Initial benchmark of Glucose/i965 with expedite

[José Fonseca]

On 10/19/07, Carl Worth <cworth at cworth.org> wrote:
> On Thu, 18 Oct 2007 15:53:04 -0700, Carl Worth wrote:
> > On Thu, 18 Oct 2007 08:10:32 +0000 (UTC), José Fonseca wrote:
> > > I wrote some quick status info on
> > >
> > >   http://www.freedesktop.org/wiki/Software/Glucose
> >
> > Thanks for this. It's helpful for me, and likely for others as well.
>
> So I got it all to compile and started running some benchmarks. I
> tried a new benchmark this time which is the "expedite" program that
> comes with evas.
>
> I've written up more details with a bar chart here:
>
>         http://cworth.org/glucose/i965/first_look/
>
> The short summary is that I'm not seeing any benefit to running
> Glucose over XAA on my i965 machine.
>
> I might have botched things entirely, so I'd be glad if anyone could
> tell me things I could do to verify that I'm actually getting expected
> results.
>
> I did find that one of the tests failed due to a failed symbol in some
> glucose module, (I don't have the details now, as I'm not near that
> machine and expedite sadly requires user interaction to run---I'll
> post more details when I'm back at that machine later).
>
> As always, any feedback is welcome,

Hi Carl. Interesting results indeed. My involvement with glucose so
far consisted in updating and stabilizing the code, so I can't really
comment much about performance. (I used unoptimized code with
debugging options most of the time.)

Nevertheless, it strikes me as odd the excessive resemblance between
Glucose and XAA timings for most tests, since they essentially use
different code paths.

Things to check on .../var/log/Xorg.0.log are:

  AIGLX: Loaded and initialized .../lib/dri/i915_dri.so

  intel(0): Using GLUCOSE for acceleration

If either is missing then you're either using glucose with GL software
rendering, or no glucose at all.

José