Analysis of LLM Vulnerability to GPU Soft Errors: An Instruction-Level Fault Injection Study

TL;DR

This paper investigates the vulnerability of large language models to GPU soft errors through an instruction-level fault injection study, revealing how architecture, size, and task complexity influence reliability.

Contribution

It is the first to systematically analyze LLM resilience to soft errors at the instruction level, providing new insights into fault tolerance for these models.

Findings

Model architecture affects error susceptibility.

Larger models exhibit different fault tolerance characteristics.

Task complexity influences error impact.

Abstract

Large language models (LLMs) are highly compute- and memory-intensive, posing significant demands on high-performance GPUs. At the same time, advances in GPU technology driven by shrinking transistor sizes and lower operating voltages have made these devices increasingly susceptible to soft errors. While prior work has examined GPU reliability, most studies have focused on general-purpose applications or conventional neural networks mostly used for vision tasks such as classification and detection. In contrast, systematic analysis of modern large-scale LLMs remains limited, despite their rapid adoption in diverse application scenarios. Given the unique characteristics of LLMs, their resilience to soft errors may differ substantially from earlier models. To bridge this gap, we conduct the first instruction-level fault injection study of LLM inference. Our approach reveals reliability characteristics from multiple perspectives, highlighting the effects of model architecture, parameter scale, and task complexity. These findings provide new insights into LLM reliability and inform the design of more effective fault tolerance mechanisms.