CASTalk.com

fpgabuilder · Guest

I am wondering if it is possible to reduce switching activity in the
next state logic if I use consensus theorem (term).

Any thoughts group? I am interested in this discussion from an
academic standpoint and not in the fact that most of the tools will
optimize away the consensus term in a synchronous design.

Thanks.
-sanjay

JustJohn · Guest

fpgabuilder · Guest

Hi John,

Sorry for the delay in responding. I am not sure about what you mean
by overlapping logic terms. By consensus term I mean use of the
expression -

ab + a'c = ab + a'c + bc. In this case, bc is the consensus term.

As you probably know, bc prevents any glitches in the output when a
changes. So what I am implying is that the switching activity on the
output is reduced if we use the bc term.

Thanks.
-sanjay

Peter Alfke · Guest

Sanjay, if you are talking about Look-Up-Table logic, there is no
difference between the approaches.
In your case of three inputs, abc, you must define the output for all 8
possible input codes in the Table (or call it a ROM).
Whether you think of "consensus" has no relevance. LUT logic is a
brute-force implementation.
You get no glitch on the output when you change one input, (well, you
may get a change, but no glitch !) and you also get no glitch when you
change two inputs, provided all 4 permutations of these two iputs
create the same output (not true in your specific example).
Peter Alfke, Xilinx Applications

	CASTalk.com Forum Index -> FPGA	All times are GMT
Page 1 of 1