On Thursday 10 February 2005 00:15, Daniel Phillips wrote: > On Wednesday 09 February 2005 17:36, Timothy Miller wrote: > > On Wed, 9 Feb 2005 23:23:47 +0100, Lourens Veen wrote: > > > Timothy, what configurations are possible for these RAM > > > blocks? > > > > 36x512, 18x1024, 9x2048, 4x4096. And only the 36x512 case loses us > > the multiplier. > > OK, so forget the precalculated difference idea, it loses on two counts: > 1) won't fit 1024 entries in a block and 2) loses a multiplier. The > first is probably not a big problem but the second is a killer.
512 entries is not enough anyway, so we'll need to use two blocks. If both of them work in 18x1024 mode then we can read 36 bits twice per clock, if we make sure that we read 36 bits from one, and 36 bits from the other. That's easier with a value + difference scheme. But see my other mail. 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)
