[Intel-gfx] [PATCH v2 3/3] drm/i915/vlv: WA for Turbo and RC6 to work together.

Jesse Barnes jbarnes at virtuousgeek.org
Wed Jan 22 17:34:15 CET 2014


On Tue, 21 Jan 2014 17:18:59 +0200
Ville Syrjälä <ville.syrjala at linux.intel.com> wrote:

> On Mon, Jan 20, 2014 at 06:40:26PM +0530, deepak.s at intel.com wrote:
> > From: Deepak S <deepak.s at intel.com>
> > 
> > With RC6 enabled, BYT has an HW issue in determining the right
> > Gfx busyness.
> > WA for Turbo + RC6: Use SW based Gfx busy-ness detection to decide
> > on increasing/decreasing the freq. This logic will monitor C0
> > counters of render/media power-wells over EI period and takes
> > necessary action based on these values
> 
> Do we have any idea what kind of performance impact this should
> have?

So aside from the code review comments, it sounds like there are two
high level issues:
  1) keeping existing boost code from Chris (as mentioned by Ville)
  2) power measurements vs current upstream

Given that upstream is a bit different than when this code was forked
off, it could be that we don't need this.  Can you sanity check things
by getting some power measurements with and without this patch?  It
looks like 1/3 and 2/3 will be required in any case though.

Assuming 3/3 does show a benefit, the other question is whether keeping
the current turbo boost code makes sense, so you'd have to port those
changes from the gen6_pm_rps_work() from Chris and measure again...

Thanks,
-- 
Jesse Barnes, Intel Open Source Technology Center



More information about the Intel-gfx mailing list