上海期智研究院PI，上海交通大学计算机科学与工程系教授。
2007年于上海交通大学获得计算机科学与技术学士学位，2013年于香港中文大学计算机科学与工程系获得博士学位。研究方向为芯片设计自动化，计算机体系结构等，在ISCA，MICRO，HPCA，DAC等国际会议与期刊上发表论文120余篇。
个人荣誉
2023年，入选国家级青年人才计划（QB）
2022年，吴文俊人工智能奖（芯片类）二等奖
2019年，CCF（中国计算机学会）集成电路Early Career Award
2019年，中国图灵大会，ACM上海新星奖
存算一体架构：构建存储内计算的非冯诺依曼系统架构的关键科学问题与技术挑战，突破内存墙瓶颈
高性能AI系统：算法、编解码、编译与架构跨层协同的设计与优化，解决大模型的算力、访存与通信瓶颈
成果2：软硬协同的稀疏存内计算架构
随着数据为中心的人工智能应用不断增长，传统处理器面临着处理能力、性能和能耗瓶颈的挑战。为了突破内存墙和能效墙的限制，存算一体架构成为领域专用处理器设计的趋势之一。蒋力团队在2023年进一步扩展了对存算一体架构的研究，不仅进一步探索细粒度动静态稀疏性数据，从而提升计算性能和能效。特别是将超维计算（HDC）的特点——高维数据表示和并行处理能力——与存算一体架构的优势结合起来。这一融合旨在应对传统处理器在处理大量非结构化、非规则稀疏数据时面临的性能和能效瓶颈。开发新型存算一体架构，该架构可以更有效地利用细粒度动静态稀疏性数据，从而提升计算性能和能效。
图. 存算一体架构动静态稀疏利用示意图
忆阻器阵列在存算一体架构中，因其高度结构化的操作数与数据流紧耦合问题，虽提供了计算的高并行性，但对细粒度的稀疏性支持不足。针对这一问题，2023年度的研究中，我们提出了一种新的架构设计方法，有效结合了静态稀疏和动态稀疏。对于静态稀疏，通过解耦数据流和操作数，我们设计了一种灵活的数据路径，使得架构能够在不牺牲并行性的同时，更高效地支持细粒度压缩。对于动态稀疏，利用层次化结构化稀疏范式，以有效地加速深度神经网络模型，特别是在面对不同、运行时稀疏性。这种方法通过将张量值分成更小的块并赋予每个块简单的稀疏模式，实现了模型性能的优化这种方法不仅提高了计算性能，也降低了能耗，为存算一体架构可以更高效地处理大规模数据集中的稀疏性提供了更好的支持。
图. 类脑融合的安全性测试示意图
类脑计算范式，特别是超维计算（HDC），因其在模拟人脑处理信息的方式上具有独特的优势，正在被广泛研究以优化边缘端部署的安全性。这种计算范式使用高维向量（即超向量）来表示数据和操作，可以通过简单的几何和代数操作来执行学习和认知任务。HDC的核心优势之一在于其超向量的稳健性，即使在噪声或错误数据的情况下也能维持其功能。在存算一体架构研究中嵌入HDC，不仅提高了计算效率，也极大地加强了边缘设备的安全性。
团队针对以存算一体架构为代表的非冯计算架构，通过细粒度稀疏算法和架构的协同优化，显著提高了神经网络计算性能。将算法进一步融合AI编译框架中，通过多层IR机制，用一种特殊的IR来表示稀疏张量算子。通过编译lowering与lifting的两次pass结合，自动根据芯片架构的张量运算单元的粒度（如忆阻器阵列大小）进行稀疏算子的拆分，以及最优化的算子融合（重新融合成不包含或只包含少量零的张量算子）。
图示：细粒度的稀疏化算法与映射方法
课题组的相关成果发表于2023年的欧洲设计自动化与测试学术会议（DATE）、IEEE计算机汇刊 （TC）和设计自动化会议（DAC），并荣获最佳论文提名奖。稀疏编译器提升了华为昇腾处理器2倍性能，合入了华为Mindspore开源AI编译框架，并荣获华为火花奖。
------------------------------------------------------------------------------------------------
成果1：多比特神经形态计算阵列精度自适应编程
针对现有基于ReRAM神经形态计算系统中普遍存在的高阻值不稳定性的问题，亟需一种高效高精度的ReRAM的权值映射方案，这也是所有基于忆阻器件的存算系统所必须要解决的关键性问题。蒋力团队提出了多比特精度自适应编程的电路级解决方案，并深度融合现有存算系统，使得ReRAM存算系统的存储密度和算力提升2倍，同时神经网络权值部署速度和能效分别提升4.7倍和2倍。从早期SAWM编程电路工作出发，与现有的ReRAM-based存算一体加速系统深度融合，在最大程度复用现有外围电路的基础之上，设计实现了极致紧凑的自适应编程电路，单编程通道只需要额外增加7个晶体管，就可以同时实现对SET和RESET操作的自适应截止，同时紧凑的电路设计还保证了反馈电路的高效性，实现更加精确的阻值控制。该成果发表在2022年的欧洲设计自动化与测试学术会议（Design, Automation and Test in Europe, DATE），并荣获Test and Dependability领域最佳论文奖。
36. INSPIRE: Accelerating Deep Neural Networks via Hardware-friendly Index-Pair Encoding, Fangxin Liu, Ning Yang, Zhiyan Song, Zongwu Wang, Haomin Li, Shiyuan Huang, Zhuoran Song, Songwen Pei and Li Jiang, DAC 2024.
35. EOS: An Energy-Oriented Attack Framework for Spiking Neural Networks, Yilong Zhao, Mingyu Gao, Fangxin Liu, Yiwei Hu, Zongwu Wang, Han Lin, Ji Li, He Xian, Hanlin Dong, Tao Yang, Naifeng Jing, Xiaoyao Liang, Li Jiang, DAC 2024.
