Evaluation of Imprecise Subtractors into Test Zone Search for VVC Encoding
Rafael Ferreira, Luciano Volcan Agostini, Cláudio Diniz, Bruno Zatt · 2023
The error-tolerant behavior of some video encoding algorithms has motivated researchers to exploit approximate computing in the seek for energy-efficient coding systems. If approximate adders have been widely evaluated, approximate subtractors did not receive much attention. Subtractors, however, are extensively used during the encoding process and optimizing them represents an opportunity for further improving energy efficiency. This study explores the adoption of approximate subtractors in Versatile Video Coding (VVC) encoders, with a specific focus on the distortion metric used in the Test Zone Search (TZS) algorithm for inter-picture prediction. The detailed analysis reveals the impact of imprecise subtractors on each stage of TZS, presenting unprecedented data about the number of calls and energy consumption per stage. The results show that the use of approximate subtractors leads to changes in the number of calls for each TZS stage, induces a minimal impact in terms of coding efficiency, and enables nearly 6% reduction in energy consumption. Therefore, this work provides substantial contributions to the understanding and to the optimization of video coding processes, energy-constrained scenarios.