Improving the Performance of a NoC-based CNN Accelerator with Gather Support
This addresses data movement bottlenecks for designers of CNN accelerators, but it is an incremental improvement focused on a specific architecture.
The paper tackled the inefficiency of handling many-to-one traffic in many-core CNN accelerators by proposing gather packet support in mesh-based NoCs, resulting in improved latency and power compared to repetitive unicast methods as evaluated on AlexNet and VGG-16 convolution layers.
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.