Compact Parallel Hash Tables on the GPU

Steef Hegeman (Corresponderende auteur), Daan Wöltgens, Anton Wijs, Alfons Laarman

Onderzoeksoutput: Hoofdstuk in Boek/Rapport/CongresprocedureConferentiebijdrageAcademicpeer review

Samenvatting

On the GPU, hash table operation speed is determined in large part by cache line efficiency, and state-of-the-art hashing schemes thus divide tables into cache line-sized buckets. This raises the question whether performance can be further improved by increasing the number of entries that fit in such buckets. Known compact hashing techniques have not yet been adapted to the massively parallel setting, nor have they been evaluated on the GPU. We consider a compact version of bucketed cuckoo hashing, and a version of compact iceberg hashing suitable for the GPU. We discuss the tables from a theoretical perspective, and provide an open source implementation of both schemes in CUDA for comparative benchmarking. In terms of performance, the state-of-the-art cuckoo hashing benefits from compactness on lookups and insertions (most experiments show at least 10–20% increase in throughput), and the iceberg table benefits significantly, to the point of being comparable to compact cuckoo hashing—while supporting performant dynamic operation.

Originele taal-2Engels
TitelEuro-Par 2024: Parallel Processing
Subtitel30th European Conference on Parallel and Distributed Processing, Madrid, Spain, August 26–30, 2024, Proceedings, Part II
RedacteurenJesus Carretero, Javier Garcia-Blas, Sameer Shende, Ivona Brandic, Katzalin Olcoz, Martin Schreiber
UitgeverijSpringer
Pagina's226-241
Aantal pagina's16
ISBN van elektronische versie978-3-031-69766-1
ISBN van geprinte versie978-3-031-69765-4
DOI's
StatusGepubliceerd - 26 aug. 2024
Evenement30th International Conference on Parallel and Distributed Computing, Euro-Par 2024 - Madrid, Spanje
Duur: 26 aug. 202430 aug. 2024

Publicatie series

NaamLecture Notes in Computer Science (LNC S)
Volume14802
ISSN van geprinte versie0302-9743
ISSN van elektronische versie1611-3349

Congres

Congres30th International Conference on Parallel and Distributed Computing, Euro-Par 2024
Land/RegioSpanje
StadMadrid
Periode26/08/2430/08/24

Vingerafdruk

Duik in de onderzoeksthema's van 'Compact Parallel Hash Tables on the GPU'. Samen vormen ze een unieke vingerafdruk.

Citeer dit