No abstract

No abstract

No abstract

Abstract-The x265 video encoder aims at improving the speed and the computational efficiency of HEVC encoders implementation. In this paper we present a detailed energy consumption analysis, considering the consumption components of CPU, cache memories and main memory, for all x265 presets executing in a multicore system. Ten HD 1080p test video sequences with different motion and brightness characteristics are used in the experiments. Three tools are used to obtain the results: CACTI, PCM and Perf. To get more reliable time/energy results, 10 executions were performed for each preset. The results show that fast presets are 47× faster than slower presets. However, slower presets use robust configurations and achieve large reductions in bitrate. Due to this, the ultrafast preset has a bitrate 45% higher than placebo preset. Furthermore, the system energy consumption increases 45×, from ultrafast preset to placebo preset. Our experiments clearly present the dependence between bitrate and energy consumption for all encoding presets, which allows us to choose the best bitrate/energy trade-off for each platform at hand.

The digital video coding process imposes severe pressure on memory traffic, leading to considerable power consumption related to frequent DRAM accesses. External off-chip memory demand needs to be minimized by clever architecture/ algorithm co-design, thus saving energy and extending battery lifetime during video encoding. To exploit temporal redundancies among neighboring frames, the motion estimation (ME) algorithm searches for good matching between the current block and blocks within reference frames stored in external memory. To save energy during ME, this work performs memory accesses distribution analysis of the test zone search (TZS) ME algorithm and, based on this analysis, proposes both a multi-sector scratchpad memory design and dynamic management for the TZS memory access. Our dynamic memory management, called neighbor management, reduces both static consumption-by employing sectorlevel power gating-and dynamic consumption-by reducing the number of accesses for ME execution. Additionally, our dynamic management was integrated with two previously proposed solutions: a hardware reference frame compressor and the Level C data reuse scheme (using a scratchpad memory). This system achieves a memory energy consumption savings of 99.8% and, when compared to the baseline solution composed of a reference frame compressor and data reuse scheme, the memory energy consumption was reduced by 44.1% at a cost of just 0.35% loss in coding efficiency, on average. When compared with related works, our system presents better memory bandwidth/energy savings and coding efficiency results.

No abstract

No abstract

No abstract