A methodology for comparing optimization algorithms for auto-tuning

Adapting applications to optimally utilize available hardware is no mean feat: the plethora of choices for optimization techniques are infeasible to maximize manually. To this end, auto-tuning frameworks are used to automate this task, which in turn use optimization algorithms to efficiently search...

Full description

Saved in:
Bibliographic Details
Published in:Future generation computer systems Vol. 159; pp. 489 - 504
Main Authors: Willemsen, Floris-Jan, Schoonhoven, Richard, Filipovič, Jiří, Tørring, Jacob O., van Nieuwpoort, Rob, van Werkhoven, Ben
Format: Journal Article
Language:English
Published: Elsevier B.V 01.10.2024
Subjects:
Auto-tuning
Methodology
Optimization algorithms
Performance comparison
Performance optimization
Optimization algorithms
Performance optimization
Performance comparison
Methodology
Auto-tuning
ISSN:0167-739X, 1872-7115
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Adapting applications to optimally utilize available hardware is no mean feat: the plethora of choices for optimization techniques are infeasible to maximize manually. To this end, auto-tuning frameworks are used to automate this task, which in turn use optimization algorithms to efficiently search the vast searchspaces. However, there is a lack of comparability in studies presenting advances in auto-tuning frameworks and the optimization algorithms incorporated. As each publication varies in the way experiments are conducted, metrics used, and results reported, comparing the performance of optimization algorithms among publications is infeasible. The auto-tuning community identified this as a key challenge at the 2022 Lorentz Center workshop on auto-tuning. The examination of the current state of the practice in this paper further underlines this. We propose a community-driven methodology composed of four steps regarding experimental setup, tuning budget, dealing with stochasticity, and quantifying performance. This methodology builds upon similar methodologies in other fields while taking into account the constraints and specific characteristics of the auto-tuning field, resulting in novel techniques. The methodology is demonstrated in a simple case study that compares the performance of several optimization algorithms used to auto-tune CUDA kernels on a set of modern GPUs. We provide a software tool to make the application of the methodology easy for authors, and simplifies reproducibility of results. •Optimization algorithm performance in auto-tuning research is infeasible to compare.•We propose a community-driven methodology for standardized comparison on elapsed time.•This includes novel methods for comparison using an independent baseline.•The methodology aggregates and quantifies results across experiments.
ISSN:0167-739X
1872-7115
DOI:10.1016/j.future.2024.05.021