Bringing Auto-Tuning to HIP: Analysis of Tuning Impact and Difficulty on AMD and Nvidia GPUs

Milo Lurati, Stijn Heldens, Alessio Sclocco, Ben van Werkhoven*

*Corresponding author for this work

Research output: Chapter in Book / Report / Conference proceedingConference contributionAcademicpeer-review

Abstract

Many studies have focused on developing and improving auto-tuning algorithms for Nvidia Graphics Processing Units (GPUs), but the effectiveness and efficiency of these approaches on AMD devices have hardly been studied. This paper aims to address this gap by introducing an auto-tuner for AMD’s HIP. We do so by extending Kernel Tuner, an open-source Python library for auto-tuning GPU programs. We analyze the performance impact and tuning difficulty for four highly-tunable benchmark kernels on four different GPUs: two from Nvidia and two from AMD. Our results demonstrate that auto-tuning has a significantly higher impact on performance on AMD compared to Nvidia (10x vs 2x). Additionally, we show that applications tuned for Nvidia do not perform optimally on AMD, underscoring the importance of auto-tuning specifically for AMD to achieve high performance on these GPUs.

Original languageEnglish
Title of host publicationEuro-Par 2024: Parallel Processing
Subtitle of host publication30th European Conference on Parallel and Distributed Processing, Madrid, Spain, August 26–30, 2024, Proceedings, Part I
EditorsJesus Carretero, Javier Garcia-Blas, Sameer Shende, Ivona Brandic, Katzalin Olcoz, Martin Schreiber
PublisherSpringer Science and Business Media Deutschland GmbH
Pages91-106
Number of pages16
Volume1
ISBN (Electronic)9783031695773
ISBN (Print)9783031695766
DOIs
Publication statusPublished - 2024
Event30th International Conference on Parallel and Distributed Computing, Euro-Par 2024 - Madrid, Spain
Duration: 26 Aug 202430 Aug 2024

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume14801 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349
NameEuropean Conference on Parallel Processing
PublisherSpringer
Volume2024

Conference

Conference30th International Conference on Parallel and Distributed Computing, Euro-Par 2024
Country/TerritorySpain
CityMadrid
Period26/08/2430/08/24

Bibliographical note

Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.

Keywords

  • Auto-tuning
  • CUDA
  • GPU Programming
  • HIP

Fingerprint

Dive into the research topics of 'Bringing Auto-Tuning to HIP: Analysis of Tuning Impact and Difficulty on AMD and Nvidia GPUs'. Together they form a unique fingerprint.

Cite this