Profiling of HEVC encoder

Abstract

One of the serious challenges facing smartphones and other portable devices nowadays is battery life. A typical smartphone user will be running several applications on the device at any given time, and most of these applications consume plenty of energy. In fact, some of the most energy-consuming applications are those related to video conferencing and streaming. There is a lot of work to improve the energy efficiency of such applications by improving the energy efficiency of the underlying encoders. However, to accomplish that, full profiling of the encoder must be performed first. The results of profiling the High Efficiency Video Coding (HEVC) are presented. Results are presented in terms of types of assembly level instructions in every function of the encoder. Thus, depending on the environment on which the encoder is running, the energy consumed for any type of instruction can be obtained. The results show that, in an HEVC encoding process, about 48% of the instructions are memory related and about 18-20% are arithmetic operations. This result is very important for any work in energy analysis and optimisation of the HEVC encoder © 2014 The Institution of Engineering and Technology.