Custom Algorithm-based Fault Tolerance for Attention Layers in Transformers

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Titopoulos, Vasileios, Alexandridis, Kosmas, Dimitrakopoulos, Giorgos
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866908461140279296
author Titopoulos, Vasileios
Alexandridis, Kosmas
Dimitrakopoulos, Giorgos
author_facet Titopoulos, Vasileios
Alexandridis, Kosmas
Dimitrakopoulos, Giorgos
contents Transformers and large language models (LLMs), powered by the attention mechanism, have transformed numerous AI applications, driving the need for specialized hardware accelerators. A major challenge in these accelerators is efficiently detecting errors caused by random hardware faults. Traditional algorithm-based fault tolerance (ABFT) techniques verify individual matrix multiplications but fall short in handling the full attention mechanism, particularly due to intermediate softmax normalization. This work proposes Flash-ABFT, a novel method that computes an online checksum across the entire three-matrix product of query, key and value matrices, of an attention layer, including the softmax operation, with a single check. This approach significantly reduces overhead by eliminating redundant checks while maintaining high fault-detection accuracy. Experimental results demonstrate that Flash-ABFT incurs only 5.3% hardware area overhead and less than 1.9% energy overhead, making it a cost-effective and robust solution for error detection in attention accelerators.
format Preprint
id arxiv_https___arxiv_org_abs_2507_16676
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Custom Algorithm-based Fault Tolerance for Attention Layers in Transformers
Titopoulos, Vasileios
Alexandridis, Kosmas
Dimitrakopoulos, Giorgos
Machine Learning
Hardware Architecture
Transformers and large language models (LLMs), powered by the attention mechanism, have transformed numerous AI applications, driving the need for specialized hardware accelerators. A major challenge in these accelerators is efficiently detecting errors caused by random hardware faults. Traditional algorithm-based fault tolerance (ABFT) techniques verify individual matrix multiplications but fall short in handling the full attention mechanism, particularly due to intermediate softmax normalization. This work proposes Flash-ABFT, a novel method that computes an online checksum across the entire three-matrix product of query, key and value matrices, of an attention layer, including the softmax operation, with a single check. This approach significantly reduces overhead by eliminating redundant checks while maintaining high fault-detection accuracy. Experimental results demonstrate that Flash-ABFT incurs only 5.3% hardware area overhead and less than 1.9% energy overhead, making it a cost-effective and robust solution for error detection in attention accelerators.
title Custom Algorithm-based Fault Tolerance for Attention Layers in Transformers
topic Machine Learning
Hardware Architecture
url https://arxiv.org/abs/2507.16676