Re: [REGRESSION] drm/sched: FAIR policy causes serious performance degradation at max GPU load on 9070XT

[Philipp Stanner <phasta@mailbox.org>]

On Wed, 2026-08-12 at 13:29 +0100, Tvrtko Ursulin wrote:
> Not sure if you are calling me out here or not, since apart from a 
> direct mention, the lore link above is also a reply to my email, so for 
> the record, I did give my ack for those patches. Plus I spent time 
> testing them. So if you were in fact calling me out, I don't see how 
> that is warranted. And on the wider topic I also never shied away from 
> the non-glamorous work.

I was addressing what I perceive to be the general problems within DRM
– I still believe the general points are valid, but want to stress that
it's a collective problem in which we all play a role, though no one is
the protagonist. I mentioned the media outlet (which quoted your name)
as part of the problem because there are external cheerleaders for
performance (whose unconditional priority is *the* reason for the
design mistakes in drm_sched) and features.

Nevertheless, I apologize for the rough tone and want to emphasize that
you're indeed the most engaged contributor and have improved many
things. The unit tests alone are worth their weight in silver.

The wider, problematic patterns in DRM might be worth addressing with
the appropriate tenderness in a suitable format; I will try to collect
my thoughts on that.


Regards,
Philipp