On Thu May 11 08:30:04 2023 +0000, Jan Sikorski wrote:
If QueryPerformanceCounter() is too heavyweight, couldn't we just RDTSC and eyeball the threshold? We don't really care about the exact millisecond value, but I guess we could make a slightly more educated estimate by reading the CPU clock speed.
In theory QueryPerformanceCounter() can itself use rdtsc. I think someone (Rémi?) has wip patches for that even. Perhaps upstreaming those would be better than using rdtsc directly.
I suppose the operative word is "if", though.