Back to Search Start Over

Evaluating low-level software-based hardening techniques for configurable GPU architectures.

Authors :
Goncalves, Marcio M.
Condia, Josie E. Rodriguez
Reorda, Matteo Sonza
Sterpone, Luca
Azambuja, Jose Rodrigo
Source :
Journal of Supercomputing. Apr2022, Vol. 78 Issue 6, p8081-8105. 25p.
Publication Year :
2022

Abstract

The high processing power of GPUs makes them attractive for safety-critical applications, where transient effects are a major concern, and resilience must be enforced without compromising performance. Configurable softcore GPUs are a recent technology that allows detailed reliability assessment capable of bringing directions to the design of reliable GPU applications. This work investigates the reliability of the register files and the pipeline of a softcore GPU under radiation-induced faults. It proposes software-based fault tolerance techniques to mitigate errors. Faults are simulated at the register transfer level in four case-study algorithms, and the Architectural Vulnerability Factor (AVF) and Mean Workload to Failure (MWTF) are checked over different GPU configurations. Results indicate that software-based techniques efficiently reduce AVF. In terms of MWTF, results show that the best cases depend on an optimized balance between GPU configuration, application runtime, and AVF. [ABSTRACT FROM AUTHOR]

Details

Language :
English
ISSN :
09208542
Volume :
78
Issue :
6
Database :
Academic Search Index
Journal :
Journal of Supercomputing
Publication Type :
Academic Journal
Accession number :
156108751
Full Text :
https://doi.org/10.1007/s11227-021-04154-z