Deep Learning Performance Optimization with Profiling Tools

Learn to improve the performance of deep learning models on GPUs using tools such as DCGM, DLProf and NSight Systems. We will walk through various GPU monitoring and profiling tools, from the simple to the advanced. We’ll describe how to use profiling tools and source code instrumentation to assess the performance and identify bottlenecks faced while training ML and DL models. We’ll also discuss commonly faced issues and bottlenecks, their root causes, and typical solutions, illustrating these concepts with some demos and example code.