Guest
Take a deep dive into the NVIDIA Nsight Compute SOL (Speed Of Light) section, one of the first and most important collections of data within Nsight Compute. SOL analysis reveals how your code performs, and device utilization compared to relevant maximums. Learn how SOL analysis helps you find and resolve limitations to the performance of your code. Key takeaways include: ◻️ Capture SOL data using the Nsight Compute command line or GUI tool. SOL is the starting point for Nsight Compute reporting, offering insights into device utilization and performance bottlenecks. ◻️ Navigate through various metrics, including Compute (SM) throughput and memory utilization. ◻️ Explore kernel duration, SM activity, memory traffic, and more, to help find the best opportunities for code optimization. ◻️ Analyze the transition from a 64-bit to a 32-bit floating-point variant in the sample code. Trace a significant reduction in kernel duration and the shift toward more “balanced“ code. Understand how the change in computation affects hardware utilization and overall GPU performance. 00:00 - Introduction 0:26 - What Is SOL? 2:12 - Sample Code 2:45 - SOL CLI Output 3:17 - SOL GUI Output 8:33 - Compute and Memory Throughput 12:41 - SOL Section with Baseline 16:28 - Conclusion This video series will help get you started with NVIDIA Nsight Developer Tools for CUDA. Grow your proficiency with the tools and apply the examples to your own development environment. Or return to specific episodes for a refresher on certain features and functionalities. We walk through analyzing performance reports, offer debugging tips and tricks, and show you the best ways to optimize your CUDA code. The series will focus primarily on Nsight Compute and Nsight Systems. CUDA Developer Tools | NVIDIA Nsight Tools Ecosystem: CUDA Developer Tools | Intro to NVIDIA Nsight Compute: CUDA Developer Tools | Intro to NVIDIA Nsight Systems: Thanks for watching, and stay tuned for more episodes. Learn more about CUDA Developer Tools: Get started with NVIDIA Nsight Compute: Join the NVIDIA Developer Program: Dive deeper and ask questions on the NVIDIA Developer forums: Read and subscribe to the NVIDIA Technical Blog: #CUDA #Nsight #developertools #NVIDIA #HPC #LLM #CUDAtutorials
