SAS

SAS (Statistical Analysis System)is a software suite developed by SAS Institute for advanced analytics, multivariate analyses, business intelligence, data management, and predictive analytics.

Usage

SAS requires a license checked out from SLURM to function. See the examples below for the proper syntax.

In the first example, we request an allocation of a single node for 2 hours with 10G of RAM on Interactive partion. Then within the allocation, we load the SAS module to use SAS interactively. Note that to use SAS interactively with GUI, you need to use X forwarding in SSH. Finally, we execute SAS:

[userid@lewis4-r710-login-node223 ~]$ srun -p Interactive --qos interactive --licenses sas:1 -N 1 --mem 10G -t 2:00:00 --pty /bin/bash
srun: job 5687214 queued and waiting for resources
srun: job 5687214 has been allocated resources
[userid@lewis4-s2600tpr-hpc4-node248 ~]$ module load sas/sas-9.4
INFO: License(s) sas:1
[userid@lewis4-s2600tpr-hpc4-node248 ~]$ sas

If all worked as intended, five SAS windows will pop up on your local terminal: SAS:Program Editor, SAS:ToolBox, SAS:Explorer, SAS:Log, SAS:Output, SAS:Explorer

In the second example, we will submit a job using sbatch to the queue that allocates 10G and will have a two day time limit on a single node with one cpu by creating a job file called sas_batch_job.sh:

#!/usr/bin/bash
#SBATCH -J SAS_sample
#SBATCH -o SAS_sample.o%j
#SBATCH -t 2-00:00:00
#SBATCH -N 1
#SBATCH -n 1
#SBATCH --mem 10G
#SBATCH --licenses sas:1

# Load in the SAS module to update our path and libraries
module load sas/sas-9.4

# Execute the serial based SAS program
srun sas -memsize 9G -filelocks none -noterminal hello_sas.sas

When hello_sas.sas is:

data _null_;
file print;
put 'Hello World';

We can use the following to submit the job:

sbatch sas_batch_job.sh

To learn more about requesting resources or submitting jobs review Introduction to Lewis and Clark Clusters.