Introduction to Computing Time
The computing resources provided by the CLAIX system are valuable and limited. To ensure optimal utilization, resources are primarily granted through computing time projects.
Table of Content
Accounting of Computing Time
In the HPC world, the consumption of computing resources is measured in core hours (core-h). As the name suggests, one core-h typically represents the usage of one CPU core for one hour. The use of a GPU for one hour (GPU-h) is billed as 24 core-h.
Every time a job is submitted to Slurm (except for the devel partition) the used computing time is accounted for.
The accounting of computing time can be monitored with r_wlm_usage.
Computing time projects enable the fair distribution of computing resources. Projects ensure that resources are not overbooked and that resource needs are justified.
Although small-scale tests can be performed without computing time project (see below) - any substantial work will require you to apply for one. The most suitable project depends on several factors:
- Your affiliation: Are you a member of RWTH, a different German university or research facility, or from industry?
- The type resources you need: CPU or GPU?
- The amount of computing resources you need, i.e how many core-h.
- The purpose of the computing time: Is it for research or teaching?
Depending on the type of computing time project you apply for, your proposal may undergo a technical and scientific review. For more information review our Quick Reference for an illustration of the application, approval and allocation process.
After allocation, the project utilization is managed by the person of contact (PC).
All users are provided with a default monthly computing quota that can be used outside of computing time projects. These quotas are as follows:
- Students: 500 core-h
- Non-students (e.g. scientific employees): 2000 core-h
If you require computing time beyond these default quotas, you will need to apply for a computing time project.
In addition, the login nodes and the devel partition can also be used for testing purposes without the need for a computing time project.