Optick: C++ Profiler for Games
github.com
github.com
Tracy is still a frame-based profiler, though.
If you want general system-wide profiling more focused on throughput rather than latency, then I've had a good experience generating flame graphs[2] using plain Linux perf.
[1] https://github.com/wolfpld/tracy
[2] https://www.brendangregg.com/FlameGraphs/cpuflamegraphs.html
I did something similar in a few macros myself by creating nested scopes which record the current time on enter and exit. However I'm leveraging the UI built into chromium for the visualization.
https://github.com/ensisoft/gamestudio/blob/master/base/trac...
https://github.com/google-research/tiny-differentiable-simul...