[Openchrome-users] openchrome-users Digest, Vol 10, Issue 1

[Bart Hartgers]

> Date: Mon, 31 Jul 2006 21:58:37 +0200
> From: Benno Schulenberg <bensberg at justemail.net>
> Subject: Re: [Openchrome-users] Curious GL problem
> To: openchrome-users at openchrome.org
> Message-ID: <200607312158.38135.bensberg at justemail.net>
> Content-Type: text/plain; charset=iso-8859-1
> 
> Bart Hartgers wrote:
>> direct rendering is working in the hardware case, but glxgears
>> shows no significant difference in FPS. I get about 140 fps
>> (fullscreen) in both cases.
> 
> Maybe it is still somehow using direct rendering?  Have you tried 
> running 'LIBGL_ALWAYS_INDIRECT=yes glxgears'?  For me that produces 
> about 35 fps (fullscreen), whereas direct rendering gives 115 fps 
> (fullscreen).  But fullscreen is not the normal way to compare fps 
> in glxgears: always use the default window size.  Then indirect 
> rendering produces 275 fps, and direct rendering 795 fps for me 
> (KM400, Athlon 2800+).
> 
> Benno

Hi Benno,

I don't think it is using direct rendering. I removed the via_dri driver
before testing, and glxinfo claims that direct rendering is not working
in that case. But I didn't check with LIBGL_ALWAYS_INDIRECT.

I agree that it is better to use glxgears with the default window size,
but I run ratpoison as a windowmanager (no mouse on my media-PC), so I
end up with fullscreen. And I should still see a difference in fps
between hardware and non-hardware that way.

Groeten,
Bart
-- 
Bart Hartgers - TUE Eindhoven - http://plasimo.phys.tue.nl/bart/contact/