The HPC DCCN Cluster and Linux
This wiki page is here to try to provide some concise but useful information on how to navigate the HPC DCCN cluster. It makes use of the existing resources at the DCCN (by providing links to them) in addition to experiences from lab members and workshops given by the TG. In contrast with the HPC DCCN Wiki, this document should be more digestible and accessible to researchers with different backgrounds.
Nonetheless, we will list resources that will be helpful when trying to work with the cluster:
- HPC Cluster Wiki – this contains all the relevant information, according to the TG, that you will need to use the cluster.
- Intranet HPC – this contains information relevant to the HPC at the DCCN
- Recordings of the HPC Tutorials Given by the TG – if you do not feel confident in doing things by simply reading documents. The TG organizes every six months interactive tutorials on how to use the cluster. If you do not want to wait, the link provided contains recordings from previous years. Note: to access the link you need to be on a DCCN network. Given the date of the recordings, some of the information there might be outdated.
Most of the time, scientific computations require the usage of a lot of data and computing power. A major goal for any scientist is to be able to carry out these processes as fast as possible. A way this can be achieved is through parallel computations: using multiple computers at the same time to work on your data. The DCCN has a High-Performance Cluster (HPC) which allows researchers to do exactly this. This wiki page is dedicated to provide you with information about the cluster and how to use it effectively.
The DCCN is transitioning on the software used for the cluster, so this page will contain information on both the current way of using it (Torque) and the future software manager (SLURM).
Contact
If you run into any problems, you can contact the TG help desk. They are always available to help with technical questions.
Every year, a PhD student takes on the PhD job to help researchers at the DCCN with any cluster problems they might have. Especifically, anything related to running specific software used for neuroimaging analyses on the cluster. The current person responsible for this is Lennart Oblong (Room 02.284), but always double check as these jobs are constantly changing.