You can use function entry/exit (FEE) tracing. A optimized implementation should only incur overhead around one hardware timestamp per function call which is g

You can use function entry/exit (FEE) tracing. A optimized implementation should only incur overhead around one hardware timestamp per function call which is generally around 10-100 ns. That is sufficient for benchmarking quantitative differences at the 1000 ns level and even clear qualitative differences at the 100 ns level. The bias is also fairly stable so you can generally even subtract it out with a little bit of care. And really, the microarchitectural state starts dominating at that point anyways, so anything more fine-grained than that starts depending on making sure the surrounding code did not mess up your L2 or branch predictor and such.