Re:IBIS 3.0 and legal questions

To all:

I believe I have heard some of the silliest arguments
for not distributing IBIS 3.0 models today.

Anyone who has actually dealt with IBIS models from
semiconductor vendors (including Intel and National, among
others) knows that "parser problems" are the least of
our worries. I have personally used hundreds of models, most
of which passed the "golden parser", only to discover
fundamental problems with the underlying data. Often I/V
curves and waveform tables are incorrect. Often the incorrect
units are used. Often clamp curves are extracted improperly.

It is wonderful that the IBIS committee has gotten to the point
of publishing the version 3.0. specification. If we wait for
a 3.0 golden parser ( which, by the way, is an aid to checking
compliance with the specification, not the specification itself),
we in the design community will be waiting for another year.
The argument that the specification will change in the future does
not hold water. All specifications change over time. It is
through "real life" experience with using the results of a
specification that is the impetus to change.

Currently, the design community requires the ability to model
multistage drivers, complex series elements, and modules. There
is no way in IBIS 2.1 to perform these modeling tasks. There is
currently no common public way to distribute this silicon and
packaging information amongst the design community. We have a crisis
for designs trying to achieve 100Mhz and above bus speeds with
complex products in increasingly shortening design times. Yet
members of the IBIS committee (and specifically representatives of
the companies that need to provide this sort of data to customers)
are afraid to release information because the specification (which
they approved) could somehow be imperfect.

Guess what? All specifications are imperfect! However, an imperfect
3.0 specification is far superior to a perfect IBIS 2.1 specification.
What I need as a designer is data from which I can model and
analyze my boards and chips. I choose to receive this data in the
IBIS 3.0 format. I choose to receive it as soon as possible. I
believe that most designers who actually design boards for a living
would rather have the best design information that they can have
right now.

Semiconductor vendors, please get off the stick and provide IBIS
3.0 compliant models today. It would be irresponsible and
inexcusable ( that is, unless you want to impede progress) to do
anything otherwise. If you wish, use the IBIS Disclaimer field
to qualify the preliminary status of your models. The field
is already used to disclaim the suitability of the models for
any purpose anyway.

I hope that other boards designers and consumers of IBIS models
will also speak up. I look forward to your comments.

scottmc@teleport.COM

 
Received on Fri Aug 8 15:26:43 1997