Monthly HPC Café: Profiling and Bottleneck Analysis of AI Workloads with Nsight Systems (April 14, online and watch party)

A piece of sweet pastry on a plate and a cup of coffee placed in front of a retro style keyboard.

The next HPC Café will take place on Tuesday, April 14, 2026, at 4:00 p.m. CEST as a hybrid event. As always, there will be plenty of time to get in touch with your favorite HPC group. We invite you to come to NHR@FAU to enjoy coffee, cake, and computing.

The event starts at 4:00 p.m. CEST with an open coffee chat. The presentation is scheduled for 4:30 p.m.

Slides

Topic: Profiling and Bottleneck Analysis of AI Workloads with Nsight Systems

Speaker: Robert Dietrich, NVIDIA

Abstract: This session presents practical approaches to performance analysis of AI inference and training workloads using Nsight Systems. It covers essential profiling techniques alongside advanced features for capturing and interpreting system-wide traces across the full stack, including CPU, GPU, I/O, and network components. Through two demos, one focused on an AI inference workload and the other on an AI training benchmark; the session demonstrates how profiling insights can be used to diagnose bottlenecks and guide optimization efforts.

Material from past events is available at: https://hpc.fau.de/teaching/hpc-cafe/