Around 14 o'clock on Aug 21, Owen Taylor wrote:

> I have some code for MMX acceleration of common Render operations
> around, that I keep meaning to clean up and give to Keith. :-(

Those would be nice to have, Owen.

> (And there is a lot of optimization of the software fallbacks that can
> be done without reference to MMX... you can get maybe 5x improvement
> over the current code with plain C, and then MMX gives you an extra 50%.)

I'd like to get some help in identifying and optimizing code paths needed 
by applications.  It would also be nice if we could figure out a clean 
framework for plugging these optimizations in; right now, you have to go 
and whack a long sequence of conditional branches for each new special 
case.  That's fraught with peril, but I haven't figured out a 
datastructure that can do the same job as quickly.  

I suspect what we need is a predicate language that describes the
conditions each optimization needs and then preprocess that into a BDD so 
that we can plug new code in without breaking existing bits.

Another separate optimization is to migrate pixmaps in and out of display 
memory so that Render operations needn't be done entirely over the AGP bus.

Keith Packard        XFree86 Core Team        HP Cambridge Research Lab


_______________________________________________
Render mailing list
[EMAIL PROTECTED]
http://XFree86.Org/mailman/listinfo/render

Reply via email to