COIN: Communication-Aware In-Memory Acceleration for Graph Convolutional Networks
arXiv:2205.07311 · doi:10.1109/JETCAS.2022.3169899
Abstract
Graph convolutional networks (GCNs) have shown remarkable learning capabilities when processing graph-structured data found inherently in many application areas. GCNs distribute the outputs of neural networks embedded in each vertex over multiple iterations to take advantage of the relations captured by the underlying graphs. Consequently, they incur a significant amount of computation and irregular communication overheads, which call for GCN-specific hardware accelerators. To this end, this paper presents a communication-aware in-memory computing architecture (COIN) for GCN hardware acceleration. Besides accelerating the computation using custom compute elements (CE) and in-memory computing, COIN aims at minimizing the intra- and inter-CE communication in GCN operations to optimize the performance and energy efficiency. Experimental evaluations with widely used datasets show up to 105x improvement in energy consumption compared to state-of-the-art GCN accelerator.
arXiv commentIEEE Journal on Emerging and Selected Topics in Circuits and Systems (2022)
References in corpus (10)
- Hierarchical Graph Representation Learning with Differentiable Pooling
- PyTorch-BigGraph: A Large-scale Graph Embedding System
- Deep Gaussian Embedding of Graphs: Unsupervised Inductive Learning via Ranking
- DeepScaleTool : A Tool for the Accurate Estimation of Technology Scaling in the Deep-Submicron Era
- A Framework for General Sparse Matrix-Matrix Multiplication on GPUs and Heterogeneous Processors
- SIAM: Chiplet-based Scalable In-Memory Acceleration with Mesh for Deep Neural Networks
- Impact of On-Chip Interconnect on In-Memory Acceleration of Deep Neural Networks
- GRIP: A Graph Neural Network Accelerator Architecture
- COIN: Communication-Aware In-Memory Acceleration for Graph Convolutional Networks
- Rubik: A Hierarchical Architecture for Efficient Graph Learning