On Mon, 21 Mar 2005 13:59:34 +0100, Nicolai Haehnle <[EMAIL PROTECTED]> wrote:

> Some hardware support for writing/reading register states into a predefined
> area of video memory could help a lot here, but I don't know if that's
> feasible.
> And even if it is feasible, the video memory management issues still remain.
> They can obviously be reduced by allowing each GPU program to lock memory
> in place so that it will not be moved by the video memory manager until a
> certain part of the program has been executed.

The state information isn't huge.  Perhaps we could have a convention
where each user of the GPU keeps a data structure containing state
information in shared memory.  This is doubly useful.  For one thing,
the client can use the contents of the data structure to decide which
writes NOT to make when generating its commands, and secondly, the
kernel can use that information to context-switch to the client. 
You'll need two copies for that to work, but the amount of information
is small.

Saving a state is bad because of the time required to read it, but if
clients track their state information correctly, it doesn't matter,
because we always have a copy in the host.
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)

Reply via email to