NVIDIA Nsight Systems
NVIDIA's system profiler correlates threads, CUDA, graphics APIs, OS scheduling, and custom NVTX markers.
Why it’s here
It reveals CPU/GPU bubbles, synchronization waits, and cross-thread pipeline issues beyond single-frame debugging.
Best fit
Testing, performance, accessibility, and diagnostics expose friction before launch.
Before you commit
Some deep metrics require NVIDIA hardware; configure capture overhead and remote-device support per target.