GShuttle: Optimizing Memory Access Efficiency for Graph Convolutional Neural Network Accelerators

Abstract
Graph convolutional neural networks (GCNs) have emerged as an effective approach to extending deep learning for graph data analytics, but they are computationally challenging given the irregular graphs and the large number of nodes in a graph. GCNs involve chain sparsedense matrix multiplications with six loops, which results in a large design space for GCN accelerators. Prior work on GCN acceleration either employs limited loop optimization techniques, or determines the design variables based on random sampling, which can hardly exploit data reuse efficiently, thus degrading system efficiency. To overcome this limitation, this paper proposes GShuttle, a GCN acceleration scheme that maximizes memory access efficiency to achieve high performance and energy efficiency. GShuttle systematically explores loop optimization techniques for GCN acceleration, and quantitatively analyzes the design objectives (e.g., required DRAM accesses and SRAM accesses) by analytical calculation based on multiple design variables. GShuttle further employs two approaches, pruned search space sweeping and greedy search, to find the optimal design variables under certain design constraints. We demonstrated the efficacy of GShuttle by evaluation on five widely used graph datasets. The experimental simulations show that GShuttle reduces the number of DRAM accesses by a factor of 1.5 and saves energy by a factor of 1.7 compared with the stateoftheart approaches.

