RE: Cat Box

From: Clay Cowgill <ClayC_at_diamondmm.com>
Date: Tue Jul 21 1998 - 13:38:57 EDT

> > I was going to ask that... What are your logic requirements? Is it
> > mostly combinatorial or clocked register stuff or do you need lots
> of
> > memory elements?
>
> Lots of memory elements. Off the top of my head, there is a
> 16-bit latch, an 8-bit latch, a 16-bit counter, a 16-bit shift
> register, and I think maybe another 16-bit latch and another 8-bit
> latch. Another counter or two, but not necessarily 16-bit.
>
Hmmmm. I bet 2-3 MACH211's would do it. For the MultiPac I have 15
bits of latch, a three bit counter and a bunch of combinatorial stuff
fitting in a 111 with room to spare (I ran out of I/O pins first). With
a 211 I could have included 2 more 16 bit counters in there without
problems... Sounds do-able, but I'd be inclined to make it work on the
435 and maybe just cost reduce it if you wanted to later...

> Not much other logic. The counters can be done with T
> flip-flops, and the MACH series stuff can configure its macrocells
> that way. Most of the outputs are tri-stated, so there is some
> tri-state enable logic. Block diagram wise, there's not much too it,
> really.
>
The fitter/optimizer that AMD ('er, Vantis) did for the MACH1-2 series
seems really good. When Synario takes a first pass at my design it
bitches up a storm that the "macrocells exceed device capacity", but
when the Mach fitter runs it packs it in with like 15% macrocells free
(so that's what...4 free or something?)

> I suppose it's possible to "bit-slice" the design into some of
> those...
>
Yeah. Or just do it for the 435 and then if you ever switch to Synario
you can build a schematic part out of the old design and just let the
fitter figure out device partitioning...

-Clay
Received on Tue Jul 21 12:39:16 1998

This archive was generated by hypermail 2.1.8 : Fri Aug 01 2003 - 00:31:25 EDT