The HammerHAI Training Platform provides HammerHAI users GPU-based high-performance computing resources for AI training tasks in research, development and technology.
Target users
AI Factory HammerHAI users
Description
The HammerHAI training platform, a service provided by KISSKI (AI service centre for critical and sensitive infrastructures), provides GPU-based high-performance computing (HPC) resources for AI training tasks in research, development, and technology. The platform offers HammerHAI users direct access to current NVIDIA A100 and H100 GPUs and supports workflows requiring shell and batch system access.
Installed in Göttingen, Germany, the training platform includes 35 nodes equipped with 4 NVIDIA A100 GPUs each, and 11 nodes equipped with 4 NVIDIA H100 GPUs each. The compute nodes are interconnected through a high-performance InfiniBand network.
How to get access
To use the training platform, users require an up-to-date SSH client, as well as (depending on the application) a local X server. Additional access requirements may apply when the platform is used through other services provided by KISSKI.
To use this service, go to: https://kisski.gwdg.de/leistungen/2-01-04_trainingsplattform-hammerhai/
Service provided by
Contact
Dr. Christian Boehme
GWDG, Deputy Head of the Computing Working Group
