Improving GPU Matrix Multiplication by Leveraging Bit Level Granularity and Compression

Research output: Contribution to book/Conference proceedings/Anthology/ReportConference contributionContributedpeer-review

Contributors

Abstract

In this paper, we introduce BEAM as a novel approach to perform GPU based matrix multiplication on compressed elements. BEAM allows flexible handling of bit sizes for both input and output elements. First evaluations show promising speedups compared to an uncompressed state-of-the-art matrix multiplication algorithm provided by Nvidia.

Details

Original languageEnglish
Title of host publicationDatenbanksysteme fur Business, Technologie und Web, BTW 2023
EditorsBirgitta Konig-Ries, Stefanie Scherzinger, Wolfgang Lehner, Gottfried Vossen
PublisherGesellschaft fur Informatik (GI)
Pages763-772
Number of pages10
ISBN (electronic)9783885797258
Publication statusPublished - 2023
Peer-reviewedYes

Publication series

SeriesLecture Notes in Informatics (LNI), Proceedings - Series of the Gesellschaft fur Informatik (GI)
VolumeP-331
ISSN1617-5468

Symposium

Title20th Conference on Database Systems for Business, Technology and Web
Abbreviated titleBTW 2023
Conference number20
Duration6 - 10 March 2023
LocationTechnische Universität Dresden
CityDresden
CountryGermany

External IDs

ORCID /0000-0001-8107-2775/work/194824066

Keywords

ASJC Scopus subject areas

Keywords

  • GPU, Matrix multiplication