Instruction-aware learning-based timing error models through significance-driven approximations

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)
33 Downloads (Pure)

Abstract

The adoption of aggressively down-scaled voltages along with worsening process variations, render nanometer devices prone to timing errors that threaten system functionality. The increased vulnerability of nanometer circuits to these errors attracted recent efforts in the development of timing error prediction models using machine learning (ML) methods. However, the majority of such models may be inaccurate, since they either neglect important microarchitecture properties and workload-dependent parameters, affecting timing error manifestation, or are constrained to limited operating areas. In this paper, we propose microarchitecture- and workload-aware ML models for timing error prediction that jointly consider various instruction types as well as all in-flight instructions in a pipeline. Our proposed models are able to predict the exact time and location (i.e., cycle, instruction and bit position) of timing errors with over 98% accuracy across multiple, critical operating regions. To circumvent the increased model complexity due to the considered features, we apply for the first time significance-driven approximations. Evaluation results for various workloads and voltage reduction levels show that our significance-driven precision scaling improves the models’ inference time up to 4.66×, with less than 4% accuracy loss. Finally, we use the proposed model to accurately and realistically inject timing errors during the evaluation of application resiliency. When compared to prior timing error evaluation frameworks that rely on workload-agnostic models, our framework improves the output quality estimation up to 82.6%.
Original languageEnglish
Title of host publicationProceedings of the 40th IEEE International Conference on Computer Design, ICCD 2022
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages455-462
ISBN (Electronic)9781665461863
ISBN (Print)9781665461870
DOIs
Publication statusPublished - 19 Dec 2022
Event IEEE International Conference on Computer Design - Lake Tahoe, United States
Duration: 23 Oct 202226 Oct 2022
Conference number: 40
https://www.iccd-conf.com/Home.html

Publication series

NameIEEE International Conference on Computer Design: Proceedings
ISSN (Print)1063-6404
ISSN (Electronic)2576-6996

Conference

Conference IEEE International Conference on Computer Design
Abbreviated titleICCD
Country/TerritoryUnited States
Period23/10/202226/10/2022
Internet address

Fingerprint

Dive into the research topics of 'Instruction-aware learning-based timing error models through significance-driven approximations'. Together they form a unique fingerprint.

Cite this