The advances in parallel computing architectures, such as many-core systems, are revolutionizing the computing industry. Such radical changing in computing has allured computer scientists by a series of challenges, spanning from software engineering solutions to computer theory development (e.g. algorithm complexity analysis). In computing application research, such as computer graphics, the biggest concern is how to revisit existing sequential algorithms and develop optimized (or brand new) algorithms to achieve better performance on the parallel computing architecture. It's not a simple implementation issue, since the theoretical foundations of sequential and parallel computing are fundamentally different. In this presentation, I will discuss several challenges in parallel algorithms development for many-core systems, including algorithm design paradigm, distributed load balancing, concurrent memory management, and pipelining for tasking parallel processing. I will show some of our recent efforts, mostly in computer graphics, that attempt to address these challenges.
Yong Cao is an Assistant Professor of Computer Science at Virginia Tech. He is the director of Animation and Gaming Research Lab. His research interests include computer graphics, high performance computing and video game based learning. He is currently involved in several projects on data intensive visualization and simulation, In-Situ visualization, general purpose computing with GPUs, and game-based learning/traing. He serves his research community by participating in several conference communities, including EuroGraphics, Siggraph Asia Sketch, and IEEE Conference on Automatic Face and Gesture Recognition. He also serves in the editoral board of ACM Computer in Entertainment.