K-means is a pivotal clustering algorithm widely accelerated through specific architectures to mitigate the computational load. Nevertheless, existing solutions lack a comprehensive co-design approach for optimization. By combining AI engines with the reconfigurability of FPGAs, this research focuses on Versal systems and proposes a co-design process that addresses both software algorithm optimization and hardware bottleneck analysis. The proposed accelerator achieves a speedup of up to 21.41× compared to state-of-the-art solutions.

A Hardware/Software Co-Design Approach for Versal-Based K-means Acceleration

Cabai, Eleonora;Sorrentino, Giuseppe;Santambrogio, Marco Domenico;Conficconi, Davide
2025-01-01

Abstract

K-means is a pivotal clustering algorithm widely accelerated through specific architectures to mitigate the computational load. Nevertheless, existing solutions lack a comprehensive co-design approach for optimization. By combining AI engines with the reconfigurability of FPGAs, this research focuses on Versal systems and proposes a co-design process that addresses both software algorithm optimization and hardware bottleneck analysis. The proposed accelerator achieves a speedup of up to 21.41× compared to state-of-the-art solutions.
2025
2025 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW)
File in questo prodotto:
File Dimensione Formato  
KMeans_RAW25_onepage.pdf

accesso aperto

Descrizione: post print
: Post-Print (DRAFT o Author’s Accepted Manuscript-AAM)
Dimensione 116.29 kB
Formato Adobe PDF
116.29 kB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11311/1297543
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 1
  • ???jsp.display-item.citation.isi??? 0
social impact