Production Supercomputing

	The graph shows the amount of sustained computing capacity (in gigaflops) available to the NCAR community over time. Note that substantial increases are provided every 16 months on average. CISL provides this growing resource in response to the requirements of the scientific community. View large image.
	This chart shows the average utilization of the production computing platforms provided by CISL. Utilization is defined as the percent of compute nodes assigned to a user at any one time. Blackforest was an IBM POWER3 system that was decommissioned in January 2005. Bluesky is the IBM POWER4, and bluevista is the IBM POWER5. Lightning is a small Linux cluster available for production computing. Through community-directed project allocations and efficient job scheduling, these powerful resources are distributed as equitably as possible to all users. View large image.

The hallmark of supercomputing at NCAR has been the provision of robust, reliable, effective, and efficient production computing, and of state-of-the-art storage, data analysis, data visualization services and tools for its community of users. To fulfill such a mission, CISL provides an environment that gives equitable access to all users, yet strives to maximize resource utilization. These goals are achieved by balancing the allocation and scheduling of resources with a properly designed tactical queue structure and single-job resource limits.

In addition, CISL attempts to increase the computational capacity available to the community on a regular basis. The acquisition and deployment during FY2006 of the IBM POWER5 system bluevista is an example of one such increase. Peripheral resources that complement and supplement the high-end computing environment are upgraded and enhanced as appropriate to match the growth in compute capacity. Capability and capacity will increase significantly with the ICESS procurement in early FY2007.

The resources in this facility are provided in support of the NCAR strategic priority "Enhancing capability and capacity of NCAR supercomputing." NCAR's supercomputers are managed by CISL under the UCAR/NSF Cooperative Agreement and are supported by NSF Core funds including CSL funding.