为了解决金融FAST (financial information exchange adapted for streaming)协议面临的纯软件解码延迟高，FPGA (field programmable gate array)硬件解码开发周期长、更新困难的问题，提出了基于OpenCL和HLS的硬件解码模式。通过对FAST数据解码的标记、切分、合并、解码模块进行流水优化，对切分和字段解码进行并行操作，将数据的输入输出改为流式接口减少I/O口的延时以及对切分数组进行分割映射等优化方式实现了解码过程低延迟、低抖动。实验结果表明，相比纯软件解码，本文提出的解码器处理速度提升了11倍，解码延迟缩短至1/6，抖动幅度控制在10 ns之内。相比传统HDL方式的FPGA定制硬件开发，开发效率可提升3～4倍，从而更好地满足产品更新换代的需求。
In order to solve the problems of high delay of pure software decoding, long development period of FPGA (field programmable gate array) hardware decoding and difficult update of financial FAST (financial information exchange adapted for streaming) protocol, a hardware decoding mode based on OpenCL and HLS was proposed. By optimizing the marking, segmentation, merging, and decoding modules of FAST data decoding through pipelining, parallel operations are performed on segmentation and field decoding. The input and output of the data are changed to a streaming interface to reduce I/O port latency, and the segmentation and mapping of the array segmentation are carried out to achieve low latency and low jitter in the decoding process. Experimental results show that compared with pure software decoding, the processing speed of the proposed decoder is improved by 11 times, the decoding delay is shortened to 1/6, and the jitter amplitude is controlled within 10ns. Compared with the traditional HDL custom FPGA hardware development, the proposed approach improves development efficiency by 3～4 times, thus better meeting the needs of product updates.
[1] 徐广斌, 武剑锋, 王泊, 等. 基于扩展FAST协议的金融消息压缩方法[J]. 计算机工程与应用, 2012, 48(1):138-141. Xu G B, Wu J F, Wang B, et al. Financial message compression method based on extended FAST protocol[J]. Computer Engineering and Applications, 2012, 48(1):138-141. (in Chinese)
[2] Toft J K, Nannarelli A. Energy efficient FPGA based hardware accelerators for financial applications[C]//2014 NORCHIP, 2014:1-6.
[3] Dvořák M, Kořenek J. Low latency book handling in FPGA for high frequency trading[C]//17th International Symposium on Design and Diagnostics of Electronic Circuits & Systems, 2014:175-178.
[4] Lockwood J W, Gupte A, Mehta N, et al. A low-latency library in FPGA hardware for high-frequency trading (HFT)[C]//2012 IEEE 20th Annual Symposium on High-Performance Interconnects, 2012:9-16.
[5] Puranik S, Barve M, Shah D, et al. Key-value store using high level synthesis flow for securities trading system[C]//2020 International Conference on Computing, Electronics & Communications Engineering (iCCECE), 2020:237-242.
[6] Fu H H, He C H, Luk W, et al. A nanosecond-level hybrid table design for financial market data generators[C]//2017 IEEE 25th Annual International Symposium on Field-Programmable Custom Computing Machines (FCCM), 2017:227-234.
[7] Boutros A, Grady B, Abbas M, et al. Build fast, trade fast:FPGA-based high-frequency trading using high-level synthesis[C]//2017 International Conference on ReConFigurable Computing and FPGAs (ReConFig), 2018:1-6.
[8] Yu L L, Fu Y Z, Liu T. A hardware structure for FAST protocol decoding adapting to 40Gbps bandwidth[J]. Computer Science and Engineering, 2017:290-296.
[9] Hu J X, Wang J F, Li R G. Low-latency ultra-wideband high-speed transmission protocol based on FPGA[J]. Journal of Physics:Conference Series, 2020, 1621(1):012066.
[10] 刘峰, 赵俊峰. 基于区块链的云存储数据完整性验证方案[J]. 应用科学学报, 2021, 39(1):164-173. Liu F, Zhao J F. Cloud storage data integrity verification scheme based on blockchain[J]. Journal of Applied Sciences, 2021, 39(1):164-173. (in Chinese)
[11] Bansod R, Virk R, Raval M. Low latency, high throughput trade surveillance system using in-memory data grid[C]//12th ACM International Conference on Distributed and Event-based Systems, 2018:250-253.
[12] 张延彬, 张凤麒, 王忠勇. 基于FPGA的期货行情数据并行处理设计[J]. 计算机工程与设计, 2019, 40(7):1866-1871. Zhang Y B, Zhang F Q, Wang Z Y. Design of futures market data parallel processing based on FPGA[J]. Computer Engineering and Design, 2019, 40(7):1866-1871. (in Chinese)
[13] Kinsner M, Seynhaeve D. High performance asynchronous host-device communication through the Intel® FPGA host pipe extension for OpenCLTM applications[C]//International Workshop on OpenCL, 2018:1.
[14] Kobayashi R, Oobata Y, Fujita N, et al. OpenCL-ready high speed FPGA network for reconfigurable high performance computing[C]//International Conference on High Performance Computing in Asia-Pacific Region, 2018:192-201.
[15] Li H, Fu Y Z, Liu T, et al. Fast protocol decoding in parallel with FPGA hardware[C]//2014 IEEE 17th International Conference on Computational Science and Engineering, 2014:1669-1673.
[16] Leber C, Geib B, Litz H. High frequency trading acceleration using FPGAs[C]//201121st International Conference on Field Programmable Logic and Applications, 2011:317-322.