34. RTSA: An RRAM-TCAM based In-Memory-Search Accelerator for Sub-100 μs Collision Detection, Jiahao Sun, Fangxin Liu, Yijian Zhang, Li Jiang, and Rui Yang, DATE 2024.
33. UM-PIM: DRAM-based PIM with Uniform & Shared Memory Space, Yilong Zhao, Mingyu Gao, Fangxin Liu, Yiwei Hu, Zongwu Wang, Han Lin, Ji Li, He Xian, Hanlin Dong, Tao Yang, Naifeng Jing, Xiaoyao Liang, Li Jiang, ISCA 2024.
32. Haomin Li, Fangxin Liu, Yichi Chen and Li Jiang, HyperNode: An Efficient Node Classification Framework Using HyperDimensional Computing, ICCD, 2023 查看PDF
31. Fangxin Liu, Ning Yang and Li Jiang, PSQ: An Automatic Search Framework for Data-Free Quantization on PIM-based Architecture, ICCD, 2023 查看PDF
30. Fangxin Liu, HaominLi, Yongbiao Chen, TaoYang and Li Jiang, HyperAttack: An Effcient Attack Framework for HyperDimensional Computing, DAC, 2023 查看PDF
29. Tao Yang, YiyuanZhou, QidongTang, FengXu, HuiMa, JieruZhao and Li Jiang, SpM
MPlu: A Compiler Plug-in with Sparse IR for Efficient Sparse Matrix Multiplication, DAC, 2023 查看PDF
28. Fangxin Liu, Wenbo Zhao, Zongwu Wang, Yongbiao Chen, Xiaoyao Liang and Li Jiang, ERA-BS: Boosting the Efficiency of ReRAM-based PIM Accelerator with Fine-Grained Bit-Level Sparsity, IEEE Transactions on Computers , 2023 查看PDF
27. Fangxin Liu, Wenbo Zhao, Zongwu Wang, XiaokangYang and Li Jiang, SIMSnn:A Weight-Agnostic ReRAM-based Search-In-Memory Engine for SNN Acceleration, DATE, 2023 查看PDF
26. Tao Yang, HuiMa, Yilong Zhao, Fangxin Liu, Zhezhi He, Xiaoli Sun and Li Jiang, PIMPR:PIM-based Personalized Recommendation with Heterogeneous Memory Hierarchy, DATE, 2023 查看PDF
25. Tao Yang, Dongyue Li, Fei Ma, Zhuoran Song, Yilong Zhao, Jiaxi Zhang, Fangxin Liu and Li Jiang, PASGCN: An ReRAM-Based PIM Design for GCN With Adaptively Sparsified Graphs., TCAD, 2023 查看PDF
24. Fangxin Liu, Zongwu Wang, Yongbiao Chen, Zhezhi He, Tao Yang, Xiaoyao Liang and Li Jiang, SoBS-X: Squeeze-Out Bit Sparsity for ReRAM-Crossbar-Based Neural Network Accelerator, TCAD, 2023 查看PDF
23. Yanan Sun, Chang Ma, Zhi Li, Yilong Zhao, Jiachen Jiang, Weikang Qian, Rui Yang, Zhezhi He, Unary Coding and Variation-Aware Optimal Mapping Scheme for Reliable ReRAM-based Neuromorphic Computing, (TCAD) IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2021 查看PDF
22. Tao Yang, Dongyue Li, Yibo Han, Yilong Zhao, Fangxin Liu, Xiaoyao Liang, Zhezhi He, Li Jiang, PIMGCN: A ReRAM-Based PIM Design for Graph Convolutional Network Acceleration, (DAC) in ACM/IEEE Design Automation Conference, 2021 查看PDF
21. Yilong Zhao, Zhezhi He, Naifeng Jing, Xiaoyao Liang, Li Jiang, Re2PIM: A Reconfigurable ReRAM-based PIM Design for Variable-sized Vector-Matrix Multiplication, (GLSVLSI) in ACM Great Lakes Symposium on VLS , 2021 查看PDF
