The most recent version of Gaussian which we support is Gaussian 98 Revision A9. and it is found on 6 machines; G98 Revision A7 also exists.
A 48 node IBM SP/2 (with 80 cpus) (danube.nist.gov)Do you have accounts on them? Click here for information on how to get accounts. You cannot do any work with these packages without getting the accounts, so get yours and then come back here.
The NIST RISC/6000 cluster: four IBM RS/6000 (sci-apps.nist.gov)
A 32 node SGI Origin 2000 (amur.nist.gov)
A 32 node SGI Origin 2000 (neva.nist.gov)
A 32 node SGI Origin 2000 (tay.nist.gov)
An 8 node SGI Origin 2000 (arno.nist.gov)
If you already have accounts on these machines, then you can begin running Gaussian98. It is possible for you to run small interactive or large memory batch calculations. You can also set your environment such that your account is always ready to run Gaussian98 jobs or you may select the version as you wish.
When running a Gaussian98 job, it is crucial to get the correct
setting for the environmental variable g98root. This variable
is needed to proceed with internal Gaussian settings and also to
set the correct PATH variable. In this
example, we will proceed using G98 Rev. A9, but examine the following table
to determine what you should set g98root to on the
|Machine||Revision A9||Revision A7|
One sets g98root like this:
setenv g98root /usr/local/apps/g98a9
To run a quick interactive job, you will prepare your system to run Gaussian98 calculations on a given G98 input file named, g98file.inp, by typing the following 3 lines at the command prompt. Danube is different because it needs a different PATH variable.
Interactive jobs are processed at the command line in real time. These jobs should never be large or consume a lot of machine resources. Interactive job limits are already in place on all machines and if they are exceeded, your session will be logged out.
Jobs which require large amounts of computing power (memory and time) are selected for batch processing. These are submitted to the machine, enter a run queue and execute when the resources become available. There are many details about running a job in a batch environment and special attention has been paid to them in the document: "Using the NIST Central Computing Facility Effectively for Computational Chemistry"
At this point, you can execute Gaussian on g98file.inp and put the results into g98file.out in the background by typing:
Note: use "g98l" on danube and "g98" on all others.
Examine the output file with the standard UNIX "cat" command.
To make your system default to Gaussian98 Rev A9,
Use your favorite editor and add the previous group of 3 lines to your .cshrc file.
Activate the changes by typing this at the command line:
or you may log out and it will be set up correctly the next time you log in.
Submitting a GAUSSIAN job using WebSubmit on the central systems.
To avoid learning about how a batch system operates, you can learn about WebSubmit instead. Briefly,
this web based facility allows one to submit Gaussian jobs securely to the machines without learning
the additional syntax of the batch facility. For further information on WebSubmit click
Now how do you set up the input files ???
Back to COMPUTATIONAL CHEMISTRY at NIST
NIST Scientific Applications and Visualization Group