On Tue, Nov 05, 2002 at 10:32:03AM -0500, 
[EMAIL PROTECTED] wrote:
> Well, I have built a new kernel (and disabled dual head support in the
> MGA framebuffer) and included the new vsync patch from the DirectFB
> "patches" directory -- this one including vsync interrupt support for
> CRTC2, but cpu usage is still 4x as high as compared to the mga driver.

Turns out I'm not kernel developer material yet :( The patch is broken.
I'll submit a fixed patch ASAP and I already put it up on my website.
Hopefully this new patch is correct. At least it works for me (TM).

> So some questions.  Should this be expected?  If not what could
> possibly be causing it?  Is there BES support on the CRTC2 using this
> new matrox-crtc2 mode?  I am playing a 320x240 mpeg1 with the "-fs"
> mplayer option, so perhaps the software scaler is being used?  But
> even if I don't use the "-fs" option and the picture on the television
> is shown at only 320x240, it still uses upwards of 60% CPU.

Some of it might be due to the broken patch. Lot of it is due to clearing
the sub-title surface even when it's not needed. I've just submitted
a new vo_dfbmga patch to mplayer-dev which implements direct rendering and
also avoids the sub-picture clearing. Hopefully these tohether will reduce
your CPU iusage. Oh and if you want to wait for vsync between frames
you'll have to pass -vsync to mplayer. This use to be the only possible
behavior but I made it configurable just in case people want to control it
themselves.

-- 
Ville Syrj�l�
[EMAIL PROTECTED]
http://www.sci.fi/~syrjala/


-- 
Info:  To unsubscribe send a mail to [EMAIL PROTECTED] with 
"unsubscribe directfb-dev" as subject.

Reply via email to