Re: Survey - Version 0.2

From: John Shields <John_Shields_at_.....> Date: Thu Feb 15 2007 - 17:20:36 PST · This archive was generated by hypermail 2.1.8 : Thu Feb 15 2007 - 17:21:07 PST

Hi Logie and Kevin,

I think a) is within our charter and b) is not.  I think the boundary
conditions of a) are already separatable into 2 parts:

  One, that which can be defined without requiring or implying changes
to the grammer/semantics of other languages.

  Two, that which can only be defined if grammar/semantic changes are
made to the other languages.

Some will argue to just do One, because it is within SV to deliver a
result.  I prefer Two remain on the table, with the understanding that
its dependencies will have to be brought back to VHDL, SC, and AMS and
standardized there in order to be achievable.

Regarding b), I want to be open to the important problem being
identified, but I think this is the wrong place to solicit input on
it.  I'd prefer it be taken to the DASC/Accellera first.  At the very
least, it should be taken back to the SV chair.  I agree with Logie's
comments about the size of the survey.  I also see a more fundamental
concern.  Asking a number of these kinds of questions will suggest that
we have perhaps lost sight of the subcommittee charter.  That may not
only stifle input, but it could just slow us down in general.

I felt asking people open-ended questions at the end of the survey was
the place for someone to say that what Kevin wants was very important
to them. That is where someone can say that a different kind of
standardization effort is needed. Despite this, there is an early
question in the survey that does refer to multiple simulators,
different languages, and from different vendors.  We discussed it and I
accept it as an explicit suggestion to raise its priority, in deference
to Kevin. That clearly lets responders say they want more than a single
vendor environment ( and others say that is unreasonable and out of
scope for this subcommittee, too).  I think it would be wrong to ask
any more questions than this.  If answers to this one indicates that
people want this, the whole matter should be referred back to the DASC
for consideration.  Another measure is whether there is some will in
the technical community to work on the problem now.  (to say nothing of
business interests). It always starts with a champion and I applaud
Keven taking that role.  I have some patience to see whether Kevin can
generate the needed interest.

If the need arises, the interested parties arise, the DASC accepts a
PAR to take it on,etc. -- great!  Regardless and in the meantime, we
would still continue working on just a) anyway.  We don't need a survey
to recognize that a) is important and primary in our charter.  The
survey is aimed to help us attack it effectively.

Regards, John

Hi Kevin,

Thanks for the clarification.  

I realize that there are two important parts here. 
a) Defining the language interoperability (similar
   to the questions we have framed as a committee)
b) Defining the implementation/tool interoperability
   (the topics that you are raising). 

Although both are equally important we should clearly
separate the two goals. As a committee we can come
up with standards on how we make
the languages interoperable. This is within our charter.

On the other hand any questions on tool interoperability
is beyond the scope of the committee. Tools may choose
to implement part of the proposed standard. 

Moreover the survey is already quite big as we are dealing
with 4 different languages. Adding more questions that
encompass tool interoperability may reduce the number
of responses and may defeat the purpose of the survey

What do others feel?

Thanks

Logie. 

-----Original Message-----
From: Kevin Cameron [mailto:kevin@sonicsinc.com] 
Sent: Wednesday, February 14, 2007 9:38 AM
To: Logie Ramachandran
Cc: sv-xc@eda.org
Subject: Re: Survey - Version 0.2

Logie Ramachandran wrote:

Hi Kevin, 

Thanks for quick response. My comments below. 

Logie.  

-----Original Message-----
From: Kevin Cameron [mailto:kevin@sonicsinc.com] 
Sent: Tuesday, February 13, 2007 3:54 PM
To: Logie Ramachandran
Cc: sv-xc@eda.org
Subject: Re: Survey - Version 0.2

Logie Ramachandran wrote:

Hi Team,

I have incorporated the feedback and created Version 0.2 of 
survey. Please give it a last minute read, and we can send it out
if there is no major changes proposed in the next 2 days. 

I still don't think it's clear whether you talking about a single 
simulator/vendor environment or multiple tools from different vendors 
talking to each other.

LOGIE>> I assume that you are talking about 1.4. This was the wording
LOGIE>> we agreed to in the meeting. Can you propose changes to 
LOGIE>> the current wording

LOGIE>>  1.4      In the context of interoperability, is it important

to

LOGIE>>           have the capability to cosimulate multiple

simulators

LOGIE>>           (supporting different languages) from different

vendors?          

Not so much that question specifically, just the general impression is 
that the questions refer to a co-simulation environment rather than a 
more academic exercise in making the language semantics compatible.

Also, we have the issue that we currently have to generate models for 
multiple SystemC environments (different headers/compilers), and would

much rather be able to ship a single object file that was usable with 
multiple SystemC environments (as well as SV). So maybe a question or 
two about interoperability of versions for compiled IP would be

useful.

LOGIE>>  I am concerned that we are expanding the scope of the survey
LOGIE>>  beyond what was original discussed. This question mainly 
LOGIE>>  addresses the compatibility of SystemC versions as it

pertains

LOGIE>>  to a compiler  (gcc version x).

I'm just raising issues we have. If the result of the survey is that 
no-one else cares about that particular issue (or co-simulation in 
general) then we can go tackle the issue somewhere else. If most people 
do care then we may need to reassess the goals of the committee.

Since we are moving into an era of increasing reuse, we need 
methodologies that support black-box IP from multiple vendors, defining 
the inter-language semantics is only half the solution. I think the 
language issues and the mechanics of co-simulation can be tackled by 
separate subcommittees if we need to do both, so it probably won't 
impact schedule that much, although I'm sure looking at the 
implementation issues will help prioritize the language issues.

Kev.

Somdipta, do you want to attempt changing the format to 
Excel? 

Thanks

Logie. 

--------------------------------
Loganath Ramachandran   
Director, R&D
Verification Group,      
Synopsys
Mountain View, CA 94043

Ph: 650-584-4891
Em: logie@synopsys.com

 >snip<