Closed screwzm closed 1 year ago
Google benchmark is bad since the state.PauseTiming/ResumeTiming have heavy overhead themselves, hence cuase the result not usable.
no because this can't be done well. Create two benchmarks: one with setup, another one with setup & actual benchmark code, and subtract the first.
Closed in favor of #86
I am wondering whether superb nanobench supports setup that is not part of the benchmar? For my cases, I would like to setup some context before running the actual part that needs to be benchmarked.
Basically I am looking for something like this:
int main() {
}