EUVIMEDEUVIMED entdecken
Crossref

Benchmarking Optimization Algorithms for Auto-Tuning GPU Kernels

Richard Arnoud Schoonhoven, Ben van Werkhoven, Kees Joost Batenburg

IEEE Transactions on Evolutionary Computation · 2023-06 · 27 (3) · 550-564

DOI: 10.1109/tevc.2022.3210654 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Evolutionary Computation, 2023-06. Verfasst von Richard Arnoud Schoonhoven und 2 weiteren Autor:innen. Die Arbeit wurde bisher 15-mal zitiert (Crossref) und nennt 68 Literaturangaben. Gefördert von: Netherlands Organization for Scientific Research, Dutch Research Council (NWO) in the framework of the NWA-ORC Call.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Recent years have witnessed phenomenal growth in the application, and capabilities of graphical processing units (GPUs) due to their high parallel computation power at relatively low cost. However, writing a computationally efficient GPU program (kernel) is challenging and, generally, only certain specific kernel configurations lead to significant increases in performance. Auto-tuning is the process of automatically optimizing software for highly efficient execution on a target hardware platform. Auto-tuning is particularly useful for GPU programming, as a single kernel requires retuning after code changes, for different input data, and for different architectures. However, the discrete and nonconvex nature of the search space creates a challenging optimization problem. In this work, we investigate which algorithm produces the fastest kernels if the time-budget for the tuning task is varied. We conduct a survey by performing experiments on 26 different kernel spaces, from nine different GPUs, for 16 different evolutionary black-box optimization algorithms. We then analyze these results and introduce a novel metric based on the PageRank centrality concept as a tool for gaining insight into the difficulty of the optimization problem. We demonstrate that our metric correlates strongly with the observed tuning performance.

Förderung

  • Netherlands Organization for Scientific Research
  • Dutch Research Council (NWO) in the framework of the NWA-ORC Call

Zitieren

APA

Schoonhoven, R. A., Werkhoven, B. V., & Batenburg, K. J. (2023). Benchmarking Optimization Algorithms for Auto-Tuning GPU Kernels. IEEE Transactions on Evolutionary Computation, 27(3), 550-564. https://doi.org/10.1109/tevc.2022.3210654

Vancouver

Schoonhoven RA, Werkhoven BV, Batenburg KJ. Benchmarking Optimization Algorithms for Auto-Tuning GPU Kernels. IEEE Transactions on Evolutionary Computation. 2023;27(3):550-564. doi:10.1109/tevc.2022.3210654

BibTeX
@article{Schoonhoven2023,
  author = {Schoonhoven, Richard Arnoud and Werkhoven, Ben van and Batenburg, Kees Joost},
  title = {Benchmarking Optimization Algorithms for Auto-Tuning GPU Kernels},
  journal = {IEEE Transactions on Evolutionary Computation},
  year = {2023},
  volume = {27},
  number = {3},
  pages = {550--564},
  doi = {10.1109/tevc.2022.3210654}
}
Kostenlos starten →