Quoting Ville Syrj�l� ([EMAIL PROTECTED]):
> Ok here's a new triple buffering patch.
> 
> What's different to the previous patch:
> - I left DSCAPS_FLIPPING as a double buffer only cap. The previous patch
> had DSCAPS_TRIPLE imply DSCAPS_FLIPPING. Not sure which way is better
> though.

We could have DSCAPS_FLIPPING only, but optionally DSDESC_BUFFERS setting the
number of buffers (currently just two or three allowed).

I'm not sure what the most beautiful and logical way is, but the current
API is already sufficient. Maybe we can arrange the bits so that the following
expression is true for double and triple buffering: (caps & DSCAPS_FLIPPING).

> - fbdev support. It's pretty much _untested_ as of yet. I've been running
> XDirectFB w/ desktop-buffer-mode=frontonly and mplayer w/
> single,double,triple buffering for a few days without problems but I
> haven't tested the fbdev double and triple buffering and changing between
> buffer modes.

I'm running it with backvideo, didn't test switching yet.

> - I paid more attention to buffer allocation,deallocation etc.

Ah, you mean the surface_* functions that need a rewrite? ;)

> - I like this one a whole lot more :) I think it could be commited to cvs
> if the fbdev part doesn't break somehow. Comments, testing and code review
> are welcome.

Triple buffering works perfectly here. I tested single buffered
and triple buffered both without vsync.

> I've also attached few patches to mplayer. First one implements new
> command lined options -triple and -notriple. Second one is for dfbmga and
> include alut44 changes, single, double and triple buffering changes and
> some general cleanups.
> 
> - If you use -triple you shouldn't need -vsync.
> - With -double you need -vsync to avoid tearing.
> - Without -double and -triple you get single buffering and obviosly -vsync
> is needed to avoid tearing.
> 
> CRTC2 seems to like triple buffering very much but BES isn't happy with
> it. -double -vsync is much better for BES.

Why should that make a difference? Did you test it with flip-on-sync
or flip-immediate?

-- 
Best regards,
  Denis Oliver Kropp

.------------------------------------------.
| DirectFB - Hardware accelerated graphics |
| http://www.directfb.org/                 |
"------------------------------------------"

                            Convergence GmbH


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

Reply via email to