Re: [PATCH 0/5] sched: Lazy preemption muck

[Mike Galbraith <efault@gmx.de>]

On Thu, 2024-11-07 at 18:21 +0100, Thomas Meyer wrote:
>
> Mike Galbraith <efault@gmx.de> writes:
> > Full per run summaries attached, high speed scroll version below.
> >
> > desktop util 18.1% static voluntary - virgin source
> > desktop util 18.3% static voluntary - +lazy patches
> > desktop util 17.5% lazy             - ditto...
> > desktop util 17.0% lazy
> > desktop util 16.8% lazy
> > desktop util 17.8% full
> > desktop util 17.8% full
> > desktop util 17.8% full
>
> Can you please elaborate a bit more were those values, e.g. 18,1%, come from?
> How to get those? I couldn't find a connection to your raw data.

I use a slightly twiddled perf to profile, ala perf sched record
<whatever load>, perf sched lat to emit the profile, then do the total
runtime accumulation vs competitor runtime arithmetic. Both competitor
and desktop load are selected for maximal hands off consistency, poke
start, watch yet another 5 minutes of BigBuckBunny to make sure the
internet doesn't hiccup during recording.. and done.

> Sorry for asking this probably stupid question,

Nah, the only silly question is one not pondered first.

	-Mike