The table below presents various partition configurations along with their respective limitations. Note that additional constraints (*), not listed here, also apply. If the limitations presented here pose a problem for your use case, we invite you to contact our support team at cbrz-support@hsu-hh.de.
| Partition Name | Nodes per Job | Wall-clock Limit | Concurent Jobs Limit | Nodes | Remarks |
|---|---|---|---|---|---|
dev | 1-2 | 1h | 1 | 1-571 | For testing purposes only, max. two queued jobs |
small | 1-5 | 72h | - (*) | 3-571 | Regular nodes, exclusive node reservation |
small_shared | 1-5 | 72h | - (*) | 3-571 | Same settings as small but node resources are by default shared |
small_fat | 1-5 | 24h | - (*) | 572-576 | Fat memory nodes, exclusive node reservation |
small_gpu | 1-5 | 24h | - (*) | gpu 1-5 | Up to two GPUs can be allocated per node |
medium | 6-256 | 24h | 1 | 3-571 | Will be removed in January 2025 |
medium-short | 6-16 | 12h | 20 | 3-571 | Regular nodes, exclusive node reservation |
medium-s | 6-32 | 24h | 5 | 3-571 | Regular nodes, exclusive node reservation |
medium-m | 33-64 | 24h | 3 | 3-571 | Regular nodes, exclusive node reservation |
medium-l | 65-256 | 24h | 1 | 3-571 | Regular nodes, exclusive node reservation |
large | >256 | 24h | - (*) | 3-571 | Regular nodes, exclusive node reservation ! available to selected users only ! |
(*): HSUper has a general maximum of concurrent jobs per user, which can be found here. Other limitations are specified there as well.
The ‘Concurrent Jobs Limit’ column specifies the maximum number of concurrent jobs allowed per user within the partition by default. If no value is given, the general HSUper-wide setting restricting the maximum amount of jobs per user that may run at any given time concurrently applies.
Note: Also a limit for the amount of submitted jobs by a single user exists to prevent slow job scheduling.
Both constraints are described on the limitations page page. You can circumvent some of those limitations by using a different Quality of Service (see below) than the default one.
HSUper offers different Quality of Services (QOS) that impose varying limitations. Each QOS may loosen some restrictions while enforcing others.
preempt QOS)You can use the preempt QOS (e.g., #SBATCH --qos=preempt) to run up to 1000 jobs concurrently. However, be aware that your job may be preempted (canceled) within 30 seconds if a higher-priority job is queued.
To minimize the impact of preemption:
The many-jobs-small_shared QOS removes partition limits for concurrent jobs per user while enforcing a maximum of 15 nodes in use per user at any given time.
Accessing this QoS: Contact our support team at cbrz-support@hsu-hh.de to request access.