[Intel-gfx] [PATCH 2/2] drm/i915: add async flip support on gen7

Jesse Barnes jbarnes at virtuousgeek.org
Wed Oct 31 20:24:38 CET 2012


On Wed, 31 Oct 2012 10:47:03 -0700
Eric Anholt <eric at anholt.net> wrote:

> Jesse Barnes <jbarnes at virtuousgeek.org> writes:
> 
> > Signed-off-by: Jesse Barnes <jbarnes at virtuousgeek.org>
> 
> So, these days if you go to render to a buffer that's been flipped from,
> your rendering gets stalled (since the buffer may still be scanned out
> when the rendering starts happening), which is the thing we need to
> avoid.  I don't see this patch series avoiding that in the async case.

The currently displayed buffer should be marked idle shortly (i.e.
when the parser gets to your new flip command) after you queue your
async flip.  So the delay should be much shorter than with today's
sync'd flips, but may still be an issue if you have rendering queued up
before the flip.  We can certainly get rid of that...

Or we could just fix our triple buffering to do what you want and get
the best of both worlds.

-- 
Jesse Barnes, Intel Open Source Technology Center



More information about the Intel-gfx mailing list