[Mesa-dev] [PATCH] i965/icl: Allow headerless sampler messages for pre-emptable contexts

Anuj Phogat anuj.phogat at gmail.com
Wed Aug 22 18:18:52 UTC 2018


On Mon, Aug 20, 2018 at 2:13 PM Kenneth Graunke <kenneth at whitecape.org> wrote:
>
> On Monday, August 20, 2018 10:26:29 AM PDT Anuj Phogat wrote:
> > On Mon, Aug 20, 2018 at 12:18 AM Kenneth Graunke wrote:
> [snip]
> > > I don't know if people are trying to enable pre-emption during GPGPU
> > > work on pre-Gen11.  If so, that probably will not work, and we'd either
> > > need to avoid headerless messages (gross) or disable preemption (maybe
> > > also bad...)
> > >
> > Why do you think it'll be problem for Pre-Gen11? I don't see a bit disallowing
> > preemption+headerless message for pre-Gen11.
>
> What's the more likely scenario?
>
> Timeline A:
>
> 1. Preemption support was newly added; it didn't work with headerless
> 2. Somebody realized it was important and added a fix to make them work,
>    but added it as a chicken bit that requires manual intervention.
> 3. They enabled the chicken bit by default, so things work right away.
> 4. They remove the chicken bit; the hardware just works.
>
> Timeline B:
>
> 1. Preemption support was added and headerless works perfectly
> 2. Someone breaks headerless but adds a chicken bit to work around it
> 3. Someone enables the fix by default
> 4. They remove the chicken bit; the hardware just works.
>
> IMHO, the first scenario is the most likely sequence of events.
Makes sense. Thanks for explaining.


More information about the mesa-dev mailing list