Important! The Chemistry Department at UNR has two versions of Gaussian, Gaussian09 and Gaussian16, and 3 different hardware platforms: AMD Istanbul, Intel Sandy Bridge, and Intel Nahalem. You will need to know which queue you are using, and what the corresponding version of Gaussian in the module is. The matrix below shows the proper module and corresponding queue:
| Gaussian Module | Submission Queue |
|---|---|
| gaussian/g09-AMD | amd-old.q | gaussian/g09-intel | intel-old.q OR intel-lgdisk.q OR intel.q |
| gaussian/g16-AMD | amd-old.q |
| gaussian/g16-intel-old | intel-old.q OR intel-lgdisk.q |
| gaussian/g16-intel | intel.q |
Please see the documentation below for specifying the queue, and loading the correct module in your submission script
1) Setup account on annie
Contact Mike Zierten, College Sysadmin for an account on annie
After you login type groups at the prompt it should list gaussian
If you're not in the gaussian group and you use the command g09 it will state command unknown. If this is the case, email The sysadmin again and he will fix it.
2)
a) Windows users: Install Putty, http://www.putty.org/, on your computer and configure it
b) mac users
open a terminal and use ssh and sftp
for example ssh [email protected]
3) use this basic script for job submission (type it into a text file using nano, vi, etc... and save it)
You can also use the scripts under /share/apps/scripts for examples and change them according to your needs
# RUN ALL JOBS UNDER SCRATCH
# IF YOU NEED EXTRA SPACE, SEND EMAIL TO [email protected] or [email protected]
##$ -pe smp 4 this is the number of processors per node that you need (so e.g. 1, 2, 3, …. 16, try to use multiples of 2 or 4)
#!/bin/bash
#$ -N yourJobName
#$ -cwd
#$ -j y
#$ -q amd-old.q
#$ -S /bin/bash
#$ -pe smp 4
echo "Current working directory is $(pwd)"
echo "Running on $(hostname)"
echo "Starting run at: $(date)"
module purge
use /share/apps/modulefiles
module load gaussian/g09-AMD
g09 <inputFile.gjf >output.out
The module load command in the script above uses the gaussian/g09-AMD module which loads the version of gaussian09 compiled for the AMD cpus, and uses the amd-old.q queue. Please see the table above for the correct queues to use with your version of Gaussian
The line "#$ -pe smp 4" uses 4 cores for the job. The amd-old.q queue machines have only 8 cores, the intel-old.q and intel-lgdisk.q machines have 12 cores, and the intel.q machines have 16 cores, so please use the number of cores in your jobs sparingly
4) submit your job to the queue
type qsub scriptname
use whatever name you gave to the script in step (4)
5) monitor you submission using qstat
under state r is running
under state qw is waiting and queued
6) grab a copy of Avogadro for visualization and input file creation