From the Macro to the Micro: CUDA Developer Tools Find and Fix Problems at Any Scale

Whether you need to understand how your multi-node CUDA workload is scaling across machines or how a GPU assembly instruction is moving through the pipeline, the latest developer tools have new features for you. We'll start with a brief overview of the tools available for free to developers, and then dive deep on the latest features enabling even more insights into application behavior. These include new expert systems for optimizing cluster-scale GPU workloads, profiling the latest NVIDIA Grace Hopper Superchip, and key performance metrics for inline functions. Additionally, debugging tools have added several improvements to make bug detection easier. Learn how to get the exact information you need at the scale you want with the information presented in this session.