U recognized for computer science, receives $800,000 grant

By By Keith Chalmers, Staff Writer

By Keith Chalmers, Staff Writer

Thanks to a new computer processing system, computer scientists at the U will be able to speed up the process of making complicated graphic illustrations and computations that usually take days.

One of the largest manufacturers of graphic processing units in the world recently named the U a Center for Excellence and is donating an $800,000 processing unit to help with research.

Compute Unified Device Architecture, a computer programming language for GPUs, recognized the U for linked computer programs and the computer science curriculum that teaches CUDA’s computer programming language to students. The company makes programming cards that are useful to the scientific community as well as the video game industry.

Chris Johnson, director of the U’s Scientific Computing and Imaging Institute, hopes that CUDA will help researchers work on projects varying from seismology to Alzheimer’s disease.

Steven Parker, principal research scientist at NVIDIA’s, inventor of the GPU, Salt Lake City center and former U professor, said the computer processing system will be among the 500 fastest computers in the world.

“This machine can get results in minutes that used to require days of computation, or in days that used to take years,” he said.

The scientific community has struggled for years to create programs that could help with research using GPUs instead of central processing units.

Johnson said early graphics cards were not set up to do scientific computations, but give thanks to CUDA’s programming language for the new technology.

Andrew Humber, a spokesman for NVIDIA, said they chose the U as a Center for Excellence due to its commitment to teaching CUDA’s programming language as a part of its curriculum and incorporating the language into their research programs.

Humber said the University of Illinois is the only other university with such a large GPU system at this time.

“But recently, NVIDIA created this new programming language called CUDA which makes it much easier to program on the graphics card,” Johnson said. “So now it is much easier for scientists to take their programs that … run on CPUs and make them run on GPUs.”

Abe Stephens, a U doctoral student in computer science, said he uses the programming language for scientific computing and visualization every day.

“GPUs are designed to draw pictures very quickly,” Stephens said. “CUDA exposes several mechanisms inside the GPU that are useful for solving problems outside of graphics.”

[email protected]