Highly parallel HEVC decoding for heterogeneous systems with CPU and GPU
作者：
Highlights：
•
• A highly parallel HEVC decoder for CPU+GPU systems is proposed.
• Multiple levels of parallelism are exploited on CPU and GPU simultaneously.
• Compared to state of the art, 167 fps (a 2.2× speedup) is achieved for 4K videos.
• An energy efficiency analysis for CPU-only and CPU+GPU decoding is performed.
• In CPU+GPU decoding, tuning GPU clocks can lead to an energy reduction up to 36%.
摘要
•A highly parallel HEVC decoder for CPU+GPU systems is proposed.•Multiple levels of parallelism are exploited on CPU and GPU simultaneously.•Compared to state of the art, 167 fps (a 2.2× speedup) is achieved for 4K videos.•An energy efficiency analysis for CPU-only and CPU+GPU decoding is performed.•In CPU+GPU decoding, tuning GPU clocks can lead to an energy reduction up to 36%.
论文关键词：
论文评审过程：Received 2 August 2017, Revised 16 December 2017, Accepted 21 December 2017, Available online 30 December 2017, Version of Record 8 January 2018.
论文官网地址：https://doi.org/10.1016/j.image.2017.12.009
