Cnvlutin2: Ineffectual-Activation-and-Weight-Free Deep Neural Network Computing
Cnvlutin2: Ineffectual-Activation-and-Weight-Free Deep Neural Network Computing
复制标题
DOI：
--
复制
发表时间：
2017-04
期刊：
影响因子：
--
通讯作者：
Patrick Judd;Alberto Delmas Lascorz;Sayeh Sharify;Andreas Moshovos
中科院分区：
文献类型：
--
作者：
Patrick Judd;Alberto Delmas Lascorz;Sayeh Sharify;Andreas Moshovos
我们讨论了对先前提出的CNV加速器的一些修改和扩展，用于深度学习网络的卷积和全连接层。我们首先描述了被认为无效的激活的不同编码。这些编码具有不同的存储开销和能量特性。我们建议在从内存访问激活时使用间接级别，通过仅存储有效的激活来减少内存占用。我们还提出了一个修改后的组织，它在从内存中提取激活时检测被认为无效的激活。这与在前一层的输出处检测它们的原始设计不同。最后，我们提出了一个扩展的CNV，它也可以跳过无效的权重。
We discuss several modifications and extensions over the previous proposed Cnvlutin (CNV) accelerator for convolutional and fully-connected layers of Deep Learning Network. We first describe different encodings of the activations that are deemed ineffectual. The encodings have different memory overhead and energy characteristics. We propose using a level of indirection when accessing activations from memory to reduce their memory footprint by storing only the effectual activations. We also present a modified organization that detects the activations that are deemed as ineffectual while fetching them from memory. This is different than the original design that instead detected them at the output of the preceding layer. Finally, we present an extended CNV that can also skip ineffectual weights.