Improving GPU Matrix Multiplication by Leveraging Bit Level Granularity and Compression

dc.contributor.author	Fett, Johannes
dc.contributor.author	Schwarz, Christian
dc.contributor.author	Kober, Urs
dc.contributor.author	Habich, Dirk
dc.contributor.author	Lehner, Wolfgang
dc.contributor.editor	König-Ries, Birgitta
dc.contributor.editor	Scherzinger, Stefanie
dc.contributor.editor	Lehner, Wolfgang
dc.contributor.editor	Vossen, Gottfried
dc.date.accessioned	2023-02-23T14:00:04Z
dc.date.available	2023-02-23T14:00:04Z
dc.date.issued	2023
dc.description.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.	en
dc.identifier.doi	10.18420/BTW2023-49
dc.identifier.isbn	978-3-88579-725-8
dc.identifier.uri	https://dl.gi.de/handle/20.500.12116/40356
dc.language.iso	en
dc.publisher	Gesellschaft für Informatik e.V.
dc.relation.ispartof	BTW 2023
dc.relation.ispartofseries	Lecture Notes in Informatics (LNI) - Proceedings, Volume P-331
dc.subject	GPU
dc.subject	Matrix multiplication
dc.title	Improving GPU Matrix Multiplication by Leveraging Bit Level Granularity and Compression	en
dc.type	Text/Conference Paper
gi.citation.endPage	772
gi.citation.publisherPlace	Bonn
gi.citation.startPage	763
gi.conference.date	06.-10. März 2023
gi.conference.location	Dresden, Germany

Dateien

1 - 1 von 1