H100 GPU Policy Change on the MedicineBow

Subject: Updated SLURM Policy for H100 GPU Nodes — 8 GPU Concurrent Use Limit

Due to high demand and limited availability of our H100 GPU nodes, we are implementing a new usage policy effective June 26th, 2026.

What's changing:
Each project on the Medicine Bow compute cluster will be limited to 8 H100 GPUs in concurrent use at any given time.

Why this change:
H100 capacity is a shared, finite resource, and recent demand has made it difficult for all groups on campus to get fair access. This limit is intended to ensure broader availability of H100 resources across the research community, rather than allowing a small number of projects to consume the bulk of capacity at once.

What this means for you:

• Jobs requesting more than 8 GPUs concurrently for a single project will be rejected at submission
• Multiple jobs from the same project that together exceed 8 GPUs will be subject to the same restriction

Questions or concerns:
if this policy creates a significant hardship for your research please reach out to arcc-help@uwyo.edu to discuss options.

We appreciate your understanding as we work to make these resources available to the broader campus community. Thank you for your continued cooperation.