Blue Gene/L system operations
The IBM Blue Gene/L computer at NCAR
(right) is designed for high performance and low power consumption. CISL
and its collaborators performed significant R&D to make it a useful tool
for geosciences research. Frost is now operating at a high level in multiple
roles for universities, NCAR, and the TeraGrid.
The IBM Blue Gene/L (BG/L) supercomputing system named frost consists of a
single BG/L rack containing 2,048 compute processors for 5.73 teraflops peak
performance. This low-power system (25 KW) is managed by CISL's Research
Systems Evaluation Team (ReSET) and is now a production supercomputer on the
TeraGrid.
Frost was purchased as an experimental system to support researchers from NCAR, the University of Colorado at Boulder, and the University of Colorado at Denver who are investigating and addressing the technical obstacles to achieving practical petascale computing in geoscience, aerospace engineering, and mathematical applications. The opportunity to experiment with systems like BG/L is absolutely essential for NCAR to maintain its ability to provide capability and capacity supercomputing to the community.
In FY2008, frost was enhanced with additional storage and networking equipment. This additional equipment tripled the storage capacity from 6 TB to 23.2 TB and is used to support special projects, such as the collection, distribution, and analysis of the results of high-resolution CCSM runs. Frost was also used as a platform for evaluating the performance of a variety of new networking and storage systems at scale. As part of the evaluation, frost was connected to several storage systems, including a 100 TB storage cluster and several commercial storage solutions, using both Infiniband and 10 Gigabit Ethernet networking.
Also in FY2008, ReSET developed software to support the new IBM BlueGene/L High-Throughput Computing mode, which is being used to provide the capability for processing grid workflows consisting of large numbers of small tasks. ReSET continues to collaborate with Argonne National Laboratory to further develop Cobalt, the queuing system currently used on frost, by incorporating alternate scheduling strategies and interfacing it with the Coordinated TeraGrid Software and Services software stack.
Frost continues to be used by a broad community of researchers. Frost is also used as one of the primary computational resources for the University of Colorado's High Performance Scientific Computing course, including 38 doctoral students, 19 graduate students, and 12 undergraduate students. Frost was also used as the IMAGE Summer School in Geophysical Turbulence.
Frost continues to be available as a TeraGrid resource. The system now has 13 active users that were created through the TeraGrid's central account request system. Frost will be maintained through FY2009 to support users at NCAR and the University of Colorado as well as the larger TeraGrid community.
As an advanced supercomputing tool, frost enhances NCAR's capability and capacity to "conduct computer science, computational science, and applied mathematics research and development." The acquisition and operation of frost was made possible through NSF MRI Grants CNS-0421498, CNS-0420873, and CNS-0420985; through the IBM Shared University Research (SUR) program with the University of Colorado; and NSF core funding.