Skip to content

Partitions

Partitions are Slurm's mechanism for dividing the cluster's limited GPU resources fairly among jobs. Each partition applies a different combination of resource limits, maximum run time, and priority. Use the Slurm flag --partition/-p to select which partition a job runs in.

Partition details

Each job assigned with a priority can preempt jobs with a lower priority: unkillable > main > long. Once preempted, the job is killed without notice and is automatically re-queued on the same partition until resources are available. To leverage a different preemption mechanism, see the Handling preemption page.

Partition Max Resource Usage Max Time Note
unkillable 6 CPUs, mem=32G, 1 GPU 2 days
unkillable-cpu 2 CPUs, mem=16G 2 days CPU-only jobs
short-unkillable mem=1000G, 4 GPUs 3 hours Large but short jobs.
Restricted to 4-GPU nodes only
main 8 CPUs, mem=48G, 2 GPUs 5 days
main-cpu 8 CPUs, mem=64G 5 days CPU-only jobs
long No limit of resources 7 days
long-cpu No limit of resources 7 days CPU-only jobs
Important: H100 GPUs Partition Restrictions

H100 GPUs are ONLY available in the short-unkillable partition.

The short-unkillable partition is restricted to 4-GPU nodes only, specifically:

  • cn-g nodes: A100 80GB GPUs (4 GPUs per node)
  • cn-l nodes: L40S GPUs (4 GPUs per node)

As an exception, it also contains the H100 nodes:

  • cn-n nodes: H100 GPUs (8 GPUs per node, but only 4 can be used per job)

For a complete list of node specifications and GPU details, see Node profile description.

Note

As a convenience, requesting the unkillable, main, or long partition for a CPU-only job automatically translates it to the -cpu equivalent.

When to use each partition and GPU availability

The following table provides a quick reference guide for choosing partitions and understanding GPU availability:

Partition When to use Available GPUs
unkillable High-priority jobs that cannot be interrupted. All GPU types
short-unkillable Large short jobs that need high priority and cannot be interrupted. Restricted to 4-GPU nodes only
main Standard priority jobs with moderate runtime needs. All GPU types
long Long-running jobs that can tolerate preemption. All GPU types except H100
*-cpu CPU-only jobs (no GPU required). N/A (CPU-only nodes)

Comments