GCN-ABFT: Low-Cost Online Error Checking for Graph Convolutional Networks

TL;DR

This paper introduces GCN-ABFT, a low-cost error detection method for GCN accelerators that computes checksums for entire matrix products in a single step, reducing overhead while maintaining accuracy.

Contribution

GCN-ABFT is the first approach to directly compute a checksum for the full three-matrix GCN layer, lowering error detection costs in hardware accelerators.

Findings

Reduces checksum computation operations by over 21% on average.

Maintains fault-detection accuracy through fault-injection analysis.

Applicable to representative GCN applications.

Abstract

Graph convolutional networks (GCNs) are popular for building machine-learning application for graph-structured data. This widespread adoption led to the development of specialized GCN hardware accelerators. In this work, we address a key architectural challenge for GCN accelerators: how to detect errors in GCN computations arising from random hardware faults with the least computation cost. Each GCN layer performs a graph convolution, mathematically equivalent to multiplying three matrices, computed through two separate matrix multiplications. Existing Algorithm-based Fault Tolerance(ABFT) techniques can check the results of individual matrix multiplications. However, for a GCN layer, this check should be performed twice. To avoid this overhead, this work introduces GCN-ABFT that directly calculates a checksum for the entire three-matrix product within a single GCN layer, providing a cost-effective approach for error detection in GCN accelerators. Experimental results demonstrate that GCN-ABFT reduces the number of operations needed for checksum computation by over 21% on average for representative GCN applications. These savings are achieved without sacrificing fault-detection accuracy, as evidenced by the presented fault-injection analysis.