20. Fangxin Liu, Wenbo Zhao, Zongwu Wang, Tao Yang, Li Jiang, IM3A: Boosting Deep Neural Network Efficiency via In-Memory Addressing-Assisted Acceleration, (GLSVLSI) in ACM Great Lakes Symposium on VLSI, 2021 查看PDF
19. Zhuoran Song, Dongyue Li, Zhezhi He, Xiaoyao Liang, Li Jiang, ReRAM-Sharing: Fine-Grained Weight Sharing for ReRAM-Based Deep Neural Network Accelerator, (ISCAS) in International Symposium on Circuits and Systems , 2021 查看PDF
18. Xingyi Wang, Yu Li, Yiquan Chen, Shiwen Wang, Yin Du, Cheng He, YuZhong Zhang, Pinan Chen, Xin Li, Wenjun Song, Qiang xu, Li Jiang, On Workload-Aware DRAM Failure Prediction in Large-Scale Data Centers, (VTS) in IEEE VLSI Test Symposium, 2021 查看PDF
17. Fangxin Liu, Wenbo Zhao, Zhezhi He, Zongwu Wang, Yilong Zhao, Yongbiao Chen, Li Jiang, Bit-Transformer: Transforming Bit-level Sparsity into Higher Preformance in ReRAM-based Accelerator, (ICCAD)International Conference on Computer-Aided Design, 2021 查看PDF
16. Fangxin Liu,Wenbo Zhao,Zhezhi He,Yanzhi Wang,Zongwu Wang, Changzhi Dai, Xiaoyao Liang, Li Jiang, Improving Neural Network Efficiency via Post-training Quantization with Adaptive Floating-Point, (ICCV)International Conference on Computer Vision, 2021 查看PDF
15. Dongyue Li,Tao Yang,Lun Du,Zhezhi He,Li Jiang, AdaptiveGCN: Efficient GCN Through Adaptively Sparsifying Graphs, (CIKM)International Conference on Information and Knowledge Management, 2021 查看PDF
14. Hanchen Guo, Zhehan Lin, Yunfei Gu, Chentao Wu*, Li Jiang*, Jie Li, Guangtao Xue, Minyi Guo, Lazy-WL: A Wear-aware Load Balanced Data Redistribution Method for Efficient SSD Array Scaling, (CLUSTER)IEEE International Conference on Cluster Computing, 2021 查看PDF
13. Fangxin Liu, Wenbo Zhao, Zhezhi He, Zongwu Wang, Yilong Zhao, Tao Yang, Xiaoyao Liang, Naifeng Jing and Li Jiang, SME: ReRAM-based Sparse-Multiplication-Engine to Squeeze-Out Bit Sparsity of Neural Network, (ICCD)International Conference on Computer Design, 2021 查看PDF
12. Fangxin Liu, Wenbo Zhao, Zongwu Wang,Qidong Tang, Yongbiao Chen,Zhezhi He,Naifeng Jing,Xiaoyang Liang and Li Jiang, EBSP: Evolving Bit Sparsity Patterns for Hardware-Friendly Inference of Quantized Deep Neural Networks, (DAC) in ACM/IEEE Design Automation Conference, 2022 查看PDF
11. Fangxin Liu, Wenbo Zhao,Yongbiao Chen,Zongwu Wang,Zhezhi He,Rui Yang,Qidong Tang, Tao Yang,Cheng Zhuo and Li Jiang, PIM-DH: ReRAM-based Processing-in-Memory Architecture for Deep Hashing Acceleration, (DAC) in ACM/IEEE Design Automation Conference, 2022 查看PDF
