Re: VIS: pseudo-inputs, primary-inputs and clock

From: Fabio Somenzi (Fabio@colorado.edu)
Date: Wed Mar 13 2002 - 13:30:29 MST

  • Next message: Emil: "ZBDD support"

    >>>>> "HH" == Hess Hodge <hhodge@cray.com> writes:

     HH> Fabio,
     HH> On Wed, 13 Mar 2002, Fabio Somenzi wrote:

    >> Hess,
    >>
    >> >>>>> "HH" == Hess Hodge <hhodge@cray.com> writes:
    >>
     HH> Hello,
     HH> First, here is some background: I'm using VIS-1.4, and I'm using synopsys
     HH> and edif2blif to create a blif file from verilog.
    >>
    >>
     HH> Here are a few VIS questions for you:
    >>
     HH> 1) Once I read in the design, I notice that there are a number of
     HH> pseudo-inputs associated with my registers, their names end in $INIT. So,
     HH> these pseudo-inputs are there to model nondeterminism on the initial-state
     HH> of these registers, right? How can I set these registers to a known
     HH> initial-state?
    >>
    >> Do you have an "initial" statement in your verilog description? I
    >> don't know exactly what the synopsys edif filter will do to such an
    >> initial statement, but that would be the first thing to look at.

     HH> Yes, I tried it with both an "initial" statement in the verilog, and
     HH> without. Apparently synopsys ignores the initial values when writing the
     HH> edif file.

    I guess this is because "initial" is ignored for synthesis.

     HH> According to the "ntk" package documentation, the pseudo-input takes on
     HH> values from a table. Is there any way to manipulate this table to
     HH> effectively set a specific initial state for the design?

    The easiest approach should be to edit the .blif file, replacing each
    "2" in the .latch directives with the appropriate value.

    Plain blif, as opposed to blif-MV, provides no way to say "the initial
    states are 00 and 11 (but not 10 and 01)." Hopefully, this is not
    your case.

    Fabio



    This archive was generated by hypermail 2b30 : Wed Mar 13 2002 - 13:39:55 MST