SuperMUC-NG AI Hackathon on Intel Data Center GPUs


Contents

The aim of this multi-day event is to help users get up and running on the SuperMUC-NG Phase 2 partition, which features Intel Ponte Vecchio accelerators. The first in a series of upcoming Hackathons, this edition will focus on user projects in the Machine Learning and Deep Learning domains. Participants will benefit from direct access to the cluster and hands-on support from LRZ and Intel experts. Together, we'll work on:

  •  Understanding the system's architecture and available resources
  •  Exploring the software stack for Machine Learning and Deep Learning applications
  •  Getting user code running efficiently on the system
  •  Identifying performance bottlenecks and planning optimisation strategies

To ensure a smooth start, a preparatory online dry-run session will be held in advance to familiarise users with cluster access and navigation. The on-site Hackathon will include brief introductory presentations, followed by ample time for hands-on work on participants' project codes in close collaboration with Intel and LRZ experts.

The Hackathon is co-organised by LRZ and Intel.

Lecturers, Support and Organisation Team

  • Intel:
    • Alexander Pöppl
    • Massoud Rezavand
    • Nikolai Solmsdorf
    • Stefana Railean
  • LRZ:
    • Ajay Navilarekal
    • Darshan Thummar
    • Gerald Mathias
    • Navdar Karabulut
    • Volker Weinberg

Agenda

  • 15 October 2025, 14:00-17:00 CEST Dry-Run Session, online
  • 12 – 13 November 2025, 10:00 – 17:00 CET, Hackathon, LRZ Seminarraum 2

Tentative Agenda Dry-Run Session

  • 14:00-14:05 Welcome (Volker Weinberg, LRZ)
  • 14:05-14:20 Round of Introductions
  • 14:20-14:40 Introduction SuperMUC-NG Phase 2 (Navdar Karabulut, LRZ)
  • 14:40-15:40 Overview of AI Framework Status on Intel Platforms (Stefana Raileanu, Intel)
  • 15:40-16:00 Mini Presentations of Teams
  • 16:00-17:00 Q&A

Slides

Mini Presentations by Teams

Useful Links