FARe: Fault-Aware GNN Training on ReRAM-based PIM Accelerators

TL;DR

FARe is a fault-aware training framework for GNNs on ReRAM-based PIM accelerators that significantly improves accuracy despite hardware faults, with minimal timing overhead.

Contribution

The paper introduces FARe, a novel fault-aware training framework that effectively mitigates hardware faults in ReRAM-based GNN accelerators, enhancing accuracy and efficiency.

Findings

Restores GNN accuracy by 47.6% on faulty hardware.

Achieves approximately 1% timing overhead.

Outperforms existing fault-tolerance approaches.

Abstract

Resistive random-access memory (ReRAM)-based processing-in-memory (PIM) architecture is an attractive solution for training Graph Neural Networks (GNNs) on edge platforms. However, the immature fabrication process and limited write endurance of ReRAMs make them prone to hardware faults, thereby limiting their widespread adoption for GNN training. Further, the existing fault-tolerant solutions prove inadequate for effectively training GNNs in the presence of faults. In this paper, we propose a fault-aware framework referred to as FARe that mitigates the effect of faults during GNN training. FARe outperforms existing approaches in terms of both accuracy and timing overhead. Experimental results demonstrate that FARe framework can restore GNN test accuracy by 47.6% on faulty ReRAM hardware with a ~1% timing overhead compared to the fault-free counterpart.