Re: Multi-dimensional arrays?

From: Sri Chandra <srikanth.chandrasekaran_at_.....>
Date: Tue Sep 13 2005 - 15:16:55 PDT

Hi Arpad,

We are trying to fix some of the problems (given by the recent questions that you have posted on the reflector) that you have highlighted. I understand some of the inconsistencies that you have highlighted in the language, and also the inconsistencies that it leads to within different simulation tools vendors.

The AMS LRM as it stands today reflects the 1995 standards which is outdated for most/all the users that are using digital/mixed signal today. As part of the upcoming revision we are trying to merge with 1364-2005 which will resolve some of the highlighted differences. Ofcourse this by no means is a small task given the divergence and try to document correctly the semantic differences between analog and digital for the same construct. At the same time we are trying to address some of the key issues that are required by analog designers today. Having said that, implementations have had to move forward with varying degrees and hence very difficult to judge what is supported where - really depends on the needs of the customers who have been using and driving the tool.

Ofcourse, this is not a great situation to be in, where interoperability between simulation tools are not guaranteed but hopefully with LRM2.3 (1364) and 3.0 (p1800) work we will be able to get there. One might ask why wasnt analog kept upto-date with digital, and about 3-4 years ago the participation and bandwidth on the committee was bit limited and also as Geoffrey mentioned most of the people in the current committee havent been here when the effort started.

regards,
Sri

Mirmak, Michael wrote:

It might be helpful to observe that there is a significant difference
between how various specification committees operate.  Practices in one
may not carry over to others.  This may be happening here, as I believe
Arpad is doing work on IBIS-VerilogAMS links.

We in IBIS actually do have several reflectors and other forums
operating as "specification Supreme Courts," offering model checking and
interpretation of the IBIS rules by experts.  Areas which may be subject
to tool-specific interpretation are pointed out and tool vendors can
always provide additional comments.  We also enforce syntax checking
though a single parser program provided by the committee.  Where
significant disagreement exists on interpretation, the specification
gets revised to clarify the issue.  

We, however, are a very small "pond" and I doubt that a user base as
large as that for Verilog-AMS could be accommodated in the same way.
Differing expectations for how specification teams operate may be
causing confusion here.

- Michael Mirmak
  Intel Corp.
  Chair, EIA IBIS Open Forum

-----Original Message-----
From: owner-verilog-ams@eda.org [mailto:owner-verilog-ams@eda.org] On
Behalf Of Muranyi, Arpad
Sent: Tuesday, September 13, 2005 12:43 PM
To: VerilogAMS Reflector
Subject: RE: Multi-dimensional arrays?

George,

I apologize for irking you again.  It was not my intent.
In fact when I wrote the sentence you quoted I was not
in an angry mood or shouting it, I was writing it in a
begging "tone of voice" which unfortunately didn't come
across the monitor you were reading it through... :-)

I really don't know the size of this email reflector,
so I may be expecting more than there really is.  And
my intension is certainly not to overload this forum
with "personal" questions either.  I usually post a
question here when I feel the LRM is missing something,
or when it is not clear on something, hoping that I
could get an answer, and/or that the LRM will get fixed.

However, I am not looking for and asking for answers about
simulators.  I am trying to find out what is officially
available in the language so that I can write my models
according to the LRM and not according to implementations
in certain tools.  I need to know what the intent of the
LRM is so I can talk to the tool vendors if necessary.
Where else can I go to find out what the official word
is in the LRM?

Sincerely,

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




-----Original Message-----
From: geoffrey.coram@analog.com [mailto:geoffrey.coram@analog.com] 
Sent: Tuesday, September 13, 2005 9:53 AM
To: Muranyi, Arpad
Cc: VerilogAMS Reflector
Subject: Re: Multi-dimensional arrays?

I know I'm the one who ticked you off ... 

To be honest, your original message irked me (again)
when it said

  

Could someone please
answer the question?
    

as though the sole reason for this e-mail list's existence
was to answer your questions, and that people on the list
all know the answer to your question for every simulator in 
existence, and we're not answering you just out of spite.

I wasn't around for the original AMS definition, and I don't
work for any of the simulator vendors; no one on the list
works for all the vendors, so there's simply no one who can
answer your question about what works in every simulator.
It's not that we in the group are "forgetting" the role of
the spec.

...
...
...

So, in summary, I apologize for ticking you off, but please
keep in mind that we're all volunteers here, and we don't
know everything.  The LRM is the spec, and if you think
that it is ambiguous in some way, then by all means, let's
open up a Mantis item and get the committee to work on 
language to clarify it, consistent with how the vendors
have implemented it (to the extent possible).

-Geoffrey


  

Received on Tue Sep 13 15:17:10 2005