On 9/15/05, Attila Kinali <[EMAIL PROTECTED]> wrote: > On Wed, 14 Sep 2005 00:04:53 -0400 > Timothy Miller <[EMAIL PROTECTED]> wrote: > > > So, here's the solution: Of all 64 combinations, go through and > > identify which ones are meaningful and which ones are meaningless or > > redundant. Add a thin layer of logic that reduces those combinations > > to a much smaller number. Now, the state machine only has to compute > > that number of possible "next states". (Whether those states are > > numbered in binary or one-hot is something to be determined as a > > result of trying to synthesize it, but I'm going to start with > > binary.) > > Whithout knowing the pci specs, i think this nummber can be > further decreased. If you draw the state machine as a graph > you'll see that there are only a limited number of edges from > one state to the next. This means, that the maximum number > of next states you'll have to compute, is the maximal degree > of a node. > Ofcourse edges that go to the same nodes should be collapsed > to decrease the degree of the nodes. > > The only drawback of this system is, that we need to compute > first what states could come next from the current state, > but that could be hardcoded into a few LUTs.
Yeah. Computing the next state numbers based on the current state is exactly what I wanted to avoid, because it puts more layers of logic between the slow inputs and the state registers. This design creates redundancy and extra logic, but it does it for the purpose of dealing with certain signals in a special way. The idea is to have all next-state info that's based on internal information computed independently of all selection info that's based on slow inputs (and minimize the logic for the latter), and then dove-tail them at the last possible point. As such, the selection logic based on slow inputs has to be dumb and fixed. In fact, for the sake of speed, I may allow more redundancy than is absolutely necessary. _______________________________________________ Open-graphics mailing list [email protected] http://lists.duskglow.com/mailman/listinfo/open-graphics List service provided by Duskglow Consulting, LLC (www.duskglow.com)
