Zum Hauptinhalt springen
  • Intern
  • DINI-Account
Deutsche Initiative für Netzwerkinformation e.V.Deutsche Initiative für Netzwerkinformation e.V.
  • DINI
    • Über uns
    • Gremien
    • Mitglieder
    • Satzung
    • Wahlordnung
    • Geschichte
    • DINI-Logos
  • Arbeitsgruppen
    • DINI/nestor-AG Forschungsdaten
    • Elektronisches Publizieren (E-Pub)
    • Forschungsinformationen & Systeme (FIS)
    • Kompetenzzentrum Interoperable Metadaten (KIM)
    • Künstliche Intelligenz
    • Lernräume
    • AG Personal
    • Videokommunikationstechnologien und ihre Anwendungsszenarien (VIKTAS)
    • ehemalige Arbeitsgruppen
  • Dienste & Projekte
    • DINI-Zertifikat
    • Liste der Publikationsdienste
    • Projekte
  • Veranstaltungen
    • Veranstaltungskalender
    • Aktuelle DINI-Veranstaltungen
    • Vergangene DINI-Veranstaltungen
  • Wettbewerbe
    • Wettbewerb 2025-2026
    • Beendete Wettbewerbe
  • Publikationen
    • DINI-Thesen
    • DINI-Publikationen
    • Mastodon und Blogs
    • DINItus Podcast

Sie sind hier:

  1. Favicon DINICreated with Sketch.
  2. Nachrichten
  3. Nachricht
  4. General-Purpose GPU Hashing Data Structures and their Application in Accelerated Genomics
23.11.2021

General-Purpose GPU Hashing Data Structures and their Application in Accelerated Genomics

A broad variety of applications relies on associative data structures that exclusively support insert, retrieve, and delete operations. Hashmaps represent such a class of effective dictionary implementations.
Properties such as amortized constant time complexity for these table operations as well as a compact memory layout make them versatile data structures with manifold applications in data analytics and artificial intelligence. The rapidly growing amount of data emerging in many scientific fields can often only be tackled with modern massively
parallel accelerators such as GPUs. Numerous GPU hash table implementations have been proposed over the recent years. However, most of these implementations lack flexibility in order to be used in existing analytics pipelines or suffer from significant performance degradation for certain application scenarios. As a more recent
approach, the WarpCore framework aims to alleviate these aforementioned restrictions by placing a focus on both versatility and performance. In
this talk we reflect the key concepts of the WarpCore library and provide a performance evaluation against the state-of-the-art. We further explore how WarpCore can be used for accelerating two bioinformatics applications (metagenomic classification and k-mer counting) with significant speedups. WarpCore is open source software
written in C++/CUDA-C and can be downloaded at
https://github.com/sleeepyjack/warpcore.

Short Bios:

Daniel Jünger is a Ph.D. candidate in the Parallel and Distributed Architectures group at JGU Mainz in Germany. Daniel’s main focus is accelerating bioinformatics applications targeting massively parallel accelerators and designing associated sparse in-memory data structures
such as bloom filters and hash maps. Daniel’s research has been published in the Cluster Computing journal, the prestigious IEEE International Parallel & Distributed Processing Symposium (IPDPS), and
the IEEE International Conference on High Performance Computing, Data, and Analytics (Best Paper Award winner 2020).

Zoom link: https://fau.zoom.us/j/68603607738

  • Kontakt
  • Impressum
  • Datenschutzerklärung