should work... Best wishes, Huub ========================================================== Huub van Dam (h.j.j.vandam@dl.ac.uk, +44-1925-603933) ========================================================== -----Original Message----- From: Dr. Jose Luis Rodriguez-Lopez [mailto:jlrdz@ipicyt.edu.mx] Sent: 28 July 2006 18:10 To: Van Dam, HJJ (Huub) Cc: nwchem-users@emsl.pnl.gov Subject: Re: [NWCHEM] IBM Power 5 compilation error Dear Huub van Dam, thanks for your cooperation. Here you have some more information. I gues that the target has to be LAPI64, and the system consists of 12 nodes each with 8 processors, aggregating to a total of 96 processors. %> lslpp -l | grep -i lapi rsct.lapi.nam 2.4.2.0 COMMITTED RSCT LAPI NAM Files rsct.lapi.rte 2.4.2.6 COMMITTED RSCT LAPI Runtime rsct.lapi.samp 2.4.2.0 COMMITTED RSCT LAPI Samples rsct.lapi.nam 2.4.2.0 COMMITTED RSCT LAPI NAM Files rsct.lapi.rte 2.4.2.6 COMMITTED RSCT LAPI Runtime Hope this helps. Best JoseLuis Van Dam, HJJ (Huub) wrote: >Dear Jose, > >I am afraid you'll need to provide some more information on the machine your are working on. You will realize that there are 4 different NWCHEM_TARGET values available for IBM AIX; IBM, IBM64, LAPI and LAPI64. If you are working on power 5 you will probably want a 64-bit binary (64-bit pointers eliminate some memory addressing issues). This means that there are still 2 options left IBM64 or LAPI64. Which one you want depends on whether your machine has the lapi library installed for the inter-node communications or not. Your query however does not give any information on the configuration of your system (I can't even be sure it is a multi-processor machine) so I cannot advice which of the two to use. Could you find out whether your system has lapi, please? > >You should be able to answer this question by running the command: > > lslpp -l | grep -i lapi > >On your system (I hope this is a generally available command on IBM AIX), if this produces a non-empty output than you probably have lapi. Anyway it would be good to send that output so we know what we are talking about. > >Best wishes, > > Huub van Dam > >========================================================== >Huub van Dam (h.j.j.vandam@dl.ac.uk, +44-1925-603933) >========================================================== > > >-----Original Message----- >From: owner-nwchem-developers@emsl.pnl.gov [mailto:owner-nwchem-developers@emsl.pnl.gov] On Behalf Of Dr. Jose Luis Rodriguez-Lopez >Sent: 28 July 2006 03:11 >To: nwchem-users@emsl.pnl.gov >Subject: [NWCHEM] IBM Power 5 compilation error > >Hello all.... >Does anyone had compiled v4.7 on a n IBM Power5 system? >It runs under AIX 5.3 operating system. >I had problems with compilation, so i'd appreciate if somebody can give any advice and input guesses for to compile nwchem > >Best regards >jose Luis > >-- >******************************************************************** >* I P I C y T * Instituto Potosino de Investigación Científica y Tecnológica, AC >* Camino Presa San José 2055 >* Lomas 4a. Secc., 78216 >* San Luis Potosí, S. L. P. * M E X I C O >* Tel.: [+52] (444) 834-2000, ext. 7217 * Fax: [+52] (444) 834-2010 >* http://www.ipicyt.edu.mx http://materials.ipicyt.edu.mx >******************************************************************** > > > > -- ******************************************************************** * I P I C y T * Instituto Potosino de Investigación Científica y Tecnológica, AC * Camino Presa San José 2055 * Lomas 4a. Secc., 78216 * San Luis Potosí, S. L. P. * M E X I C O * Tel.: [+52] (444) 834-2000, ext. 7217 * Fax: [+52] (444) 834-2010 * http://www.ipicyt.edu.mx http://materials.ipicyt.edu.mx ********************************************************************