[Intel-gfx] [PATCH] drm/i915: Optimistically spin for the request completion

[Chris Wilson]

On Wed, Mar 11, 2015 at 10:30:13AM +0000, Chris Wilson wrote:
> On Wed, Mar 11, 2015 at 11:13:59AM +0100, Daniel Vetter wrote:
> > Also do you have microbenchmark numbers for something midly ridiculous
> > like a loop of very short batches (enough ofc to cause a bit of delay) and
> > immediately stalling for them? It's definitely an awesome idea given that
> > every other lock and sync primitive does it too.
> 
> I guess I have
> an exciting morning of letting synmark run on one machine in various
> configs.

So looking at the basics on HSW:gt3e, comparing mesa + throttle
improvements and this patch, we see a consistent improvement throughout
the synmark suite, of anything between 4 - 16%. That is actually
reassuring as one of my main worries is that this patch will acerbate
fallout from thermal throttling. However, a quick look at a few games
suggests that any improvement here is well into the noise, if any.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre