Re: Model name question

Lynne,

My suggestion is to formalize the concept that a unique model name is
described by the following:

manufacturer.component.model.revision.

... and that simulator vendors should attempt to manage models
down to this level of detail. However, understanding that no vendor
currently is capable of doing this, that this is desired for future designs
with IBIS-X. It should be recommended that vendors document the current
method for model management and acknowledge whatever current
flaws and name space colisions which can occur. (this is not to point
fingers, but rather so that users can correctly accomodate the various
tool "qwirks" and guarantee valid simulations.

An example "qwirk" might occur when multiple files are loaded into
a simulation tool, both of which have the same component name.
This can happen easily when multiple revisions of an IBIS model exist,
or when there are multiple vendors for the same component.

regards,

scott

Lynne Green wrote:

> Hi, Scott and Arpad,
>
> I like the idea of formalizing this. At the
> same time, should we consider allowing
> more than 20 characters is a buffer name?
>
> Cadence uses COMP + MODEL now, and
> also makes the combination unique within
> a simulation (default setting).
>
> - Lynne
>
> At 03:28 PM 7/12/2001 -0700, Scott McMorrow wrote:
> >Arpad,
> >
> >I believe many simulator vendors already identify models
> >internally by the following tuple:
> >
> >(component_name, model_name)
> >
> >I would recommend formalizing this in future IBIS specifications.
> >We need not specify the exact format of the naming convention,
> >but we do need to clarify that a model should be identified by
> >component_name and model_name, since model_names are
> >not unique.
> >
> >If we were all real sticklers, we would require tools to identify
> >a model by the following tuple:
> >
> >(component_name, model_name, revision)
> >
> >Many a time I have been bit by multiple revisions of models laying
> >around ... especially with automatic path searches of model directories,
> >where the last model loaded is the one that is used.
> >
> >regards,
> >
> >scott
> >
> >
> >"Muranyi, Arpad" wrote:
> >
> > > All,
> > >
> > > I would like to find out whether there is a need to include the
> > > component name as part of the buffer names for the [Model]
> > > keyword in an IBIS file. The reasoning goes like this:
> > >
> > > If there are two different IBIS files with buffer names inside
> > > them which are the same (while the electrical characteristics
> > > of the models are different) and these two models are used in
> > > the same simulation, could the tool lose track of which one is
> > > which?
> > >
> > > Do tools make these names unique for simulations to prevent
> > > this from happening so that me as a model maker wouldn't have
> > > to worry about this?
> > >
> > > Thanks,
> > >
> > > Arpad
> > > =================================================================
> >
> >--
> >Scott McMorrow
> >Principal Engineer
> >SiQual, Signal Quality Engineering
> >18735 SW Boones Ferry Road
> >Tualatin, OR 97062-3090
> >(503) 885-1231
> >http://www.siqual.com
> >

--
Scott McMorrow
Principal Engineer
SiQual, Signal Quality Engineering
18735 SW Boones Ferry Road
Tualatin, OR  97062-3090
(503) 885-1231
http://www.siqual.com