SURFsara's Cartesius Cluster: Overview
Cartesius is a supercomputer in the true sense of the word. All servers in the cluster are linked with high-speed connections to form a very large computer. Cartesius is best suited for calculations in which the individual components (functions and procedures) have a great interdependence. A so-called Graphical Processing Unit (GPU) is also offered in this environment. In some cases, for example with Floating Point calculations, running a program on a Graphics adapter is much more efficient than on a computer processor.
Pros
- The most rapid solution.
- Graphical Processing Unit (GPU) available.
- A huge amount of software.
- Users do not have to share resources but have exclusive access.
Cons
- Steep learning curve. Knowledge is required to be able to build an efficient job.
- No direct access to the joint network of Tilburg University.
- There may be a queue and thus waiting time.
- Less interactive because of the queue principle.
How to access SURFsara's Cartesius cluster
You can request access via a 'small compute grant' at SURF here. Applications are light-weight, and are handled within 2 weeks. Contrary to the name, the grant will award a decent amount of hours on either the HPC Cloud, a linux cluster, or the supercomputer. Larger grants are available via a more involved procedure. Support at TiU is available at digital research support, reachable via the self service portal.