hpc

This evening our cluster achieved another small mile stone, 100,000 active computational hours, or put another way, all the current jobs running at the moment have consumed about 11.4 years of computation. ... and no, that's not 36 hours in…

ICTS HPC cluster downtime: The ICTS HPC team will be upgrading various components of both clusters in early December.  We will be putting the clusters into draining mode on Friday the 6th of December.  All worker nodes as well…

ICTS HPC is very pleased to have played a small role in the development of SKA Squeeze, a data compression algorithm for the SKA radio telescope.  Use was made of the GPGPU servers in order to achieve the speeds…

Several sysadmins met today at UWC for the first HPC forum meeting.  Represented were UWC, UCT, CHPC, SANBI and the University of Stellenbosch.  There was discussion on common schedulers, file systems and supporting technologies.  Also several topics for further investigation…

An intersting article on parallel tools by Jeff Layton.  While there is a pssh utility for various transfer routines it's no longer being maintained.  Other parallel versions of rsync seem to be script based:  https://gist.github.com/rcoup/5358786 and http://blog.ciberterminal.net/2012/10/16/parallel-rsyncing-a-huge-directory-tree/…

What's up with the yellow warning lights on the cluster? These nodes have been placed on hold and the reason for us doing this is so that we can 'drain' them of jobs prior to upgrading and patching.  We prefer…

We encountered an issue where a user's jobs used far more cpus than allocated.  This occured even though ppn was set to 1, the server load went up to 400+ and the individual processes were using up to 4000% CPU…

…

…

…