Improving GPU Matrix Multiplication by Leveraging Bit Level Granularity and Compression
Autor(en):
Zusammenfassung
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.
- Vollständige Referenz
- BibTeX
Fett, J., Schwarz, C., Kober, U., Habich, D. & Lehner, W.,
(2023).
Improving GPU Matrix Multiplication by Leveraging Bit Level Granularity and Compression.
In:
König-Ries, B., Scherzinger, S., Lehner, W. & Vossen, G.
(Hrsg.),
BTW 2023.
Gesellschaft für Informatik e.V..
DOI: 10.18420/BTW2023-49
@inproceedings{mci/Fett2023,
author = {Fett, Johannes AND Schwarz, Christian AND Kober, Urs AND Habich, Dirk AND Lehner, Wolfgang},
title = {Improving GPU Matrix Multiplication by Leveraging Bit Level Granularity and Compression},
booktitle = {BTW 2023},
year = {2023},
editor = {König-Ries, Birgitta AND Scherzinger, Stefanie AND Lehner, Wolfgang AND Vossen, Gottfried} ,
doi = { 10.18420/BTW2023-49 },
publisher = {Gesellschaft für Informatik e.V.},
address = {}
}
author = {Fett, Johannes AND Schwarz, Christian AND Kober, Urs AND Habich, Dirk AND Lehner, Wolfgang},
title = {Improving GPU Matrix Multiplication by Leveraging Bit Level Granularity and Compression},
booktitle = {BTW 2023},
year = {2023},
editor = {König-Ries, Birgitta AND Scherzinger, Stefanie AND Lehner, Wolfgang AND Vossen, Gottfried} ,
doi = { 10.18420/BTW2023-49 },
publisher = {Gesellschaft für Informatik e.V.},
address = {}
}
Sollte hier kein Volltext (PDF) verlinkt sein, dann kann es sein, dass dieser aus verschiedenen Gruenden (z.B. Lizenzen oder Copyright) nur in einer anderen Digital Library verfuegbar ist. Versuchen Sie in diesem Fall einen Zugriff ueber die verlinkte DOI: 10.18420/BTW2023-49
Haben Sie fehlerhafte Angaben entdeckt? Sagen Sie uns Bescheid: Feedback abschicken
Mehr Information
DOI: 10.18420/BTW2023-49
ISBN: 978-3-88579-725-8
Datum: 2023
Sprache:
(en)
(en)
Typ: Text/Conference Paper

