RE: IBIS 3.0 and legal questions

I've read through the responses from tool and model suppliers, and the
perspective from a model user is very different. We're actively
pursuing simulation information for devices, memory modules, processor
modules, etc. that can only be accurately represented with v3.0
syntax. Our choices have been limited to the following: get the
information directly in Quad Design format, exert pressure and get it
in a painful combination of Spice and CAD (Allegro) formats, or don't
get it at all.

It's quite interesting to note that Quad XTK models are becoming more
readily available for most of the devices we care about. It's also
interesting to see some of our competitors being forced to switch
simulation tools for this very reason. I'm seeing buffer technologies
continuing to advance, simulation engines advancing along with them,
and the IBIS spec continuing to fall behind. V3.0 catches up to where
we are today, but it will be behind again in very short order.

The v3.0 spec has allowed us to start pushing the "don't get it at
all" cases. Accordingly we're starting to see v3.0 models. I
certainly wouldn't trust v3.0 models from just any vendor, but I some
level of confidence in most of these guys. From a user perspective I
generally don't trust v3.0 models any more than I trust v2.1 models
(assume they're of limited value until proven otherwise). We still
see a surprising number of v2.1 models with various problems (syntax,
questionable data, etc.).

The lack of a v3.0 parser and tool-specific translators is forcing us
to write our own. We're left with no choice.

Mark Leonard
Manager, Signal Integrity & Package Analysis
Compaq Computer Corporation

-----Original Message-----
From: c=US;a= ;p=COMPAQ;dda:ZID=<huq(a)rockie.nsc.com>;
Sent: Friday, August 08, 1997 4:21 PM
To: Leonard, Mark; c=US;a= ;p=COMPAQ;dda:ZID=<ibis(a)vhdl.org>;
Subject: Re: IBIS 3.0 and legal questions

To all:

A lot of hard work has gone into creating the v3.0 spec. A lot
of 'bad/corrupt' models can easily find their way into the
public domain if not parsed properly thru the v3.0 parser(..which
is coming soon..).

I personally think that the IBIS community should not encourage
IBISv3.0 model distribution unless the v3.0 parser is out/debugged
and approved.

If you must create a v3.0 model, do it and contain it between you
and your EDA vendor. Do not pass this to your customer or to
the public yet....An ASCII file without properly parsed will do
more harm than good.

Regards,
Syed
Vice-Chair ANSI/EIA-656
National Semiconductor Corp.

 
Received on Fri Aug 8 15:10:45 1997