Tags:
create new tag
, view all tags

Jan 2012 News (updated Oct 2012)

  • My visit w/ DJ and GVD at Herndon in December 2011 was very successful.
  • The cluster will be upgraded, s/w and h/w, as follow:

What Details When
+ upgrade NetApp filer from FAS3140 to FAS3240-R5 done (2/14/12)
+ additional NetApp scratch disk space scratch space doubled from 20TB to 40TB 30TB done so far (3/16/12)
+ upgrade of ROCKS from v5.3 to v5.4.3 done (3/16/12)
+ upgrade the head node from a 4 cores to a 24 cores one done (2 login nodes, 1 front-end node; 3/16/12)
+ changes in queues and limits stay tuned done (3/16/12)
+ add extra memory to nodes to have 2GB/cores on each compute node done (not all nodes; h/w config. limitations)
+ add 10 R715 nodes, each 24 cores augment compute cores from 2420 to 2752 done (3/16/12)
+ add an InfiniBand fabric to part of the cluster 3 x 36 = 108 IB ports 2 switches and 55 nodes installed, for 844 cores (Jul/12)
Expect to extend it to ~70 nodes for ~1,500 cores
+ add 10 R815 nodes, each 64 cores, 4GB/core upgrade rack#0, replacing 32n x 4c by 10n x 64c, on IB
augment total # of cores 2648 to 3196
expected in Oct/12
+ deploy a parallel file system (pvfs2) to augment scratch space and I/O performances proof-of-concept deployed for testing (1/10/12)

I will keep updating this table as the upgrades progress.

  • The set of primers on how to use the cluster are now available,
    and have been re-organized to consolidate them and avoid duplication.
  • The compilers have been upgraded to support the IB fabric.
  • The queues have been reconfigured to simplify them and to support the IB fabric.

-- SylvainKorzennikHPCAnalyst - 19 Jan 2012

Topic revision: r5 - 2012-10-03 - SylvainKorzennikHPCAnalyst
 
This site is powered by the TWiki collaboration platform Powered by PerlCopyright © 2008-2015 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback