Spring 2007 Update

Over the last six months several upgrades and additions have taken place at the HPCC.  I wanted to take the opportunity to update the user community on these activities.  We have significantly increased the storage, compute-power, and back-up capabilities of the HPCC.  The recent upgrades have raised our total number of 64-bit cores to 1000 up from 432 just one year ago. 

First we have implemented a parallel file system (Lustre) and added 12 Terabytes of parallel storage to Hrothgar.   This replaces our old NFS storage that was both slow and prone to failure.  In addition we have added 20 terabytes of back-up disk space.   This means that we are backing up Hrothgar and Antaeus on a 2 week cycle.  Should you have files that cannot be recreated through computation (or other means), we recommend that you preserve copies on you local machines.  We may not be able to retrieve files that are older than 2 weeks and have been changed in the last 14 days. 

Second we have upgraded 64 Hrothgar nodes from dual CPUs to eight CPUs increasing the number of CPUs on Hrothgar from 384 CPUs to 768 CPUs.  The new nodes have 12 GB of RAM each and can be used as eight-way shared memory machines.  For reference, each of the 64 eight-way nodes is more powerful than 3 Pleiones (our now retired 56 processor SGI machine). 

Antaeus is a new community cluster, just purchased and installed last fall.  It has 192 processors purchased by research groups in Physics, Chemistry, and by the HPCC.  This machine also has significant (Lustre) storage and is primarily used for the various grid computing projects that are underway at TTU, including OSG, LHC, SURAgrid, and TIGRE. 

Finally, we are in the process of upgrading our campus grid.  We are moving to the Condor technology from the AVAKI technology.  Condor has made great strides over the last few years and AVAKI has not kept up.  It may well be that this incarnation of TechGrid will surpass 1000 nodes in the next few months.

My thanks to the HPCC staff for making these transitions as painless as possible, and also for keeping the users? needs at the forefront of their thoughts.  We make every effort to stay on top of the technologies that are necessary for your research computing.  James Abbott has been instrumental in these updates, with assistance from David Chaffin and Srirangam Addepalli.  Jerry Perez is implementing the Condor grid.  Alan Sill and Ravi Vadapalli are working tirelessly on TIGRE.