Million Dollar Upgrade to Ocelote Research Computer
Artificial intelligence and machine learning researchers at UA have a new resource for high-speed computation. The UA's new GPU (graphics processing unit) capabilities will also serve:
- molecular dynamics, aerospace and mechanical engineering, and geoscience researchers
- climate, planetary, and materials scientists
- videographers and 3D photographers
- any UA researcher needing high-speed processing capabilities
The Ocelote high performance computing system, located in the Computer Center's first floor Research Data Center, now has 46 new nodes, each with an Nvidia Tesla P100 GPU.
"This was a special upgrade," explains Jeremy Frumkin, UITS executive director for research technologies. "We normally do not expand systems between refresh cycles, but in working with our research community, we realized the need to provide the GPU capability much sooner than the next refresh in 2021."
The GPU expansion cost approximately $800,000. At the same time, UITS installed new CPUs in Ocelote, bringing the total upgrade investment to nearly $1,000,000. The CPUs were purchased by the Department of Astronomy via the Research Data Center "buy-in" program. By funding these processors, Astronomy will have priority access to them for Dr. Peter Behroozi's research on galaxy and black hole formation.
Research Data Center resources, including the GPUs in Ocelote, are available at no charge to all UA researchers on a first-available basis. Researchers who buy in to the system by purchasing additional nodes get priority access to those nodes, rather than first-available.
Learn more about using OpenACC to program the new GPUs with XSEDE's free workshop:
XSEDE HPC Workshop: GPU Programming Using OpenACC
Tuesday, March 6
Computer Center (1077 N. Highland Ave.)