Configurable Power/Quality-Aware Hardware Design for the AV1 Directional Intra Frame Prediction

Luiz Neto, M. Corrêa, Bruno Zatt Daniel Palomino, L. Agostini, G. Corrêa
{"title":"Configurable Power/Quality-Aware Hardware Design for the AV1 Directional Intra Frame Prediction","authors":"Luiz Neto, M. Corrêa, Bruno Zatt Daniel Palomino, L. Agostini, G. Corrêa","doi":"10.1109/SBCCI53441.2021.9529997","DOIUrl":null,"url":null,"abstract":"AOMedia Video 1 (AV1) is a royalty-free video format launched in 2018 that includes several tools to achieve high coding efficiency, but also presents a high computational cost. Because of this, hardware solutions are a good alternative to speed up the encoding process and enable real-time processing. This work presents a configurable hardware design for the AV1 directional intra-frame prediction. Two operation modes were defined on a software-based analysis and implemented in the architecture: the High Quality mode, which offers high coding efficiency, and the Low Power mode, which offers a power dissipation 84.4% lower in comparison to the first mode at the cost of a compression efficiency loss of 2.48%. The proposed architecture is capable of operating in real-time for video resolutions of up to 4K@60fps. When synthesized for the TSMC 40nm technology and operating at a target frequency of 1,902 MHz for 4K@60fps videos, the High Quality and the Low Power modes presented a power dissipation of 631.76 mW and 82.76 mW, respectively.","PeriodicalId":270661,"journal":{"name":"2021 34th SBC/SBMicro/IEEE/ACM Symposium on Integrated Circuits and Systems Design (SBCCI)","volume":"579 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2021-08-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2021 34th SBC/SBMicro/IEEE/ACM Symposium on Integrated Circuits and Systems Design (SBCCI)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/SBCCI53441.2021.9529997","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 1

Abstract

AOMedia Video 1 (AV1) is a royalty-free video format launched in 2018 that includes several tools to achieve high coding efficiency, but also presents a high computational cost. Because of this, hardware solutions are a good alternative to speed up the encoding process and enable real-time processing. This work presents a configurable hardware design for the AV1 directional intra-frame prediction. Two operation modes were defined on a software-based analysis and implemented in the architecture: the High Quality mode, which offers high coding efficiency, and the Low Power mode, which offers a power dissipation 84.4% lower in comparison to the first mode at the cost of a compression efficiency loss of 2.48%. The proposed architecture is capable of operating in real-time for video resolutions of up to 4K@60fps. When synthesized for the TSMC 40nm technology and operating at a target frequency of 1,902 MHz for 4K@60fps videos, the High Quality and the Low Power modes presented a power dissipation of 631.76 mW and 82.76 mW, respectively.
面向AV1帧内定向预测的可配置电源/质量感知硬件设计
AOMedia Video 1 (AV1)是2018年推出的一种免版税视频格式,它包含了多个工具,可以实现高编码效率,但也存在较高的计算成本。因此,硬件解决方案是加快编码过程和实现实时处理的一个很好的替代方案。本文提出了一种用于AV1方向帧内预测的可配置硬件设计。基于软件分析定义了两种工作模式,并在该架构中实现:提供高编码效率的高质量模式和低功耗模式,其功耗比第一种模式低84.4%,但压缩效率损失为2.48%。所提出的架构能够在高达4K@60fps的视频分辨率下实时运行。当采用台积电40nm技术合成且工作在1,902 MHz目标频率下用于4K@60fps视频时,高质量和低功耗模式的功耗分别为631.76 mW和82.76 mW。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信