HSIM-DNN: Hardware Simulator for Computation-, Storage- and Power-Efficient Deep Neural Networks
HSIM-DNN: Hardware Simulator for Computation-, Storage- and Power-Efficient Deep Neural Networks
复制标题
DOI：
10.1145/3299874.3317996
复制
发表时间：
2019-05
期刊：
影响因子：
--
通讯作者：
Mengshu Sun-;Pu Zhao;Yanzhi Wang;N. Chang;X. Lin
中科院分区：
文献类型：
--
作者：
Mengshu Sun-;Pu Zhao;Yanzhi Wang;N. Chang;X. Lin
利用大规模深度神经网络(DNN)的深度学习在自动提取高层特征方面是有效的，但也是计算和存储密集型的。利用分块循环矩阵构造DNN可以同时实现硬件加速和模型压缩，同时保持较高的精度。提出了一种基于C++平台的精确硬件仿真器HSIM-DNN，用于模拟DNN硬件实现的准确行为，从而方便了基于块循环矩阵的DNN训练和推理程序的硬件设计。实际的FPGA实现验证了该模拟器在考虑精度、压缩比和功耗的情况下具有不同的循环块大小和数据位长度，这为硬件设计提供了极好的见解。
Deep learning that utilizes large-scale deep neural networks (DNNs) is effective in automatic high-level feature extraction but also computation and memory intensive. Constructing DNNs using block-circulant matrices can simultaneously achieve hardware acceleration and model compression while maintaining high accuracy. This paper proposes HSIM-DNN, an accurate hardware simulator on the C++ platform, to simulate the exact behavior of DNN hardware implementations and thereby facilitate the block-circulant matrix-based design of DNN training and inference procedures in hardware. Real FPGA implementations validate the simulator with various circulant block sizes and data bit lengths taking into account accuracy, compression ratio and power consumption, which provides excellent insights for hardware design.