On Wednesday 09 February 2005 23:59, Daniel Phillips wrote:
> Hi Timothy,
>
> I think I've got enough of a handle on the design to attempt a budget
> for the dedicated multipliers.  Assumptions:
>
>   - Four clocks to set up each span
>
>   - Interpolate 16 parameters (don't interpolate Z, yes I know it's
>     currently planned otherwise)
>
>   - Assume fog and LOD need 16 bit precision (they might not)
>
>   - Assume color is perspective-correct (some think it's ok to relax
>     this)
>
>   - 8 bit precision is good enough for all blending

By that you mean interpolation in texture lookups I take it, and the stuff in 
the TextureFilter() and Blend() routines?

>   - Need to double check for any overlooked multiplies
>
> So:
>
>    Interpolation correction:      4
>    (16 interpolants / 4 clocks)
>
>    Perspective reciprocal:        2
>    (1 reciprocal x 2 pixels)
>
>    Perspective correction:       30
>    (15 parameters x 2 pixels)
>
>    Not allocated:                 4
>                                 ----
>           Total:                 40  dedicated 18x18 multipliers
>
> This will deliver 2 pixels/clock most of the time, but will stall up to
> 3 clocks on short spans.  Worst case is to fill the screen with single
> pixel vertical strips, which will cut the fill rate by 75%.  Typical
> throughput reduction should be much less, perhaps in the 5-10% range.
> If there really are four multipliers left over, they could be applied
> to cutting the span setup stall to a single clock.
>
> Comments/flames?

I may be able to do the reciprocal without dedicated multipliers at all, but 
I'd already mentioned that in my other post.

Lourens
_______________________________________________
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