Skip to content

SCIS GPU Cluster Changelog

Summer Maintenance Update (10-Jul-23 to 17-Jul-23)

Memory limits are now strictly enforced. Jobs exceeding memory usage will be terminated with an "out of memory" error. Memory limits for a job can be adjusted in your job submission script under the parameter #--sbatch mem=

The maximum memory resource quota allocation (for GPUs and CPUs) can be fetched using the “myinfo” command while you are connected to the cluster. If you require more memory, feel free to reach out to us.

  • Update Nvidia drivers to 535.54.03
  • Update kernel to 4.18.0-477
  • Module changes :-
  • Add python 3.8.17
  • Add python 3.9.17
  • Add python 3.10.9
  • Add python 3.11.4
  • Add CUDA 12.1.1
  • Add CUDA 12.2.0
  • Add cuDNN 8.9.2.26
  • Add cuDNN 8.9.3.28
  • Transition to cgroupsv2

Winter Maintenance Update (1-Dec-22 to 16-Dec-22)

  • Update Nvidia drivers to 525.60.13
  • Update kernel to 4.18.0-425
  • Add container support via enroot
  • Module changes :-
  • Add python 3.10.9
  • Add python 3.9.16
  • Add python 3.8.16
  • Add python 3.7.16
  • Add CUDA 11.6
  • Add CUDA 11.7
  • Add CUDA 11.8
  • Add cuDNN 8.7.0.84