IEEE OCCS Podcast

IEEE OCCS Podcast
Sep 28, 2009
Architectures and Programming from GPUs to the Cloud Parallel computing is readily available and will soon become ubiquitous in everything from mobile phones to laptops. Mainstream microprocessors are multi-core, while powerful graphics processing units (GPUs) are adding additional computation capacity, yet very little of this capability is used. The primary culprit is the complexity of parallel programming and the sometimes limited gain despite much effort. This talk will present the common parallel computing architectures most visible to consumers, including parallel CPUs, multiprocessors, and GPUs, along with benefits and pitfalls of each. In addition, parallel programming approaches, from threads to OpenMP to CUDA and OpenCL will be introduced and compared. Performance examples from the HIPerWall research project will demonstrate parallel computing benefits and hazards. Finally, the next phase of parallel computing, cloud computing, will be discussed along with an applicable programming model.
Bio: Dr. Stephen Jenks is the Director of the Scalable Parallel and Distributed Systems Laboratory (SPDS) and manager of the Calit2 Visualization Laboratory at UCI where he is an assistant professor in the Department of Electrical Engineering and Computer Science and holds a joint appointment in the Donald Bren School of Information and Computer Sciences. He is also Co-Founder of Hiperwall, Inc. which has commercialized very high-resolution tiled display wall technology on commodity hardware. Dr. Jenks received his BS degree from Carnegie Mellon University in 1987 and the MS and PhD degrees from the University of Southern California in 1992 and 2000, respectively. He designed advanced avionics and medical systems for 14 years at Northrop Grumman Corporation. He teaches courses in advanced system software, parallel computer architecture, and processor interfacing. His research interests include parallel and distributed computing, tiled display systems, and embedded systems. He currently leads the HIPerWall project, a 200 million pixel tiled display system. He is a member of ACM and the IEEE Computer Society.
Contact information:
Dept. of Electrical Engineering and Computer Science
The Henry Samueli School of Engineering
3215 Engineering Hall (or Engineering III)
Irvine, CA 92697-2625
949-824-9072
Ubiquitous Parallel Computing