多核数据管理器仲裁机制的研究与设计
Research and Design of Arbitration Mechanism for Multi-Core Data Manager
摘要: 多核片上数据传输是指在片上系统中,通过共享存储器、片上网络等方式进行通讯,实现多个处理器协同处理任务的技术。为了达到多核片上传输技术在进行多个处理器核和多个外部设备之间数据传输时,同时兼顾数据包发送和接收、有效监控数据交换和灵活派发数据包的目标,本文提出了一种基于大规模数据的拆解/封包传输的多核数据仲裁机制,新机制采用优先级多级调整策略,实现数据包有序进出队列管理器,从而减少内部通信方面的负担,提升系统整体性能,同时完成了新机制的HDL建模和仿真,实验结果表明,所提出的设计能够保证多组数据的优先级次序,提高传输时的数据利用率,平衡响应速度,在多数据场景下,实现公平有效地传输。
Abstract: Multi-core on-chip data transmission refers to a technology that communicates through shared memory, on-chip network, etc., in a system-on-chip to realize collaborative processing tasks by multiple processors. In order to achieve the goal of multi-core on-chip transmission technology in data transmission between multi-ple processor cores and multiple external devices, while taking into account data packet sending and receiving, effectively monitoring data exchange and flexibly dispatching data packets, this pa-per proposes a method based on a multi-core data arbitration mechanism for large-scale data dis-assembly/packet transmission. The new mechanism adopts a priority multi-level adjustment strategy to realize the orderly entry and exit of data packets into the queue manager, thereby re-ducing the burden of internal communication and improving the overall performance of the system. The HDL modeling and simulation of the new mechanism are carried out. The experimental results show that the proposed design can ensure the priority order of multiple groups of data, improve the data utilization rate during transmission, balance the response speed, and achieve fair and efficient transmission in multi-data scenarios.
文章引用:马燕. 多核数据管理器仲裁机制的研究与设计[J]. 传感器技术与应用, 2022, 10(3): 369-377. https://doi.org/10.12677/JSTA.2022.103044

参考文献

[1] 樊超, 詹思维, 李博. 基于多核DSP的高清图像跟踪系统[J]. 电子技术, 2021, 50(3): 4-5.
[2] 秦昳, 史晓楠, 巨新刚. 多核处理器核间的通信研究与实现[J]. 现代电子技术, 2016, 39(16): 83-87.
[3] 陈术涛, 沈志, 王春联, 等. 多核DSP与FPGA高速数据传输系统设计与实现[J]. 电子技术应用, 2018, 44(12): 40-43.
[4] 魏智伟. 多核DSP间基于SRIO数据传输的设计与实现[J]. 微型机与应用, 2017, 36(4): 36-39.
[5] 王磊, 刘道福, 陈云霁, 等. 片上多核处理器共享资源分配与调度策略研究综述[J]. 计算机研究与发展, 2013, 50(10): 2212-2227.
[6] 蔡想伟. CPU-GPU异构系统下的片上网络仲裁机制研究[J]. 电子设计工程, 2018, 26(8): 93-101.
[7] Li, Z., Yang, G., Liu, S., et al. (2018) A Task-Based Multi-Core Allocation Mechanism for Packet Acceleration. IEICE Elec-tronics Express, 15, 9 p. [Google Scholar] [CrossRef
[8] 牛金海. TMS320C66xKeystone架构多核DSP入门与实例精解[M]. 上海: 上海交通大学出版社, 2015: 1-106.
[9] 夏际金, 赵洪立, 李川. TI C66x多核DSP高级软件开发技术[M]. 北京: 清华大学出版社, 2017: 1-131.
[10] 周超群, 周亦敏. 一种改进的基于复制的异构多核任务调度算法[J]. 电子科技, 2017, 30(6): 57-62.
[11] 钱鹏飞, 乔庐峰, 陈庆华. 一种在大容量交换中可以保证QoS的队列调度器设计与实现[J]. 通信技术, 2018, 51(11): 2655-2661.
[12] Guo, Y. and Li, Z. (2018) A Novel Priority-Allocated Scheme for Flow-Based Queue Managers. 2018 IEEE 20th International Conference on High Performance Computing and Communications; IEEE 16th International Conference on Smart City; IEEE 4th International Conference on Data Science and Systems (HPCC/SmartCity/DSS), Exeter, 28-30 June 2018, 1002-1006. [Google Scholar] [CrossRef
[13] 徐金波, 常俊胜, 李琰. 支持多优先级多输出通道的数据队列调度方法和硬件实现[J]. 计算机工程与科学, 2020, 42(10): 1749-1756.
[14] 章婷婷, 彭敏, 周清峰, 等. 无线体域网中动态分配队列长度的调度算法[J]. 合肥工业大学学报(自然科学版), 2019, 42(7): 906-911.
[15] 罗乐, 王春华, 张多利, 等. 一种多核系统改进型列表调度算法[J]. 电子科技, 2020, 33(6): 52-57.
[16] Benacer, I., Boyer, F.R. and Savaria, Y. (2017) A High-Speed Traffic Manager Architecture for Flow-Based Networking. 2017 15th IEEE International New Circuits and Systems Conference (NEWCAS), Strasbourg, 25-28 June 2017, 161-164. [Google Scholar] [CrossRef
[17] 赵海峰. 高性能队列管理与调度技术研究[D]: [硕士学位论文]. 西安: 西安电子科技大学, 2018: 1-65.
[18] 李润泽. 多核数据导航器的设计与验证[D]: [硕士学位论文]. 长沙: 国防科技大学, 2018: 1-54.