GPUs on the HPC¶
This page covers the Graphics Processing Unit (GPU) resources available on the cluster. The gpu partition is the general GPU resource for HPC users; multigpu is the alternative partition where more than one GPUs are accessible.
Anyone with a cluster account has access to the gpu partition. However, to access the multigpu partition, you must submit a ServiceNow ticket.
Note
When working with shared computational resources, it is important to remember not to leave the jobs idle.
Name |
Requires Approval? |
Time in Hours (Default/Max) |
Submitted Jobs |
GPU per Job Limit |
User Limit (No. GPUs) |
|---|---|---|---|---|---|
|
No |
4/8 |
4/100 |
1 |
4 |
|
Yes |
4/24 |
8/100 |
8 |
8 |
Important
Consider the compatibility of the GPU, as some programs do not work on the older k40m or k80 GPUs.
Execute the following command to display the non-Kepler GPUs that are available:
sinfo -p gpu –Format=nodes,cpus,memory,features,statecompact,nodelist,gres
This indicates the state (idle or not) of gpu-types and could be helpful to find one that is idle. However, the command does not give real-time information of the state and should be used carefully.