arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1802.06402cs.LGstat.ML
收藏

面向深度学习系统的超高性能和能量效率:一个硬件协同优化框架

Towards Ultra-High Performance and Energy Efficiency of Deep Learning Systems: An Algorithm-Hardware Co-Optimization Framework

Syracuse University(雪城大学);City University of New York(纽约城市大学);University of Southern California(南加州大学);Northeastern University(东北大学)

Yanzhi Wang, Caiwen Ding, Zhe Li, Geng Yuan, Siyu Liao, Xiaolong Ma, Bo Yuan, Xuehai Qian, Jian Tang, Qinru Qiu, Xue Lin

摘要:

深度学习系统的硬件加速已经在工业界和学术界得到了广泛的研究。本文的目的是实现深度神经网络(DNN)硬件实现的超高能效和性能。开发了一个算法-硬件协同优化框架,该框架适用于不同类型、规模和应用场景的DNN。算法部分采用一般的块循环矩阵,实现了精确度和压缩比之间的细粒度折衷。它适用于全连接层和卷积层,并包含该方法有效性的数学严格证明。该算法将训练和推理的计算复杂性度从O($n^2$)降低到O($n\log n$),存储复杂度从O($n^2$)降低到O($n$)。硬件部分由高效的现场可编程门阵列(FPGA)为基础的实现,使用有效的重新配置,批处理,深度流水线,资源重用,分层控制。实验结果表明,在相同的测试精度下,与IBM TrueNorth处理器相比,该框架实现了至少152倍的加速比和71倍的能效增益。与基于FPGA的参考工作相比,它实现了至少31倍的能效增益。

英文摘要:

Hardware accelerations of deep learning systems have been extensively investigated in industry and academia. The aim of this paper is to achieve ultra-high energy efficiency and performance for hardware implementations of deep neural networks (DNNs). An algorithm-hardware co-optimization framework is developed, which is applicable to different DNN types, sizes, and application scenarios. The algorithm part adopts the general block-circulant matrices to achieve a fine-grained tradeoff between accuracy and compression ratio. It applies to both fully-connected and convolutional layers and contains a mathematically rigorous proof of the effectiveness of the method. The proposed algorithm reduces computational complexity per layer from O($n^2$) to O($n\log n$) and storage complexity from O($n^2$) to O($n$), both for training and inference. The hardware part consists of highly efficient Field Programmable Gate Array (FPGA)-based implementations using effective reconfiguration, batch processing, deep pipelining, resource re-using, and hierarchical control. Experimental results demonstrate that the proposed framework achieves at least 152X speedup and 71X energy efficiency gain compared with IBM TrueNorth processor under the same test accuracy. It achieves at least 31X energy efficiency gain compared with the reference FPGA-based work.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