Optimizing Emerging Multimedia Algorithms on Multi-core Platforms   

Yen-Kuang Chen

Principal research scientist, Intel Deputy Director, Intel-NTU Connected Contextual Computing Center

Time: August 22 (Monday) 10:40 - 11:40

Multi-core processors are prevalent everywhere in desktops, graphics processors, laptops, and embedded systems. Many novel multimedia applications can be enabled by multi-core processors. More importantly, multimedia system designers must understand the nuances of a multi-core computing engine, only then can the tremendous computing power that such platforms provide be harnessed efficiently. While some people think that programming GPU's is sufficient for the future of multi-core programming, this talk will show some of the common pitfalls in GPU performance claims. Furthermore, this talk will show the real challenge in the future multi-core programming relies heavily on memory bandwidth. The winning multi-core architecture must have the most energy-efficient memory hierarchy. Therefore, it is critical to parallelize emerging applications, e.g., image processing, computer vision, and video codec, on multi-core processors with multi-level caches to take advantage of the processing power of multi-core processors.