Named after the famous Norwegian mathematician Niels Henrik Abel (1802 – 1829), the Linux cluster at UiO is a shared resource for research computing capable of 258 TFLOP/s theoretical peak performance. At the time of installation Abel reached position 96 on the Top500 list of the most powerfull systems in the world.
Abel is an allround, allpurpose cluster designed to handle multiple concurrent jobs / users with varying requirements. Instead of massively parallell applications, the primary application profile is for moderately to smaller parallel applications with high IO and/or memory demand. Frequent users of Abel are from the life sciences, astrophysics, geophysics, chemistry and also humanities. All compute nodes on the cluster have a minimum 64 GB RAM, 16 physical CPU cores and are connected by FDR (56 Gbps) Infiniband.
Short stop of queue system Wednesday 27/07 at 09:00
Jul 25, 2016 01:55 PM
The queue system on Abel will be stopped for about 10 minutes at 09:00 on Wednesday (27/07), in order to perform maintenance on its accounting data base.
Running jobs should not be affected by the stop, but no jobs will start, and no slurm commands will work, during the stop.
Correction: The date is 27. July.
Problems with /work file system on Abel
Jun 21, 2016 03:25 PM
We are currently experiencing problems with the /work file system on Abel. We are working on identifying and fixing the problem, and will update as soon as we know more.
Update: After restarting the file system daemons yesterday, /work seems to work fine again. We are in contact with beegfs support to find out why the problem occured.
May 31, 2016 03:17 PM
R 3.3.0 has been installed, and will become the default version in about two weeks.
Along with the new version of R, a few packages have been installed or upgraded in order to build R:
- bzip2 1.0.6
- xz 5.2.2
- pcre 8.38 (upgrade)
Note that as of version 3.3.0, we no longer install the BioConductor packages globally on Abel. The reason is that these packages are upgraded quite often, so globallly installed versions will quickly become outdated, and upgrading them might break user code. Therefore it is better that users install the BioConductor packages themselves.