GPU Accelerated H.264 Video Compression for Broadcast

In recent years the ability to perform massive parallel computations has become readily available on every desktop due to improved graphic processors and new programming tools like CUDA. The standard CPUs are also evolving rapidly in the same direction, making this technology even more accessible. Video compression is an illustrative example of an area where the speed / computation power trade off is especially steep. To render all the features of the modern H.264 compression standard a super-computing level of power might be necessary. The modern GPU and the next generation of the CPU could deliver the required computation power. We investigate what can be done to exploit the features of the modern GPUs for H.264 compression and how the new video compression standards like H.265 might be adopted in the future of massive parallel processing

Published

2010-10

Content type

Original Research

Keywords

Video encoding, H.264, H.265, MPEG-2, GPU, massive parallel computing, CUDA, OpenCL, codec

DOI

10.5594/M001388

ISBN

978-1-61482-944-7