Resource Management System Running
March 4--Users of NPACI Cray and IBM high-performance computing platforms can better track the usage of their allocations with the recent port by SDSC's HPC Systems group of the Resources Management System (RES) to the IBM SP platform.
RES works with the computer's accounting system to charge the running time of applications against the appropriate allocation. On the CRAY T90, RES can checkpoint jobs when an allocation becomes overdrawn. On the CRAY T3E and IBM SP, RES prevents users with overdrawn allocations from starting new batch jobs.
With the port to the SP platform, RES is now available on all NPACI systems at SDSC and the IBM SP at the University of Michigan. There are plans to implement RES on the NPACI Cray systems at the University of Texas.
RES also allows project leaders to manage the time available to multiple users working on the same project. For example, a project leader can specify that a particular user can "spend" 25 percent of an account's allocation.
"Porting RES to the Cray and IBM systems represents a milestone in linking account management across all NPACI resource systems," said Mike Vildibill, SDSC associate director of computing resources. "We are now working on porting RES to the HP Exemplar at Caltech, NOW at UC Berkeley, and the Tera MTA at SDSC."