On Monday 14 March 2005 14:06, Timothy Miller wrote: > On Mon, 14 Mar 2005 00:22:45 -0500, Daniel Phillips <[EMAIL PROTECTED]> wrote: > > The case for YUV->RGB hardware has to be made a _lot_ more clearly. > > Can somebody state this in terms of algorithms and data paths > > please, instead of just anecdotes? > > It's basically a matrix multiply, but it's really no big deal, > because some of the multiplies are by powers of two or the sum of two > powers of two.
I see your point. Another point is, this hardware could take several clocks to do the job, say 4. I only see one multiplier close enough to a binary digit to do with a shift: http://www.fourcc.org/index.php?http%3A//www.fourcc.org/fccyvrgb.php The remaining 6 multiplies fit easily into two multipliers, with only one of them being a multiply by a constant. So in the light of day, this doesn't look like much hardware at all. Meanwhile back at the ranch, do you have an idea what percentage of the 3S2000 gates are needed for each 8x8 = 8 multiplier? Regards, Daniel _______________________________________________ Open-graphics mailing list [email protected] http://lists.duskglow.com/mailman/listinfo/open-graphics List service provided by Duskglow Consulting, LLC (www.duskglow.com)
