From owner-nwchem-users@emsl.pnl.gov Thu Apr 7 09:12:15 2005 Received: from odyssey.emsl.pnl.gov (localhost [127.0.0.1]) by odyssey.emsl.pnl.gov (8.12.10/8.12.10) with ESMTP id j37GCFxQ012566 for ; Thu, 7 Apr 2005 09:12:15 -0700 (PDT) Received: (from majordom@localhost) by odyssey.emsl.pnl.gov (8.12.10/8.12.10/Submit) id j37GCFcC012565 for nwchem-users-outgoing; Thu, 7 Apr 2005 09:12:15 -0700 (PDT) Date: Thu, 07 Apr 2005 09:12:10 -0700 From: Edoardo Apra` Subject: Re: Nwchem 4.6 does not run on Linux Cluster using mpirun In-reply-to: To: Philip J Brucat , nwchem-user Message-id: <42555BDA.3030105@pnl.gov> Organization: PNNL MIME-version: 1.0 Content-type: text/plain; format=flowed; charset=ISO-8859-1 Content-transfer-encoding: 7bit X-Accept-Language: en-us, en User-Agent: Mozilla Thunderbird (X11/20050406) References: <8139338.1112844664309.JavaMail.oracle@ocsapp4.utpa.edu> <4254AD2D.2040108@pnl.gov> X-OriginalArrivalTime: 07 Apr 2005 16:12:10.0031 (UTC) FILETIME=[8D32E3F0:01C53B8C] Sender: owner-nwchem-users@emsl.pnl.gov Precedence: bulk Philip please send me the output of the following command nm $NWCHEM_TOP/bin/$NWCHEM_TARGET/nwchem |grep -i mpi_abort Thanks, Edo Philip J Brucat wrote: > All, > I am having the same problem described below with the 4.7 code > compiled with USE_MPI=y. Are there any known issues with nwchem and > mpich? > > /pjb > On Apr 6, 2005, at 11:46 PM, Edoardo Apra` wrote: > >> Nikolaos, >> I am assuming you are using the binary downloaded from the NWChem >> web site, is this correct? >> If this is the case, the NWChem binary are built with TCGMSG and not >> with MPI. >> If you wish to use MPI, please compile the NWChem source after >> setting USE_MPI=y. >> The following web page explains how to use TCGMSG based binaries >> http://www.emsl.pnl.gov/docs/nwchem/doc/user/ >> node44.html#SECTION004420000000000000000 >> >> dimakis@utpa.edu wrote: >> >>> Hello >>> >>> I am using nwchem on Linux cluster. I was able to run nwchem either as >>> nwchem >>> >>> or mpirun -np 18 nwchem . What I observed is that under >>> mpirun the job is not distributed to more than one CPU and its runs >>> at almost the same time regardless of the np parameter. The mpi is >>> mpi-ch-p4 and it has been verified that it works under other test >>> programs that it does distributes the jobs to all nodes. >>> >>> I am using nwchem included as binary into ecce distibution. >>> >>> Thank you >>> >>> >>> Nikolaos Dimakis >>> Assistant Professor >>> Physics and Geology >>> University of Texas-Pan American >>> Phone:(956) 380-8761 >>> >>> > --- > P J Brucat > Department of Chemistry and Center for Chemical Physics > University of Florida > PO Box 117200 > Gainesville, FL 32611-7200 > PH: (352) 392-2006 >