Priority queues provide elevated scheduling priority for users and groups with dedicated allocations on NVwulf. These partitions are not listed in the main queues table and are intended for users who have been specifically granted access.
Available Priority Partitions
Each priority partition mirrors a standard partition in terms of hardware and runtime limits, but jobs submitted to a priority queue receive higher scheduling priority over general-access jobs on the same nodes.
| Priority Partition | Equivalent Standard Partition | GPU Type | GPUs/Node | Max Runtime | Max Nodes |
|---|---|---|---|---|---|
| p-h200x4 | h200x4 | H200 SXM5 | 4 | 8h | 2 |
| p-h200x4-long | h200x4-long | H200 SXM5 | 4 | 48h | 1 |
| p-h200x8 | h200x8 | H200 SXM5 | 8 | 8h | 1 |
| p-h200x8-long | h200x8-long | H200 SXM5 | 8 | 48h | 1 |
| p-b40x4 | b40x4 | RTX PRO 6000 Blackwell (B40) | 4 | 8h | 2 |
| p-b40x4-long | b40x4-long | RTX PRO 6000 Blackwell (B40) | 4 | 48h | 1 |
Submitting to a Priority Partition
To submit a job to a priority partition, use the partition name in your job script or srun command exactly as listed above. For example:
All other job script parameters (GPU count, CPU count, memory, etc.) follow the same conventions as the standard partitions. See the Getting Started on NVwulf guide for general job submission instructions.