Skip to main content

NVIDIA DGX H100 – The gold standard for AI infrastructure

Resource for teachers

NVIDIA DGX H100
NVIDIA DGX H100. Photo from the NVIDIA image bank.
Published Feb 06, 2024

Teachers at EECS have access to the GPU NVIDIA DGX H100.

Access to the resources happens through a JupyterHub instance, ready to be customised according to course requirements.

Users can execute Jupyter Notebook, with access to GPU resources, directly in their browser, similar to Google CoLab.

Some teachers and teacher assistants are already testing the system to get familiar with the environment and understand how to improve it to minimise the impact on first-time users.

The system's main characteristics

The NVIDIA DGX-H100 is the latest iteration of NVIDIA DGX systems, equipped with eight top-notch NVIDIA H100 GPUs, each with 80GB of memory and providing up to 32 petaFLOPS of total FP8 computing power.

More about the characteristics

Access

Users need to be enabled to use the JupyterHub system. Get access by contacting girondi@kth.se