基于RISC‐V的FFmpeg多媒体算法库优化策略

doi:10.19678/j.issn.1000-3428.0063495

计算机工程 ›› 2023, Vol. 49 ›› Issue (4): 159-165,173. doi: 10.19678/j.issn.1000-3428.0063495

基于RISC‐V的FFmpeg多媒体算法库优化策略

张桢^1,2, 梁军¹, 贾海鹏², 张云泉², 李青¹

1. 北京联合大学北京市信息服务工程重点实验室, 北京 100101;
2. 中国科学院计算技术研究所计算机体系结构国家重点实验室北京 100190

收稿日期:2021-12-09 修回日期:2022-02-19 发布日期:2023-04-07
作者简介:张桢(1996-),男,硕士研究生,主研方向为并行算法优化;梁军(通信作者),教授;贾海鹏,高级工程师、博士;张云泉,研究员、博士;李青,副教授、博士。
基金资助:
国家自然科学基金（61972376）；北京联合大学科研项目（ZK50202002）。

Optimization Strategy of FFmpeg Multimedia Algorithm Library Based on RISC-V

ZHANG Zhen^1,2, LIANG Jun¹, JIA Haipeng², ZHANG Yunquan², LI Qing¹

1. Beijing Key Laboratory of Information Service Engineering, Beijing Union University, Beijing 100101, China;
2. State Key Laboratory of Computer Architecture, Institute of Computing Technology, Chinese Academy of Sciences, Beijing 100190, China

Received:2021-12-09 Revised:2022-02-19 Published:2023-04-07

摘要/Abstract

摘要： RISC-V处理器的广泛应用使得FFmpeg多媒体算法库在RISC-V平台上的高性能实现日益重要。提出一种基于RISC-V架构的系列优化策略，针对开源音视频多媒体FFmpeg算法库中不同特征和计算密度的算法，利用RISC-V指令集的扩展性对算法库中某些耗时的算法进行指令加速和并行优化。在深入研究RISC-V开源架构的基础上，构建一个基于RISC-V开源架构的高性能FFmpeg算法库。针对不连续访存类算法、数据依赖类算法、数据快速转换类算法，从向量单元配置、向量化访存、汇编优化、指令流水优化4个方面出发，大幅提升FFmpeg算法库在RISC-V处理器上的性能。实验结果表明，采用以上优化策略后的FFmpeg算法库在基于RISC-V架构的XT-910芯片上的性能得到明显提升，其中的不连续访存类算法、数据依赖类算法、数据快速转换类算法的加速比分别为8.20、3.67、3.62。

关键词: 开源指令集架构, FFmpeg多媒体算法库, 向量化访存, 汇编优化, 指令流水优化

Abstract: The widespread application of RISC-V processors has made the high-performance implementation of FFmpeg multimedia algorithm library on the RISC-V platform increasingly important.This study proposes a series of RISC-V architecture-based optimization strategies aimed at algorithms with different characteristics and computational densities in the open source audio and video multimedia FFmpeg algorithm library and uses the extensibility of the RISC-V instruction set to accelerate and optimize the instructions of few time-consuming algorithms in the library.Based on an in-depth study of the RISC-V open source architecture, a high-performance FFmpeg algorithm library based on RISC-V is built.The performance of the FFmpeg algorithm library on RISC-V processors is significantly improved with the aim of discontinuous memory retrieval, data dependency, and fast data conversion algorithms in four aspects:vector unit configuration, vectorized memory access, assembly optimization, and instruction pipeline optimization.The experimental results show that adoption of the aforementioned optimization strategy significantly improved the performance of the FFmpeg algorithm library on the XT-910 chip based on RISC-V architecture, and the speedup ratios of the discontinuous memory access, data dependency, and data fast conversion algorithms are 8.20, 3.67, and 3.62, respectively.

Key words: open source Instruction Set Architecture(ISA), FFmpeg multimedia algorithm library, vectorized memory access, assembly optimization, instruction pipeline optimization

中图分类号:

TP391

张桢, 梁军, 贾海鹏, 张云泉, 李青. 基于RISC‐V的FFmpeg多媒体算法库优化策略[J]. 计算机工程, 2023, 49(4): 159-165,173.

