high performance computing on graphics processing units: hgpu.org

With the rise of streaming media on the Internet and the YouTube revolution, the demand for online videos is costing companies a significant amount of bandwidth. To alleviate the resources needed for streaming media, video compression removes redundant information and minimizes the loss in quality experienced by a human audience. In response to the need of better compression for high definition video across many environments like the Internet, the ITU-T Video Coding Experts Group (VCEG) and the ISO/IEC Moving Picture Experts Group (MPEG) worked together to create a new modern standard called H.264. H.264 Advanced Video Coding is the current standard in video compression. H.264 is a powerful yet flexible approach to video encoding across many platforms and applications. H.264 encoders take the digital frames of a video and reduce the file size by applying predictions, transforms, quantization, and bitstream encoding. Inter and intra-prediction are the two prediction modes that allow H.264 to achieve high performance in compression by reducing redundancies across temporal and spatial domains. The advances techniques used in H.264 will provide a strong foundation for future video encoder standards. The author investigated speeding up motion estimation in a H.264 software implementation using multicore graphics processor (GPU) and optimizing the GPU code for different types of memory. The overhead of memory transfers to the GPU proved to be slower than the original CPU-based version.