Computational Cluster Programs

How to Run Gaussian on ATS-Hosted Clusters

Note that, by contractual agreement with Gaussian, Inc., only authorized members of the UCLA community can use Gaussian. Inquiries should be directed to Ashay Patel, UCLA Department of Chemistry or to atshpc@ucla.edu. The consultants will reply during normal business hours.

You can use any text editor to create the appropriate input files for Gaussian, or you can use gaussview for Gaussian 09, or gaussview3.09 for Gaussian 03. You must be in an X-Windows environment in order to use GaussView.

Note that your input file should not contain %NProcShared or %LindaWorkers, or %NProc or %NProcLinda instructions specifying the number of processors to be used, or the nodes you job should run on. Your job will request any needed processors at runtime.

In order for Gaussian to make use of your job's memory allocation, your input file may include a %Mem instruction for dynamic memory. The default value for %Mem is 256MB or 32MW (mega-words) per core. The %Mem value should be less than the memory per core that you request for your job. We recommend a maximum value of:

%Mem=800MB for jobs requesting 1024 megabytes per core
%Mem=3800MB for jobs requesting 4096 megabytes per core

You should modify the default %Mem value only if needed because a value that is too large may decrease your job's performance instead of improving it (see: http://www.gaussian.com/g_tech/g_ur/m_running_w.htm).

How to Run Gaussian in Batch

To Run Gaussian from the UCLA Grid Portal

To run Gaussian in batch from the UCLA Grid Portal click the Job Services tab. Then click Applications. If you have access to a cluster that has Gaussian installed and you are on the access list for Gaussian, you can run it on that cluster.

To Run Gaussian from the Cluster Login Node using the Queue Scripts

The easiest way to run Gaussian in batch from the login node is to use the queue scripts. See Queue Scripts for information about using queue scripts.

The following queue scripts are available for Gaussian:

g09.q
gaussian09.q

Asks you whether you want to run in parallel or serial.

g09.q.parallel
gaussian09.q.parallel

Runs in parallel, 8 processors per node.

g09.q.serial
gaussian09.q.serial

Runs in serial.

g03.q
gaussian03.q

Runs in serial.

To Run Gaussian from the Cluster Login Node Using SGE Comands

See Running a Batch Job on an ATS-Hosted Cluster for guidelines to follow to create the required SGE command file. Alternatively, you could create an SGE command file with one of the queue scripts listed above. After saving the command file, you can modify it if necessary, before submitting it to the job scheduler.

How to Run Gaussian Interactively

Gaussian is available interactively through GaussView. See How to Run GaussView on ATS-Hosted Clusters.

August 2011