The increasing application of deep learning technology drives the need for an efficient parallel computing architecture for Convolutional Neural Networks (CNNs). A significant challenge faced when designing a many-core CNN accelerator is to handle the data movement between the processing elements. The CNN workload introduces many-to-one traffic in addition to one-to-one and one-to-many traffic. As the de-facto standard for on-chip communication, Network-on-Chip (NoC) can support various unicast and multicast traffic. For many-to-one traffic, repetitive unicast is employed which is not an efficient way. In this paper, we propose to use the gather packet on mesh-based NoCs employing output stationary systolic array in support of many-to-one traffic. The gather packet will collect the data from the intermediate nodes eventually leading to the destination efficiently. This method is evaluated using the traffic traces generated from the convolution layer of AlexNet and VGG-16 with improvement in the latency and power than the repetitive unicast method.

中文翻译：

---

使用 Gather 支持提高基于 NoC 的 CNN 加速器的性能

深度学习技术的应用日益广泛，推动了对卷积神经网络 (CNN) 的高效并行计算架构的需求。设计众核 CNN 加速器时面临的一个重大挑战是处理处理元件之间的数据移动。除了一对一和一对多的流量之外，CNN 工作负载还引入了多对一的流量。作为片上通信的事实上的标准，片上网络 (NoC) 可以支持各种单播和多播流量。对于多对一流量，采用重复单播，这不是一种有效的方式。在本文中，我们建议在基于网格的 NoC 上使用收集数据包，采用输出静态收缩阵列来支持多对一流量。收集数据包将从最终通向目的地的中间节点收集数据。该方法使用从 AlexNet 和 VGG-16 的卷积层生成的流量跟踪进行评估，与重复单播方法相比，延迟和功率有所提高。