MASR: A Modular Accelerator for Sparse RNNs
MASR: A Modular Accelerator for Sparse RNNs
复制标题
DOI：
10.1109/pact.2019.00009
复制
发表时间：
2019-08
期刊：
影响因子：
--
通讯作者：
Udit Gupta;Brandon Reagen;Lillian Pentecost;M. Donato;Thierry Tambe;Alexander M. Rush;Gu-Yeon Wei;D. Brooks
中科院分区：
文献类型：
--
作者：
Udit Gupta;Brandon Reagen;Lillian Pentecost;M. Donato;Thierry Tambe;Alexander M. Rush;Gu-Yeon Wei;D. Brooks
递归神经网络（RNNs）正在成为语音识别的实际解决方案。rnn通过应用重复的、学习的转换来利用数据中的长期时间关系。与单向量矩阵操作的全连接（FC）层不同，RNN层由数百个这样的操作组成。这给rnn带来了卷积神经网络（cnn）或FC模型中没有的独特挑战，即大动态激活。在本文中，我们提出了MASR，这是一种原则性和模块化的架构，可以加速片上ASR的双向rnn。MASR旨在利用动态激活和静态权重的稀疏性。该架构通过一系列动态激活优化得到增强，这些优化可以实现紧凑的存储，确保在计算null操作时不浪费能量，并为高度并行的加速器设计保持高MAC利用率。与当前最先进的稀疏神经网络加速器（例如，EIE）相比，MASR提供2×area 3×energy和1.6×performance的好处。MASR的模块化特性使设计能够有效地从资源受限的低功耗物联网应用扩展到大规模、高度并行的数据中心部署。
Recurrent neural networks (RNNs) are becoming the de-facto solution for speech recognition. RNNs exploit long-term temporal relationships in data by applying repeated, learned transformations. Unlike fully-connected (FC) layers with single vector matrix operations, RNN layers consist of hundreds of such operations chained over time. This poses challenges unique to RNNs that are not found in convolutional neural networks(CNNs) or FC models, namely large dynamic activation. In this paper we present MASR, a principled and modular architecture that accelerates bidirectional RNNs for on-chip ASR. MASR is designed to exploit sparsity in both dynamic activations and static weights. The architecture is enhanced by a series of dynamic activation optimizations that enable compact storage, ensure no energy is wasted computing null operations, and maintain high MAC utilization for highly parallel accelerator designs. In comparison to current state-of-the-art sparse neural network accelerators (e.g., EIE), MASR provides 2×area 3×energy, and 1.6×performance benefits. The modular nature of MASR enables designs that efficiently scale from resource-constrained low-power IoT applications to large-scale, highly parallel datacenter deployments.