ZHANG Zhen, LIANG Jun, JIA Haipeng, ZHANG Yunquan, LI Qing. Optimization Strategy of FFmpeg Multimedia Algorithm Library Based on RISC-V[J]. Computer Engineering, 2023, 49(4): 159-165,173.

http://www.ecice06.com/CN/Y2023/V49/I4/159

图/表 8

20230417185221

20230417185223

20230417185226

20230417185229

20230417185232

20230417185236

20230417185239

20230417185242

参考文献

[1] The FFmpeg developers.FFmpeg documentation[EB/OL].[2021-11-08].http://svn.ffmpeg.org/ffmpeg-protocols.html.
[2] NISHII O, HANAWA M, NISHIMUKAI T, et al.A 1000 MIPS BiCMOS microprocessor with superscalar architecture[C]//Proceedings of IEEE International Solid-State Circuits Conference Digest of Technical Papers.Washington D.C., USA:IEEE Press, 2002:114-115.
[3] HENNESS Y.VLSI processor architecture[J].IEEE Transactions on Computers, 1984, 33(12):1221-1246.
[4] TINA G, DARUWALA R D.Performance improvement of MIPS architecture by adding new features[J].International Journal of Advanced Research in Computer Science and Software Engineering, 2013, 3(2):34-51.
[5] FEEHRER J, JAIRATH S, LOEWENSTEIN P, et al.The oracle sparc T516-core processor scales to eight sockets[J].IEEE Micro, 2013, 33(2):48-57.
[6] KONGETIRA P, AINGARAN K, OLUKOTUN K.Niagara:a 32-way multithreaded sparc processor[J].IEEE Micro, 2005, 25(2):21-29.
[7] KONSTADINIDIS G K, TREMBLAY M, CHAUDHRY S, et al.Architecture and physical implementation of a third generation 65 nm, 16 core, 32 thread chip-multithreading SPARC processor[J].IEEE Journal of Solid-State Circuits, 2009, 44(1):7-17.
[8] PAPADOPOULOS N C, KARAKOSTAS V, NIKAS K, et al.Enabling virtual memory research on RISC-V with a configurable TLB hierarchy for the rocket chip generator[EB/OL].[2021-11-08].https://arxiv.org/abs/2009.07723.
[9] HAN R, TINE B, LEE J, et al.Supporting CUDA for an extended RISC-V GPU architecture[EB/OL].[2021-11-08].https://arxiv.org/abs/2109.00673.
[10] LEE Y, WATERMAN A, AVIZIENIS R, et al.A 45nm 1.3 GHz 16.7 double-precision GFLOPS/W RISC-V processor with vector accelerators[C]//Proceedings of the 40th European Solid State Circuits Conference.Washington D.C., USA:IEEE Press, 2014:199-202.
[11] MENON A, MURUGAN S, REBEIRO C, et al.Shakti-T:a RISC-V processor with light weight security extensions[C]//Proceedings of the Hardware and Architectural Support for Security and Privacy.New York, USA:ACM Press, 2017:1-8.
[12] PATIL V, RAVEENDRAN A, SOBHA P M, et al.Out of order floating point coprocessor for RISC V ISA[C]//Proceedings of the 19th International Symposium on VLSI Design and Test.New York, USA:ACM Press, 2015:1-7.
[13] TAGLIAVINI G, MACH S, ROSSI D, et al.Design and evaluation of smallfloat SIMD extensions to the RISC-V ISA[C]//Proceedings of Design, Automation and Test in Europe Conference and Exhibition.Washington D.C., USA:IEEE Press, 2019:654-657.
[14] BAO Y G.Answer:the 1st RISC-V China summit will be held in Shanghai from June 21, 2021.Is there anything worthy of attention?[EB/OL].[2021-11-08].https://www.zhihu.com/question/466393646/answer/1955410750.
[15] 黄正伟, 刘宏伟, 徐渊.用于IToF传感器的极低功耗RISC-V专用处理器设计[J].计算机工程, 2022, 48(9):146-154. HUANG Z W, LIU H W, XU Y.Design of ultra-low-power RISC-V dedicated processor for IToF sensor[J].Computer Engineering, 2022, 48(9):146-154.(in Chinese)
