Partitions and Job Policies
This page lists job partitions, job limits, and QOS considerations for submitting jobs to the Hydro cluster.
Hydro Partitions (Queues)
Partition (Queue) |
Node/Job Type |
Max Nodes per Job |
Max Duration |
Max Running in Queue/user |
Charge Factor |
|---|---|---|---|---|---|
sandybridge |
CPU (Intel) |
TBD |
7 days |
TBD |
1.0 |
sandybridge2.9 |
CPU (Intel) |
TBD |
7 days |
TBD |
1.0 |
sandybridge2.0 |
CPU (Intel) |
TBD |
7 days |
TBD |
1.0 |
interlagos |
CPU (AMD) |
TBD |
7 days |
TBD |
1.0 |
milan |
CPU (AMD) |
TBD |
7 days |
TBD |
6.0 |
rome |
CPU (AMD) |
TBD |
7 days |
TBD |
6.0 |
a100 |
dual A100 GPU w/ any CPU |
TBD |
7 days |
TBD |
20.0 |
a100milan |
dual A100 GPU w/ Milan CPU |
TBD |
7 days |
TBD |
20.0 |
a100rome |
dual A100 GPU w/ Rome CPU |
TBD |
7 days |
TBD |
20.0 |
Sandybridge nodes have 16 cores per node, dual-socket, 384MB (2.9 and 2.0 GHz).
Interlagos nodes have 64 cores per node, quad-socket, 512MB.
Milan nodes have 56 cores per node, dual socket, 256MB.
Rome nodes have 64 cores per node, dual socket, 256MB.
sview View of Slurm Partitions
Slurm’s sview command provides a graphical view into a resource’s partition and node properties.
Node Policies
Node-sharing is not enabled; jobs currently run exclusive.
When node-sharing with jobs is enabled, node-exclusive mode can be obtained by specifying all the consumable resources for that node type or adding the following Slurm options:
--exclusive --mem=0
GPU NVIDIA MIG (GPU slicing) for the A100 and preemptive jobs will be supported at future dates.
Project and Job Accounting
There are two available commands for tracking usage depending on the project’s charging policy.
For projects that are charged monthly and do not have a specific allocated award amount, such as NGA projects, use the following command to show balance information:
accountsFor projects that are allocated a fixed amount, such as Illinois Computes projects, use the following command to show balance and deposit info:
accounts-remaining
Additional information on job use can be found using the script jobcharge_grp.py. The script requires the following arguments:
jobcharge_grp.py [-h] [-detail] accountstring daysback