This is the homepage for the supercomputing facility at the University of Southern Denmark (DCSC/SDU) run on behalf of the Danish Centre for Scientific Computing.
- Today, after serving us well in 5 years and 25 days (96222 jobs), Horseshoe5 has been powered down.
The hardware will be dismantled and removed, to make room for our new Horseshoe9 cluster.
- The express node on fe7 has been upgraded to CUDA 5.5.
- 9:00-10:00: A failure in our virtual environment has caused downtime on most of our infrastructure servers.
Clusternodes and fileservers has not been interruptet by this error, but queue management has been down.
All systems are now back to normal opration.
- New versions of several modules available gromacs/4.6.5 numpy/1.8.0, etc.
for further information.
- 15:50-18:30: All nodes including fe5-fe8 are down due to a major power outage.
- We are currently experiencing degraded performance on disk4. disk4 is
currently rebuilding due to disk failure. The rebuild is now done.