How to Profile Modern PC Games with NVIDIA Nsight Graphics

PC game performance analysis has become dramatically more complex in the DirectX 12 and Vulkan era. Between asynchronous compute, hardware ray tracing, temporal reconstruction, frame generation, and machine learning-assisted denoising, a modern game’s rendered frame is no longer a simple sequence of rasterized draw calls. If you actually want to understand where your performance is going, you need to move beyond basic performance overlays and into a tool that can expose how work is distributed across the GPU, how shaders behave at runtime, and which stages are truly limiting the frame. That is exactly where programs like NVIDIA Nsight Graphics come in.

In this guide, we’re going to walk you through a practical workflow for profiling GPU performance in modern games with NVIDIA Nsight Graphics, using CD Projekt RED‘s Cyberpunk 2077 as a case study. The focus here is not on debugging rendering correctness, but on understanding performance: where the GPU is spending time, what kinds of workloads dominate in rasterized, ray-traced, and path-traced scenes, and how to interpret the results using marquee Nsight Graphics features, such as GPU Trace Profiler, Shader Pipelines, and Hotspots. Our methodology is based on NVIDIA’s Peak-Performance-Percentage analysis approach, which starts from GPU-level evidence rather than guesses about how the game engine is using the GPU.

What Is NVIDIA Nsight Graphics?

NVIDIA Nsight Graphics is NVIDIA’s…