7T RF research facility server usage
Storinator: 272 Terabyte Raids for storage
rflab cluster: 2 NVIDIA TITAN RTX 48 total GPU RAM
Pitt CRC: Cluster: smp, Available SUs: 1700000 | Cluster: mpi, Available SUs: 50000 | Cluster: gpu, Available SUs: 50000 | Cluster: htc, Available SUs: 500000
crc-quota This command shows how much disk storage you have left.
crc-usage This command shows how much computational resources you have used.
crc-idel This command shows the idle computational resources.
squeue -u username --cluster gpu This command shows you the jobs that you are running on the gpu cluster.
crc-interactive -s -b 12These command will request resources on the smp cluster with 12 gb memory. Note: htc is slower than smp. But fewer people request htc and you could get resources allocated faster than smp if you want more cores. To leverage the amount of cores requested you could use the
parallelGNU command.Example:
ls *.nii | parallel --jobs 4 mri_synthstrip -i {} -m {.}_mask.niiThis command will list out all the nii files in your current directory and process skull stripping using mri_synthstrip 4 jobs at a time.
crc-interactive -g -p a100 -b 40 -t 6This commands will request a100 gpu 40 gb memory for 6 hours. To see available computational resources, you could usecrc-idel.