About the Condor Grid at UH
Condor is a software framework for distributing a computing workload across a variety of platforms. It can schedule and run jobs on dedicated compute clusters as well as "scavange" resources from otherwise idle desktop computers.
Our goal is to deploy a grid that includes dedicated and scavanged cycles from servers at the RCC as well as scavanged cycles from desktop computers on the UH campus.
When operational, the grid will be available to all UH researchers, graduate and undergraduate students.
Deployment of the Condor grid is at a very early stage.
Suggested Readings
Condor Version 7.0.4 Manual
Wikipedia: the Free Encyclopedia "Condor High-Throughput Computing System" |
|
 HPC operates a 59 node Dell Linux cluster with a total of 189 32-bit Intel Xeon 2.4GHz processors, over 120GB of memory and about 10TB of storage. The Beowulf compute cluster is interconnected by dual gigabit ethernet networks, one dedicated to message passing, the other to file services.
The operating system is a current, supported version of Red Hat Enterprise Linux (RHEL) layered in the Rocks cluster framework. In addition to the development tools commonly available in a clustered Linux environment (including OpenMPI, MPICH and the GNU compilers), there is a choice of specialized complier suites from the Portland Group and from Intel. Batch job scheduling and processing is managed by TORQUE, a standard for HPC at research sites worldwide.
The Beowulf cluster is available to any researcher working at the University of Houston. Please visit the account request page to apply for accounts.
Through the job scheduler, a portion of the Beowulf cluster has been partitioned to give priority to the Air Quality Modelling Group.
|
|
|
|
|
|
|
RCC Resources