Skip to content

Resource_usage.md #302

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Graviton28 wants to merge 2 commits into
base: master
Choose a base branch
from
Open

Resource_usage.md #302

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
cf88eb8
Resource_usage.md
Graviton28 Jul 30, 2026
da12a60
fix Hopper debug partition walltime, was wrong (2 days vs actual 4 ho…
Graviton28 Aug 3, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
54 changes: 19 additions & 35 deletions Resource_usage.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,49 +1,33 @@
# Storage Policy

Home directories are limited to 200 GB of storage. Project space is limited to 250 GB of storage. Scratch storage is limited to 1 TB (2 TB on Xena). Center-wide project scratch space is limited to 3 TB. To purchase additional storage please see our [pricing spreadsheet](https://carc.unm.edu/research/premium-research-computing-services.html).
Home directories are limited to 100 GB soft / 200 GB hard quota. Scratch storage (`/easley/scratch` on Easley) has no personal per-user quota enforced. To purchase additional storage please see our [pricing spreadsheet](https://carc.unm.edu/research/premium-research-computing-services.html).

# Resource Usage Policy

Soft limits and hard limits are used to provide fair scheduling without wasting resources. The job scheduler will schedule all jobs meeting the soft limit requirements. If there are additional resources after all jobs meeting the soft limits are scheduled, then jobs meeting the hard limits are scheduled. The hard limits prevent users from monopolising the cluster for long periods by starting large jobs during temporary lulls in utilisation.
Slurm schedules jobs against per-user resource limits set on each partition's QOS (Quality of Service). These limits cap how much CPU, memory, and GPU a single user can have allocated at once on a given partition, so no one user can monopolize a partition and starve everyone else's jobs.

## Xena Configuration
## Easley Configuration

Parameters | Soft Limit | Hardlimit
--- | --- | ---
Number of Processors | 64 | 128
Number of Nodes |4 (xena) <br> 2(BigMem) | 8 (xena) <br> 4 (Bigmem)
Processors per Node | 16(xena) <br> 32(BigMem) | 16 (xena) <br> 32 (BigMem)
| Partition | Max CPUs/User | Max Memory/User | GPUs/User | Walltime |
| --- | --- | --- | --- | --- |
| general (default) | 512 | ~2 TB | - | 2-00:00:00 |
| bigmem | 64 | ~2 TB | - | 2-00:00:00 |
| h100 | 64 | ~1 TB | 2 | 2-00:00:00 |
| l40s | 96 | ~168 GB | 3 | 2-00:00:00 |
| interactive | 64 | ~250 GB | 1 (on GPU nodes) | 04:00:00 |
| debug | 128 | ~500 GB | 2 (on GPU nodes) | 01:00:00 |


Node Access Policy | Single Job/Node ie (Multiple users can’t share resources across same node)
Node Access Policy | Multiple users' jobs can share the same compute node, allocated by CPU/memory rather than whole nodes.
--- | ---

## Hopper Configuration

| Partition | Max CPUs/User | Max Memory/User | Walltime |
| --- | --- | --- | --- |
| general (default) | 128 | ~380 GB | 2-00:00:00 |
| debug | 8 | 25 GB | 04:00:00 |

## Wheeler Configuration


| Queue: | Default | Default | Debug | Debug |
|----------------------:|:----------:|:----------:|:----------:|:----------:|
| parameter | soft limit | hard limit | soft limit | hard limit |
| Number of Processors | 160 | 400 | 32 | 32 |
| Number of Nodes | 20 | 50 | 4 | 4 |
| Processors per Node | 8 | 8 | 8 | 8 |
| Walltime | 48:00:00 | 48:00:00 | 01:00:00 | 01:00:00 |

Node Access Policy | Multi Job/Node ie (Multiple users may share resources across same node)
Node Access Policy | Multiple users' jobs can share the same compute node, allocated by CPU/memory rather than whole nodes.
--- | ---



## Gibbs Configuration

Parameters | Soft Limit | Hardlimit
--- | --- | ---
Number of Processors | 96 | 96
Number of Nodes |6 | 6
Processors per Node | 16 | 16
Walltime | 72:00:00 | 72:00:00

Node Access Policy | Single Job/Node ie (Multiple users can’t share resources across same node)
--- | ---
*This quickbyte was validated on 7/30/2026*