On Saturday 05 February 2005 09:05, Timothy Miller wrote:
> On Sat, 5 Feb 2005 02:08:45 -0500, Daniel Phillips 
<[EMAIL PROTECTED]> wrote:
> > On Friday 04 February 2005 19:29, Lourens Veen wrote:
> > > On Friday 04 February 2005 23:21, Daniel Phillips wrote:
> > > > But why can't we solve this with an intermediate queue that
> > > > holds the same number of entries as stages in the iteration?
> > >
> > > Because the producer and the consumer are one and the same piece
> > > of logic
> >
> > I don't believe they are, and even if they were, you'd replicate
> > it. The problem Timothy is talking about (I think) has more to do
> > with how you factor the stages and move data between them.
>
> No, Lourens is right.
>
> But you are correct in that some amount of replication can be done to
> compensate for that.

Hi guys,

Please bear with me, I'm still figuring out how to think about this.  
When I diagramed out my queue scheme, I saw how it just doesn't work. 
So I got rid of the queue and started replicating.  My new idea is to 
have two 16 bit adders, each with two one-clock stages.  One adder 
computes even steps, the other computes odd steps.  The texture pipe 
accepts interpolated results from one or the other, alternating each 
clock.

Did this solve anything?  A crude diagram is attached.  Each box 
executes in one clock.  The thick lines carry all the interpolants 
between stages.  This bit only handles horizontal stepping.  It would 
be nice but not necessary to be able to step vertically to the next 
span in a single clock as well.

I'm sure you've already solved this, Timothy.  However, I need to get a 
more accurate handle on what's possible with this hardware, so here we 
go again.

Regards,

Daniel

Attachment: horizontal.step.dia
Description: GNU Zip compressed data

_______________________________________________
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