Re: [RFC][PATCH 0/3] DynamicHZ: Configuring the timer tick rate at boot time

[Qais Yousef]

On 02/03/25 12:14, Peter Zijlstra wrote:

> The whole preemption thing could probably be replaced with HRTICK (which
> might be suffering from bitrot), but the problem has always been with
> hrtimers being too expensive (on x86). But ideally we'd move away from
> tick based preemption.

I do think HRTICK must be on by default. Precise time slices are important.
Especially if your plans to attach changing of base_slice to
sched_attr->runtime are still happening. People will find lots of surprises
when they ask for 500us runtime but end up running (much) longer, especially
with the default 4ms TICK value (which I just posted a patch to change it).

Can we re-evaluate the HRTICK impact?