Bio:  Scot Halverson is a Solutions Architect in NVIDIA’s Supercomputing team.  Scot works closely with a number of Department of Energy labs to ensure users of their High Performance Computing clusters are able to make the most of the resources available.  Scot has a MS in Computer Science, and before joining NVIDIA spent ten years developing large scale distributed multi-physics applications at LANL for a variety of supercomputing platforms.

Abstract: We’ll cover the options available to users and administrators for sharing a GPU across multiple processes and multiple users, including Multi-Process Service (MPS) and Multi-Instance GPU (MIG).  We’ll also provide an introduction to NVIDIA developer tools for assessing correctness and performance of applications running on both CPU and GPU.