Queues (Partitions)
Horizon includes CPU and GPU resources to meet a broad range of computational demands. Accordingly, users will submit jobs that differ significantly in resource type, size, duration, and priority. Horizon's batch queues—which Slurm refers to as "partitions"—have been configured so that users gain access to their preferred resources fairly and efficiently.
Here is a summary of the characteristics of the different queues anticipated for Horizon at the time of this writing (October, 2026):
| Queue Name | Max Nodes/Job (Cores/Job) |
Max Job Time |
Max Jobs per User |
Max Jobs Submitted |
Max Total Nodes per User |
Charge per Node-Hour |
|---|---|---|---|---|---|---|
| gb | 64 nodes (9,216 cores/256 GPUs) |
48 hrs | TBD | TBD | TBD | TBD |
| gb-dev | 8 nodes (1,152 cores/32 GPUs) |
2 hrs | TBD | TBD | TBD | TBD |
| gb-large | 256 nodes (36,864 cores/1,024 GPUs) |
48 hrs | TBD | TBD | TBD | TBD |
| vv | 256 nodes (45,056 cores) |
48 hrs | TBD | TBD | TBD | 0.25 SU |
| vv-dev | 32 nodes (5,632 cores) |
2 hrs | TBD | TBD | TBD | 0.25 SU |
| vv-large | 1,024 nodes (180,224 cores) |
48 hrs | TBD | TBD | TBD | 0.25 SU |
Jobs may be submitted to any of the above queues. Three of the queues feature GPUs. The nodes in the gb, gb-dev, and gb-large queues feature four Blackwell B200 GPUs per node. Detailed specifications for all the types of nodes on Horizon can be found in the Slurm Partitions section of the Horizon User Guide.
Be aware that the queue configuration in the table above is subject to change at any time.
To get up-to-date queue configurations, run this on a login node:
$ qlimits -vCVW material development is supported by NSF OAC awards 1854828, 2321040, 2323116 (UT Austin) and 2005506 (Indiana University)