[16] 李雨倩, 焦继业, 刘有耀, 等.嵌入式RISC-V乱序执行处理器的研究与设计[J].计算机工程, 2021, 47(2):261-267, 284. LI Y Q, JIAO J Y, LIU Y Y, et al.Research and design of embedded RISC-V out-of-order execution processor[J].Computer Engineering, 2021, 47(2):261-267, 284.(in Chinese)
[17] 张坤宁, 赵烁, 何虎, 等.基于RISC-V处理器的卷积加速SoC系统设计[J].计算机工程, 2021, 47(4):153-157. ZHANG K N, ZHAO S, HE H, et al.Design of SoC system for convolution acceleration based on RISC-V processor[J].Computer Engineering, 2021, 47(4):153-157.(in Chinese)
[18] MAAS M, ASANOVIC K, KUBIATOWICZ J.A hardware accelerator for tracing garbage collection[C]//Proceedings of the 45th Annual International Symposium on Computer Architecture.Washington D.C., USA:IEEE Press, 2019:38-46.
[19] 魏秋明, 梁军, 鲍泓, 等.异构计算平台图像边缘检测算法优化研究[J].计算机工程, 2017, 43(5):240-247. WEI Q M, LIANG J, BAO H, et al.Research on image edge detection algorithm optimization on heterogeneous computing platform[J].Computer Engineering, 2017, 43(5):240-247.(in Chinese)
[20] 胡敏, 卢永江, 刘兵.基于CK810处理器的汇编链接时优化[J].计算机工程, 2014, 40(11):250-254. HU M, LU Y J, LIU B.Assembly and link time optimization based on CK810 processor[J].Computer Engineering, 2014, 40(11):250-254.(in Chinese)
[21] LOUIS M S, AZAD Z, DELSHADTEHRANI L, et al.Towards Deep Learning Using TensorFlow Lite on RISC-V[EB/OL].[2021-11-08].https://projects.iq.harvard.edu/files/edge/files/carrv_workshop_submission_2019_camera_ready.pdf.
[22] TEHRANI E, GRABA T, MERABET A S, et al.Classification of lightweight block ciphers for specific processor accelerated implementations[C]//Proceedings of the 26th IEEE International Conference on Electronics, Circuits and Systems.Washington D.C., USA:IEEE Press, 2020:747-750.
[23] WATERMAN A, LEE Y, PATTERSON D A, et al.The RISC-V instruction set manual.[EB/OL].[2021-11-08].https://www.mendeley.com/catalogue/60cac0d9-5d3b-3520-918d-22acfccf69b5/.
[24] 郭金鑫, 张广婷, 张云泉, 等.Cooley-Tukey FFT算法高性能实现与优化研究[J].计算机科学与探索, 2022, 16(6):1304-1315. GUO J X, ZHANG G T, ZHANG Y Q, et al.High-performance implementation and optimization of Cooley-Tukey FFT algorithm[J].Journal of Frontiers of Computer Science and Technology, 2022, 16(6):1304-1315.(in Chinese)
[25] CHEN C, XIANG X Y, LIU C, et al.Xuantie-910:a commercial multi-core 12-stage pipeline out-of-order 64-bit high performance RISC-V processor with vector extension:industrial product[C]//Proceedings of the 47th Annual International Symposium on Computer Architecture.Washington D.C., USA:IEEE Press, 2020:52-64.

选择文件类型/文献管理软件名称

选择包含的内容

基于RISC‐V的FFmpeg多媒体算法库优化策略

Optimization Strategy of FFmpeg Multimedia Algorithm Library Based on RISC-V

RichHTML

PDF

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 8

参考文献

相关文章 2

编辑推荐

Metrics

本文评价

[1]	魏秋明,梁军,鲍泓,王晶,李论. 异构计算平台图像边缘检测算法优化研究[J]. 计算机工程, 2017, 43(5): 240-247.
[2]	胡敏,卢永江,刘兵. 基于CK810 处理器的汇编链接时优化[J]. 计算机工程, 2014, 40(11): 250-254.

模态框（Modal）标题

选择文件类型/文献管理软件名称

选择包含的内容

基于RISC‐V的FFmpeg多媒体算法库优化策略

Optimization Strategy of FFmpeg Multimedia Algorithm Library Based on RISC-V

RichHTML

PDF

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 8

参考文献

相关文章 2

编辑推荐

Metrics

本文评价