Subject: Re: Infinity in Verilog-AMS
From: Srikanth Chandrasekaran (srikanth.chandrasekaran@motorola.com)
Date: Tue Apr 29 2003 - 17:09:26 PDT
Geoffrey,
C defines a few numbers like MAX_DOUBLE, MIN_DOUBLE (in limits.h i think).
Should these be added into the language? If so, would this be sufficient for
device modeling? AMS already defines a standard header called "constants.h"
and so these could be added there. I think the constants.h file is specified
in Annex D of the LRM.
cheers,
Sri
----- Original Message -----
From: "Kevin Cameron x3251" <Kevin.Cameron@nsc.com>
To: <Geoffrey.Coram@analog.com>
Cc: <verilog-ams@eda.org>
Sent: Wednesday, April 30, 2003 3:41 AM
Subject: Re: Infinity in Verilog-AMS
> > From Geoffrey.Coram@analog.com Tue Apr 29 10:29:48 2003
> >
> > Martin -
> > The requirement I have is not related to knowing whether the
> > parameter was set or not. (Although I like the idea of
> > having a way to ask that question.)
> >
> > The question is in regards to device modeling concepts
> > like Early voltages, where the model will do something
> > with 1/VA, and it's important to know whether 1/VA
> > is really zero or whether it's just "close."
> >
> > The Cadence Verilog-A Language Reference suggests that I
> > Change all illegal references to 'inf to a large number
> > such as 1M.
> >
> > However, 1M (meaning 1e6) is not "large" enough for the
> > general case, and probably not even large enough for
> > this example.
> >
> > Perhaps you are meaning that, if a parameter such as the
> > Early voltage is not specified, then I could use the
> > $param_given construct to make the module behave as
> > though VA had a default of infinity.
> >
> > -Geoffrey
>
> I think you want C's "HUGE" value (from math.h). Like I said, there's
> no reason you can't define it in a standard header file (it might be
> platform dependent), so there's no real need for a language extension.
> I would probably just add it to the constants.vams file e.g.:
>
> `define HUGE 3.40282346638528860e+38
>
>
> My own favourite is NaN (Not a Number). I'd like it to be defined so
> that you can initialize real values with it and assign it when converting
> Xs/Zs from digital, and for conditional tests.
>
> Kev.
>
>
>
>
>
> > Martin O'Leary wrote:
> > >
> > > Geoffrey,
> > > It was removed as inf is not a valid number.
> > > Understand the requirement you have - my take is that adding a system
task such as $param_given is a better way to know if a parameter was set or
not.
> > > Thanks,
> > > --Martin
> >
> > --
> > Geoffrey J. Coram, Ph.D. Senior CAD Engineer
> > Analog Devices, Inc. Geoffrey.Coram@analog.com
> > 804 Woburn St., MS-422, Tel (781) 937-1924
> > Wilmington, MA 01887 Fax (781) 937-1014
> >
> >
> >
> >
> >
> >
>
This archive was generated by hypermail 2b28 : Tue Apr 29 2003 - 17:10:29 PDT