Paper Reading: Eyeriss v2: A Flexible Accelerator for Emerging DCNNs on Mobile Devices

1. Introduction

        This paper is published in 2019. Due to the compact or sparse models are different from the traditional large ones in that there is much more variation in their layer shapes and sizes, the researchs design a flexible accelerator, called eyeriss v2, exploits sparsity for performance improvement.

2. Innovation points

        The following is the Eyeriss v2 top-level architecture.        

        1. Propose a novel NoC, called hierarchical mesh, that is designed to adapt to a wide range of bandwidth requirements.

        2. Design a new PE architecture that exploits the sparsity in weights and activations simultaneously

3. Summary

        The Eyeriss v2 specification is shown below.