Ubuntu will use ntpd-rs (and benchmark it against chrony)

21

Ah ok, nothing else running, but that is my point, my server is running a lot more stuff at the same time.
As such I want a stable-constant scheduling.

What happens if you run an CPU-benchmark at the same time? That will put pressure on the CPU’s and should affect the difference.

I’m pretty sure I’m not the only one that runs just NTP on it, but a bunch more stuff.

I’m not so much interested in max-performance, but stable-steady performance, where other stuff has little impact on chrony.

We know the load now, with an idle machine, but how behave the different kernels idle versus pressure? I find that important. As I suspect the RT-kernel to do better when it’s not idle.

22

But that is what one aspect of this thread is about: Benchmarking ntpd-rs against other NTP server implementations.

Optimizing NTP performance in the sense of accuracy etc. is quite relevant, too. I suggest opening a separate thread for that so it is more easily found by people interested, rather than having it hidden under a topic title that indicates a different topic focus.