10. Fangxin Liu，Wenbo Zhao, Zongwu Wang,Yongbiao Chen,Li Jiang, SpikeConverter: An Efficient Conversion Framework Zipping the Gap between Artificial Neural Networks and Spiking Neural Networks, AAAI Conference on Artificial Intelligence, 2022 查看PDF
9. Tao Yang, Dongyue Li, Zhuoran Song, Yilong Zhao, Fangxin Liu, Zongwu Wang, Zhezhi He and Li Jiang, DTQAtten: Leveraging Dynamic Token-based Quantization for Efficient Attention Architecture, (DATE)Design, Automation & Test in Europe Conference & Exhibition, 2022 查看PDF
8. Tao Yang，Dongyue Li，Fei Ma，Zhuoran Song，Yilong Zhao，Jiaxi Zhang，Fangxin Liu and Li Jiang, PASGCN: An ReRAM-Based PIM Design for GCN with Adaptively Sparsified Graphs, （TCAD）IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2022 查看PDF
7. Zongwu Wang,Zhezhi He, Rui Yang,Shiquan Fan，Jie Lin, Fangxin Liu，Yueyang Jia, Chenxi Yuan，Qidong Tang and Li Jiang, Self-Terminating Write of Multi-Level Cell ReRAM for Efficient Neuromorphic Computing, (DATE)Design, Automation & Test in Europe Conference & Exhibition, 2022 查看PDF
6. Fangxin Liu, Wenbo Zhao, Yongbiao Chen, Zongwu Wang, Tao Yang and Li Jiang, SSTDP: Supervised Spike Timing Dependent Plasticity for Efficient Spiking Neural Network Training, Frontiers in Neuroscience, section Neuromorphic Engineering, 2022 查看PDF
5. Qidong Tang, Zhezhi He, Fangxin Liu, Zongwu Wang, Yiyuan Zhou, Yinghuan Zhang, Li Jiang, HAWIS: Hardware-Aware Automated WIdth Search for Accurate, Energy-Efficient and Robust Binary Neural Network on ReRAM Dot-Product Engine, (ASP-DAC)27th Asia and South Pacific Design Automation Conference, 2022 查看PDF
4. Fangxin Liu, Wenbo Zhao, Zongwu Wang, Yongbiao Chen, Tao Yang, Zhezhi He, Xiaokang Yang and Li Jiang, SATO: Spiking Neural Network Acceleration via Temporal-Oriented Dataflow and Architecture, (DAC) in ACM/IEEE Design Automation Conference, 2022 查看PDF
3. Fangxin Liu，Haomin Li，Xiaokang Yang，Li Jiang, L3E-HD: A Framework Enabling Efficient Ensemble in High-Dimensional Space for Language Tasks, (SIGIR)International Conference on Research and Development in Information Retrieval, 2022 查看PDF
2. Fangxin Liu, Zongwu Wang,Wenbo Zhao, Yongbiao Chen, Xiaokang Yang and Li Jiang, Randomize and Match: Exploiting Irregular Sparsity for Energy Efficient Processing in SNNs, IEEE International Conference on Computer Design (ICCD), 2022 查看PDF
1. Tao Yang,Hui Ma,Xiaoling Li,Fangxin Liu,Yilong Zhao,Zhezhi He and Li Jiang, DTATrans: Leveraging Dynamic Token-basedQuantization with Accuracy Compensation Mechanism for Efficien tTranformer Architecture, IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems(TCAD), 2022 查看PDF
.jpg)
