FPGA Central

Arpad,

You're bringing up two issues here.

The first is a legacy issue - do we change an existing model to use the standard as approved, do we extend the standard to meet the legacy model's requirements, or some mixture of both? I think these issues sort themselves out but can require extra support in the interim. We're working to bring a new class of models and simulators on-line; I think it's reasonable to expect we'll have a few legacy issues to deal with.

It sounds like the .ibs cases you're describing are old files that are in the process of being replaced. I didn't expect the IBIS 5.0 parser to find problems with the models we'd seen, but it did. Some of the problems were simple typos, some were places where data had been omitted. Most of the issues I saw didn't affect simulation results (omitted keywords and missing NA data comes to mind), but they were still worth correcting.

As you say, that's what the parser is for!

Todd.



--

Todd Westerhoff
VP, Software Products
SiSoft
6 Clock Tower Place, Suite 250
Maynard, MA 01754
(978) 461-0449 x24
twesterh (AT) sisoft (DOT) com
www.sisoft.com


----- Original Message -----
From: "Arpad Muranyi" <Arpad_Muranyi (AT) mentor (DOT) com>
To: si-list (AT) freelists (DOT) org
Sent: Monday, February 8, 2010 6:06:22 PM
Subject: [SI-LIST] Re: IBIS-AMI Vendor Support Help

Todd,

I would like to add a few more items to your list on why (and
how) some AMI models are non-compliant.

I don't know if I can mention the company name openly, but there
are some released AMI DLL-s which use a certain parameter format
that is not AMI spec compliant. The answer I was given when I
questioned this was that these models existed before the AMI spec
was finished, and it would be too much work for the IC vendor to
rewrite their models to obey the AMI syntax. I understand that
Walter's proposed AMI BIRD contains changes that will make these
models compliant, but for the time being they will only work with
tools which are willing to support them out-of-spec.

While I agree with your discussion about S-parameters, I would
like to mention that I have seen non-compliant .ibs files in
which there were no S-parameter references and many of the
non-compliant features could have been described by simple,
existing IBIS syntax. Also, I have seen files which do not
pass the parser because they contain obvious spec violations,
such as not having a number in the typical column of various
items, or omitting required keywords and/or subparameters,
or even because of the incorrect spelling of the OS under
the brand new [Algorithmic Model] keyword.

One could call these problems accidental mistakes or typos, but
that's what we have a parser for, right?

Arpad
================================================== =================


-----Original Message-----
From: si-list-bounce (AT) freelists (DOT) org [mailto:si-list-bounce (AT) freelists (DOT) org]
On Behalf Of Todd Westerhoff
Sent: Monday, February 08, 2010 9:25 AM
To: Eric Monteiro
Cc: si-list (AT) freelists (DOT) org
Subject: [SI-LIST] Re: IBIS-AMI Vendor Support Help

All,

....
....

Why would a model ever be non-compliant? There are two main reasons.
The first (and possibly the issue
Eric was raising) is that the simulator vendor may not fully support the

standard. The .ami file is a "control file"
that tells the simulator how to load and execute the .dll, and what
model configuration options are available.
From what I've seen, some vendors don't support this fully, and require

a tool-specific model control file instead.
It's usually the same information in a different format, but it does
require that the user understand the details of
a proprietary modeling format.

The second reason a model might be non-compliant is that is uses
extensions to IBIS-AMI that haven't been
accepted as part of the standard yet. A good example of this is the use

of S-parameter data in place of the
analog model. The 5.0 IBIS spec doesn't readily lend itself to
describing the broadband behavior of analog I/O.
IBM, Cisco and SiSoft proposed a method for dealing with this in
February 2009:

http://www.vhdl.org/pub/ibis/summits/feb09/hawes.pdf

Although the exact syntax for this capability is still in discussion,
there *ARE* models out there that use this
syntax, or something very similar. There are a number of semiconductor
vendors that use S-parameter
data to characterize the analog aspects of their I/O, and there's no
good reason, in my opinion, to compromise
model accuracy because we haven't settled on the final syntax for the
standard. The consequence of this
is that the S-parameter data gets included in a slightly different
fashion between tools, but the purpose
and the S-parameter data are the same. Claiming that a model "isn't
compliant!" and "doesn't work out of the
box!" may make for a good sound bite, but this isn't rocket science,
especially since the proposed method
and syntax have been documented publicly.

....
....

I'm sure the discussion will continue, so I'm going to leave off here
for now.

Todd.

Todd Westerhoff
VP, Software Products
SiSoft
6 Clock Tower Place, Suite 250
Maynard, MA 01754
(978) 461-0449 x24
twesterh (AT) sisoft (DOT) com
www.sisoft.com
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request (AT) freelists (DOT) org with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-request (AT) freelists (DOT) org with 'help' in the Subject field


List technical documents are available at:
http://www.si-list.net

List archives are viewable at:
http://www.freelists.org/archives/si-list

Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu


------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request (AT) freelists (DOT) org with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-request (AT) freelists (DOT) org with 'help' in the Subject field


List technical documents are available at:
http://www.si-list.net

List archives are viewable at:
http://www.freelists.org/archives/si-list

Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu