RE: Minutes from meeting

Hi Brian and Jason,

I agree with Jason's point of view which appears as user and market oriented
and shows a will of real use of the standard.

I would like also to rise some issues in defining such standard:

- a standard complexification can lead that all features will never be
strictly implemented by all vendors so we will go to a non-standard (which
products can be fairly said as 100% sce-mi compliant ?).

- an acceptance of any feature of all current emulation solutions (even if
for transactor vendors, it would be easier since any implementation could
fit for standard) would complexify the standard

- a solution oriented will restrict the standard: EDA vendor could
differentiate providing extra mechanism(s) aiming at user view
simplification


It is important to hierarchize the several aspects of the standards (e.g.
using layers if necessary), and thus to be able to precisely dissociate them
(e.g. the aspects of interfacing, modelling, debugging).

At leat one request concerning the interfacing is to provide a full SCE-MI
1.x compatibility.

Regards,

Joseph Bulone


-----Original Message-----
From: owner-itc@eda.org [mailto:owner-itc@eda.org] On Behalf Of Brian Bailey
Sent: Wednesday, March 30, 2005 5:20 AM
To: 'Jason Andrews'; itc@eda.org
Cc: 'Jason Rothfuss'
Subject: RE: Minutes from meeting


        Thanks Jason(s),
            I would also like to use this as a reminder to everyone else
that tomorrow is the deadline to get to me any issues or corrections to the
high level goals and objectives as laid out in slides 11-15 of Duaine's
presentation.
        The lack of feedback indicates your acceptance of them.
        
        Regards,
        Brian
        
        
        -----Original Message-----
From: Jason Andrews [mailto:jason@verisity.com] 
Sent: Tuesday, March 29, 2005 3:25 PM
To: brian_bailey@acm.org; itc@eda.org
Cc: 'Jason Rothfuss'
Subject: RE: Minutes from meeting
        
        Hi Brian,
        
        Here is our feedback on the goals. It comes from Jason Rothfuss
        and myself.
        
        Thanks.
        
        -----------------------------------------------------------------
        
        Verisity Feedback on SCE-MI 2.0 goals as outlined by Duaine on
slides 11-15
        
        
        1. Reference Implementation and Compliance Suite
        ---------------------------------------------------
        Our most important feedback is the goal of the committee to provide
a
        reference 
        implementation of SCE-MI and a test suite to help ensure compliance
by all
        vendors.
        We feel this is the best way to achieve the goals of enlarging the
market
        for EDA 
        vendors and enabling increased development of VIP by users and 3rd
parties.
        
        A free implementation that runs on all popular simulators is a must
to
        spread SCE-MI 
        to a wider audience. The reference may or may not include source
code, but
        it must run
        easily on popular platforms. We don't have complete approval at this
time,
        but we
        would be interested to provide this. 
        
        We feel the goal of SCE-MI is to enable a large set of models for
users that
        run the
        same way on all acceleration/emulation/prototyping systems. This
allows
        users to 
        choose products based on the features and benefits vs. choosing
based on
        available models. 
        If this is not the goal then we should all go back to proprietary
interfaces
        and selling 
        models that only work on our own platforms.
        
        2. Ease-of-Use
        --------------
        We support any activities to make the life of the model developer
easier.
        Enhancements 
        to simplify clocking, the use of variable length messages, and data
        streaming are examples. 
        This supports of the goal of not requiring A+ engineers to develop
models.
        By providing the 
        reference environment discussed in #1 more engineers can learn how
to
        develop models.
        
        3. Languages
        -------------
        We support continued use of C/C++
        Every other language used in verification can easily call and be
called by
        C.
        
        4. Acceleration Subset
        ----------------------
        We do NOT support any effort to define a new subset of Verilog,
VHDL, or
        System Verilog 
        for the purposes of accelerating modeling constructs.
        We feel the current de facto standards are sufficient for engineers
to
        develop models. 
        As new enhancements to the current languages used for design are
made they
        can be 
        adopted for modeling. SCE-MI is an interface and not about
developing
        special synthesis 
        tools for acceleration. Even though we develop such tools in
Verisity (such
        as e synthesis) 
        we don't see this as useful for improving the standard and promoting
        increased usage. 
        In fact, it would likely lead to vendor separation based on
proprietary
        synthesis tools 
        and make interoperability even worse than it is now.
        
        
        
        > -----Original Message-----
        > From: owner-itc@eda.org [mailto:owner-itc@eda.org] On Behalf 
        > Of Brian Bailey
        > Sent: Thursday, March 24, 2005 10:59 AM
        > To: itc@eda.org
        > Subject: Minutes from meeting
        > 
        >         Please find attached the minutes from today's 
        > meeting. Important action item for everyone - all proposed 
        > modification, or acceptance of the goals must be mailed to me by
3/30
        >         
        >         Thanks
        >         Brian 
        >