Join us for a continuation of the 2:10 session on Grace Hopper Nodes. This session will focus more on the System Administration of Grace Hopper nodes.
The NVIDIA GH200 Grace Hopper Superchip combines the NVIDIA Grace and Hopper architectures into a single CPU+GPU coherent memory model, with the ability to greatly accelerate data-intensive GPU workflows. A number of RMACC institutions recently acquired GH200 nodes. This thirty-minute "Birds of a Feather" will enable participants to discuss their early experiences with the GH200, or to learn from others' experiences if they are considering the Grace Hopper architecture. Both systems-facing and user-facing aspects will be discussed.