Optimizing the Analysis and Evaluation of Logic Simulation Workloads in HPC Systems

The increasing transistor density and the data-intensive nature of modern applications impede the development of exhaustive validations and reliability estimations in feasible times. Traditional evaluation strategies use simulation-based Fault Injection (FI), demanding considerable computational pow...

Full description

Saved in:
Bibliographic Details
Published in2023 IEEE 17th International Conference on Application of Information and Communication Technologies (AICT) pp. 1 - 6
Main Authors Sierra, Robert Limas, Guerrero-Balaguera, Juan-David, Condia, Josie E. Rodriguez, Reorda, Matteo Sonza
Format Conference Proceeding
LanguageEnglish
Published IEEE 18.10.2023
Subjects
Cluster Management
Distributed Computing
High performance computing
High-Performance Computing (HPC)
Information and communication technology
Instruction sets
Logic Simulation
Mathematical models
Performance evaluation
Reliability engineering
Throughput
Online AccessGet full text

Cover

More Information
Summary:The increasing transistor density and the data-intensive nature of modern applications impede the development of exhaustive validations and reliability estimations in feasible times. Traditional evaluation strategies use simulation-based Fault Injection (FI), demanding considerable computational power. Thus, computation in High-Performance Computing (HPC) systems has become relevant. However, most FIs workloads follow sequential behaviors, impacting their effectiveness and efficiency in HPCs. This work proposes a method to analyze FI workloads and exploit HPCs. The technique uses container managers to boost the throughput and performance of sequential FI workloads used in dependability analyses. Our results demonstrate that Logic Simulation Workloads performance can improve up to 600 times.
ISSN:2472-8586
DOI:10.1109/AICT59525.2023.10313156