Design Space Exploration of HW Accelerators and Network Infrastructure for FPGA-Based MPSoC

被引:0
|
作者
Dammak, Bouthaina [1 ]
Baklouti, Mouna [2 ]
Alsekait, Deema [1 ]
机构
[1] Princess Nourah Bint Abdulrahman Univ, Appl Coll, Dept Comp Sci, Riyadh 84428, Saudi Arabia
[2] Univ Sfax, Comp Embedded Syst Lab, CES Lab, Sfax 3038, Tunisia
关键词
Program processors; Task analysis; Computer architecture; Field programmable gate arrays; Space exploration; Multiprocessor interconnection; Hardware acceleration; System-on-chip; High level synthesis; Multi-processor system-on-chip; custom instructions; shared hardware accelerator; network communication; high-level exploration; PERFORMANCE; ALGORITHMS; SYSTEMS;
D O I
10.1109/ACCESS.2024.3357352
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Supercomputing systems are increasingly reliant on heterogeneous Multiprocessors System on-Chip (MPSoCs), merging multiple processors and hardware accelerators (HWAcc) on the same die to achieve power and performance needs. Due to CPU complication and timing closure challenges of tightly coupled design approach, the state-of-the art of HWAcc design methodology relies on coupling the processors with loosely coupled HWAccs. Loosely-coupled HWAccs can be shared or private accelerators running custom instructions to form a heterogeneous multi-processor system. Some works discussed the determination of the sharing degree of the HWAcc over the processors, however the impact of the integrated communication infrastructure is not discussed. Thus, we propose a high-level design exploration tool to select the accelerators and generate the adequate communication interconnect under performance and area constraints. Different homogeneous and heterogeneous multi-processor configurations are evaluated and compared running different signal processing benchmarks. Experimental results show the efficiency of the proposed exploration tool to rapidly explore and select the adequate architecture.
引用
收藏
页码:15280 / 15289
页数:10
相关论文
共 50 条
  • [31] DNNVM: End-to-End Compiler Leveraging Heterogeneous Optimizations on FPGA-Based CNN Accelerators
    Xing, Yu
    Liang, Shuang
    Sui, Lingzhi
    Jia, Xijie
    Qiu, Jiantao
    Liu, Xin
    Wang, Yushun
    Shan, Yi
    Wang, Yu
    IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS, 2020, 39 (10) : 2668 - 2681
  • [32] Design Space Exploration of The KNN Imputation on FPGA
    Al-Zoubi, Ahmad
    Tatas, Konstantinos
    Kyriacou, Costas
    2018 7TH INTERNATIONAL CONFERENCE ON MODERN CIRCUITS AND SYSTEMS TECHNOLOGIES (MOCAST), 2018,
  • [33] FPGA-based Acceleration of Neural Network Training
    Sang, Ruoyu
    Liu, Qiang
    Zhang, Qijun
    2016 IEEE MTT-S INTERNATIONAL CONFERENCE ON NUMERICAL ELECTROMAGNETIC AND MULTIPHYSICS MODELING AND OPTIMIZATION (NEMO), 2016,
  • [34] An Exploration of State-of-the-Art Automation Frameworks for FPGA-Based DNN Acceleration
    Hamanaka, Fumio
    Odan, Takashi
    Kise, Kenji
    Chu, Thiem Van
    IEEE ACCESS, 2023, 11 : 5701 - 5713
  • [35] Space Optimization on Counters for FPGA-Based Perl Compatible Regular Expressions
    Lo, Chia-Tien Dan
    Tai, Yi-Gang
    ACM TRANSACTIONS ON RECONFIGURABLE TECHNOLOGY AND SYSTEMS, 2009, 2 (04)
  • [36] Towards Open-HW: A platform to design, share and deploy FPGA accelerators in low cost
    Zhao Q.
    Amagasaki M.
    Iida M.
    Kuga M.
    Sueyoshi T.
    IPSJ Transactions on System LSI Design Methodology, 2017, 10 : 63 - 70
  • [37] ExPAN(N)D: Exploring Posits for Efficient Artificial Neural Network Design in FPGA-Based Systems
    Nambi, Suresh
    Ullah, Salim
    Sahoo, Siva Satyendra
    Lohana, Aditya
    Merchant, Farhad
    Kumar, Akash
    IEEE ACCESS, 2021, 9 : 103691 - 103708
  • [38] Detect and Replace: Efficient Soft Error Protection of FPGA-Based CNN Accelerators
    Gao, Zhen
    Qi, Yanmao
    Shi, Jinchang
    Liu, Qiang
    Ge, Guangjun
    Wang, Yu
    Reviriego, Pedro
    IEEE TRANSACTIONS ON VERY LARGE SCALE INTEGRATION (VLSI) SYSTEMS, 2025, 33 (01) : 66 - 74
  • [39] Accelerating FPGA Prototyping through Predictive Model-Based HLS Design Space Exploration
    Liu, Shuangnan
    Lau, Francis C. M.
    Schafer, Benjamin Carrion
    PROCEEDINGS OF THE 2019 56TH ACM/EDAC/IEEE DESIGN AUTOMATION CONFERENCE (DAC), 2019,
  • [40] Roofline-Model-Based Design Space Exploration for Dataflow Techniques of CNN Accelerators
    Park, Chan
    Park, Sungkyung
    Park, Chester Sungchung
    IEEE ACCESS, 2020, 8 : 172509 - 172523