Moving your AI training jobs to LUMI: A Hands-On Workshop

Result code in IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F61989100%3A27740%2F24%3A10256602" target="_blank" >RIV/61989100:27740/24:10256602 - isvavai.cz</a>

Result on the web

<a href="https://events.it4i.cz/event/285/" target="_blank" >https://events.it4i.cz/event/285/</a>

DOI - Digital Object Identifier

—

Result language

angličtina

Original language name

Moving your AI training jobs to LUMI: A Hands-On Workshop

Original language description

The two-day workshop, “Getting Started with AI on LUMI,” introduced participants to the capabilities of the LUMI supercomputer for artificial intelligence applications. It was tailored for individuals transitioning from smaller-scale computing environments to LUMI’s robust, GPU-intensive platform.Participants brought their own AI training scripts and received personalized support to adapt and run them on LUMI’s advanced GPU system, learning to leverage both single and multiple GPUs effectively.The workshop covered the LUMI-G architecture for AI training, including SLURM, ROCm, the Lustre/LUMI-O file systems, and the Slingshot 11 interconnect. Attendees learned to use existing AI containers, build custom containers with cotainr, monitor GPU efficiency, distribute workloads across multiple GPUs within a LUMI-G node, and optimize AI training processes on LUMI.

Czech name

—

Czech description

—

Type

O - Miscellaneous

CEP classification

—

OECD FORD branch

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Project

—

Continuities

—

Publication year

2024

Confidentiality

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů