Stony Brook University Computer Science

SUNY Korea, September, 2015

The NVIDIA GPU Research Center at SUNY Korea is headed by Professor Klaus Mueller of SUNY Korea’s Computer Science Department. It supports a wide range of research and education activities that involve CUDA devices and technology. GPU-accelerated computing can tremendously accelerate scientific, analytics, engineering, consumer, and enterprise applications. While a conventional CPU consists of a few cores optimized for sequential serial processing, a GPU has a massively parallel architecture consisting of thousands of smaller, more efficient cores designed for handling multiple tasks simultaneously.

One application that uses GPU-technology at SUNY Korea is computed tomography (CT) for medical and industrial applications, such as X-ray and ultrasound imaging. Here the speed of GPUs enables the realization of highly compute-intensive iterative schemes for the reconstruction from data acquired in adverse conditions, such as low radiation dose, sparse acquisition, and beam hardening. GPU-accelerated computing also enables the accurate modeling and integration of the underlying imaging physics into the reconstruction process.

Another application at SUNY Korea is in the area of visual analytics and data science where GPUs support frameworks for interactive cluster analysis of big data in health informatics, climate science, finance, and economics. GPUs also accelerate parameter learning of optimization algorithms from example data where the high computational performance of GPUs enables an interactive exploration of the search space. A more recent effort is the design of effective visual interfaces to visualize runtime performance and aid in code optimization, in particular nested loop reordering and decomposition using the polyhedral method.

These and other research efforts have been widely published by members of the Center in both the domain literature and as CUDA Computing Gems. Almost all students of the Center have used, or are using, GPUs in their research for the purpose of making their software interactive or facilitating otherwise computationally infeasible research. Courses on high performance computing with GPUs are taught at SUNY Korea on a regular basis where students are able to also hone their skills on SUNY Korea’s high-end multi-GPU K20 and K40 Tesla server.

NVIDIA has been a pioneer in visual computing. With its invention of the GPU (Graphical Processing Unit) — the engine of modern visual computing — the field has expanded to encompass video games, movie production, product design, medical diagnosis, and scientific research. In 2007, NVIDIA introduced the application programing interface CUDA (Compute Unified Device Architecture) which has allowed software developers to use GPUs for general purpose parallel computing on the desktop. But also supercomputers such as the Titan at Oakridge National Labs which is currently the 2nd fastest supercomputer in the world are using GPUs on a massive scale to accelerate their computations.

With its GPU Research Centers and other academic programs NVIDIA seeks to advance parallel computing education and research.  Among the current NVIDIA GPU Research Centers are world-class institutions such as University of California at Santa Barbara, Johns Hopkins University, University of Southern California,  Duke University, and now also SUNY Korea.