Announcements
Magellan Queues End November 30
November 17, 2011 by Francesca Verdier | 0 Comments
Schedule and Status
- Mon, Nov 28
- westmere queue
- drain - 08:00 - DONE
- remove - 10:20 - DONE
- serial queue
- add westmere nodes - 10:40 - DONE
- westmere queue
- Tue, Nov 29
- mag_xlmem queue
- drain - 08:00 - DONE
- remove - 09:15 - DONE
- reg_xlmem queue
- create - 14:00 - DONE
- enable - 17:00
- mag_xlmem queue
- Wed, Nov 30
- no queue modifications
- Thu, Dec 1
- magellan queue
- drain - 08:00 - DONE
- remove - 09:30 - DONE
- mag_serial queue
- drain - 08:00 - DONE
- remove - 09:30 - DONE
- magellan queue
Fast turnaround in Carver westmere queue
October 6, 2011 by Katie Antypas | 0 Comments
There is fast job turnaround in the new westmere queue on Carver. Nodes in the westmere queue have 12 cores per node and 48 GB of memory.
New Serial Queue on Carver
September 21, 2011 by Katie Antypas | 0 Comments
NERSC has implemented a new serial queue on Carver. A serial job is one that only requires a single core. The advantage of using the serial queue for single core jobs is that a user will only be charged for one core, rather than for all the cores on the node (which is the case for jobs submitted to the regular queue).
New 1 Terabyte Memory Node
June 17, 2011 by Francesca Verdier | 0 Comments
A new "extra large" memory node that has four 8-core Nehalem EX processors (32 cores total) and 1TB memory has been deployed in the Carver/Magellan complex. See Memory Considerations on Carver.
Purging of /global/scratch
June 13, 2011 by Francesca Verdier | 0 Comments
As previously announced, we have begun purging old files from /global/scratch. Last week we were approaching 90% full on this file system, which can severely impact file system performance and reliability. We therefore began the purge process, which brought usage down to about 75%. The "newest" file removed during this process was from November, 2010 (about 6 months old).