200 series upgrade
Node 200 had become a bit unstable over time so we reinstalled it. Same hardware, just upgraded to SL5.8. We also replaced nodes 216 and 217 with blades with 8GB of RAM to make the 200 series architecture homogeneous. After…
Cape Town power failure
Thanks to our data center UPS and generator our cluster remained up during today's city wide power failure. No jobs lost or data compromised.
…
…SEQUOIA SUPERCOMPUTER, Lawrence Livermore National Laboratory
A Blue Gene/Q supercomputer built on IBM Power architecture at the
Department of Energy’s Lawrence Livermore National Laboratory has taken pole position in the Top500.org of supercomputers in the world. The "SEQUOIA SUPERCOMPUTER " has a staggering 1.6 million cpu…
Intel releases new coprocessor aimed at HPC
Intel released its Intel Phi coprocessor yesterday which is specifically aimed at highly parallel applications. The processor boasts a whopping 1 teraflop of compute power contained on a single PCI-E bus. Petascale and Exascale computing is definitely around the …
New worker nodes
We've added two new nodes, 213 and 214, to the 200 series. Both nodes have 8GB of RAM. This brings our cluster up to 136 cores.…
RAM upgrade and more worker nodes
The RAM of worker node 209 has been upgraded from 4 to 8GB and the node moved back to the 200 series. We will be deploying a few more 200 series and at least 1 more 400M series worker nodes…
Surviving the power outage
Yesterday's Peninsula wide blackout fortunately had no effect on our HPC and Grid systems. Both data center generators kept their UPS's running until power was restored after about an hour.
As can be seen from the monitoring dashboard all devices…
Memory upgrade complete
The cluster memory upgrade seems to have gone OK, /scratch01 is remounted. Server 209 is still at 4GB as we only budgeted for 9 memory kits. Once we can confirm that the extra 4GB RAM
has made a difference for…
RAM upgrade
THE ICTS-HPC team will be upgrading the
RAM in several of the 200 series servers between 10 and 12 on Friday the
16th of September 2011. The implication is that the /scratch01 disk
area will be unavailable at this time.
…