Re: Re: FW: XFree86 future

On Sun, Mar 23, 2003 at 01:30:42PM +0100, Lukas Molzberger wrote:
> > > If direct rendered 2D would bring little to no benefit why then is the 
> > > performance of direct rendered 3D under XFree so much better than the 2D 
> > 
> > How can you compare them ? 
> I admit that this is mainly a subjective thing, but in 3D I can load really 
> complex objects and I can move them around and it feels smooth and fluent. 
> The 3D won't let me see any broken or incompletly drawing states. But that's 
> the case on the X11 desktop. Sometimes I can watch the system redrawing a 
> window. The X11 desktop just doesn't feel good.

This isn't primarily about speed, it's about
flicker/tearing/lack-of-animation, as already discussed.

What makes it harder in the X11 case vs. the 3D/direct-rendering case
is not network protocol vs. direct-rendering, but the fact that
multiple applications are involved that need to be synchronized.  If
you are doing direct rendering, you still need some way to do that.

This is a completely solveable problem.

> That's not only a XFree problem but a general problem of X11. I've
> used several X11 implementations (in Irix, AIX, Solaris) and it's
> always the same.

As far as I know, all these X11 implementations are more or less from
the same codebase, so unsurprising. ;-) But even so, the protocol
needs extending to allow apps to coordinate double buffering and such,
you can't fix this totally in the server implementation.

> My point is that the client-server model does not only have a performance 
> penalty but more importantly that it also makes X11 as a whole unnecessarily 
> complex. It makes it difficult if not impossible to fix certain issues in 
> XFree and the complexity also scares possible new developers away. Take for 
> example the problem of resizing opque windows. It looks extremly slow and 
> broken even on fast hardware. That's not really a performance problem but a 
> synchronization problem between the WM, the XServer and the application. I've 
> asked about this issue before and everybody seem to agree that it is pretty 
> much impossible to solve this issue with the current X11 protocol. 

But direct rendering doesn't solve that either, at all. What solves it
is an extension that lets us redirect frame and client drawing to
backing store, then pop them both off at the same time when the unit
is ready, or something along those lines.

We also do have a reasonably OK-looking hack in metacity/GTK that uses
the SYNC extension to ensure that we do 1 frame repaint, 1 client
repaint, 1 frame repaint, 1 client repaint, alternating, instead of 
what happens now where sometimes you redo the frame 30 times and repaint
the client only once. (The GTK patch for this hasn't been applied, so
there's no way to try this yourself yet.)

Once you apply this patch you can see some performance issues still,
but the profiler demonstrates that those are caused by image scaling
and text layout, not rendering, in metacity's case.

There's also some suckage from SHAPE when using windows with rounded
corners, but SHAPE sucks in lots of ways and if we get real alpha
channel we can stop using it.

Havoc