[sv-cc] Minutes from the 28 February 2003 SV face-to-face.


Subject: [sv-cc] Minutes from the 28 February 2003 SV face-to-face.
From: Swapnajit Mittra (mittra@juno.com)
Date: Sun Mar 02 2003 - 23:09:48 PST


        Here are the minutes from the 28 February 2003 SV
        face-to-face sent by David Smith.

+++++++++++++++++++++++++++++++++++++++
SV Meeting Minutes
28 February 2003

Phone:
Simon Davidmann,
Francoise Martinolle,
Connie O'Dell,
Ghassan Khoory
Stu Sutherland showed up near end

Here:
        Kurt Takara - 0-In
        Jay Lawrence - Cadence,
        Surrendra Dudani - Synopsys
        Karen Pieper - Synopsys
        Brad Pierce - Synopsys
        Johny Srouji - Intel
        Dave Rich - Synopsys
        Peter Flake - Synopsys
        Neil Korpusik - Sun
        Faisal Haque - Cisco
        Dennis Brophy - Model Tech
        Kevin Cameron - National
        Swapnajit Mittra - SGI
        Jayant Nagda - Synopsys
        Bassam Tabbara - Novas
        Joe Daniels - Self
        Arturo Salz - Synopsys
        David Smith - Synopsys
        Stefen Boyd - Boyd Tech/IEEE
        Vassilios Gerousis - Infineon

Agenda:
        Introduction
        Committee Updates:
        Break
        Scheduling Semantics
        Technical Presentation
                SV-BC
                SV-EC
                SV-CC
                SV-AC
        Break
        Veristy Patent
        Completion plans for SV 3.1

Action Items:
    1. SV-AC: Evaluate extending templates and features to the rest of =
Verilog.
    2. SV-AC: Asynchronous sequence for event (sequences of events for
            synchronization - $wait_order)
    3. SV-CC: Need complete list of 3.1 support for extern.
    4. SV-CC: Export/Import/Extern resolution (Surrendra wants more =
pragmas!)
    5. SV-AC: needs to propose those items to the other committees=20
        for review (Surrendra).=20
    6. Setup team to address assertion syntax (particularly ; and ()):
            David Rich
            Jay
            Surrendra
            Bassam
            Neil
            Johny
            Joao
        Charter: Make it work with Verilog. Present back to SV.
        Timeframe: 21 March
        Look at SSWG guidelines

Introduction
        Vassilios Presentation

Committee Updates
        SV-BC (Johny) Presentation

        SV-EC (David) Presentation
            Question (Brad Pierce): BNF review
            Response: Stefen is working on it and is working toward
completion
                next week. Review process will be started with new
versions
                asynchronous to the LRM.
                Jayant commented on the need to provide feedback as
vendors work
                on implementing the language

    SV-AC (Faisal) Presentation
        Question (Dave Rich): What about templates and macros being made
            applicable to rest of Verilog?
        Discussion: High value and need to evaluate the impact of adding
in
            future versus now.

        Issue (Peter/Arturo): Overlap between SV-EC/SV-AC on events and
            sequences on event. Verilog events but not really assertions.
            Need to reconcile. It is asynchronous assertion.

    SV-CC (Swapnajit) Presentation

        Question: Name of interface
        Answer: Last suggestion was the Direct Programming Interface
(DPI)...

Break

Simulation Semantics
    The Arturo and Jay show!

    Qustion: Did you look at the evaluation order the within existing
queues.
    Answer: No new determinism for Verilog code was not provided.

    Question (Karen): What is the difference between the preponed and
the postponed.
    Answer: (Jay) When they run, Preponed is at the start of next time,
        postponed is at the end of current time. (Jayant) Need for api at
        both places.
        (Joao) Preponed is when sampling occurs (conceptually).
        (Arturo) Possiblity of adding pre/post around all regions.
        (David) Current one has only some pre/post regions. Will have to
        make clear what we are voting on today.

    Question (Simon): Are the simulation vendors behind this?
    Answer (Jay): Cadence is thrilled with the scheduling semantics.
        (Dennis) We are behind this. (Arturo) Synopsys clearly supports
it.
        (David) There was a unanimous vote from the simulation vendors
        and users supporting this.

    Question (Neil): Why does the #0 not execute in the Active?
    Answer (Jay): This is a #0 skew on the sampling domain.

    Question (Francoise): Why do we do the sampling and drive in the NBA
    region?
    Answer (Jay): We don't. If there is a problem in slide we need to
fix it.

    Vote on version 6 of the document
    Motion: Arturo Vote on acceptance of version of the SSWG document
    Second: Jay
    Against: None
    For: Unanimous
    Passed

Patent discussion:
    Basic tenor is that while SystemVerilog provides some of the pieces =
that
    could be used to create a system that could violate the Verisity =
patent
    it was felt that SystemVerilog itself does not. No notion of =
transition
    coverage. We do not even have the state machine construct.

    Note: there are no lawyers present. Each company must determine =
their
    liability for themselves.

LRM Completion:
    Question (Dave Rich): Then what?
    Answer: Party

    Question (Jayant): They go to both TCC and Board in parallel?
    Answer: Yes. This is a new process.

    Question (Joe): How are the issues with CC going to be frozen by
    the end of the first week when the material is not even in the LRM =
yet.
    Answer: The technical issues should all be known by now and be able =
to
    be resolved.

    Question (Dave): Who are you considering?
    Answer: People from existing participants. Purpose is for =
consistency
    review.

Committee Technical Presentations
    SV-EC review of content (David)
        Taking notes so no questions captured.

    SV-CC review of content (Swapnajit)
        Question (Jay): Is the data layout for the SV data types defined?
        Answer: (Joao) Canonically yes. (Jay) Still issue here.

        Question (Arturo): C code can access only functions and not
tasks?
        Answer: yes

        Question (Jay): Is there a decompiler for the sequences
available?
        Answer: Not yet. Can see some detail but a full decompiler will
come
        later.

        Issue (Dave): Look at putting attributes on aliases. Can create
group =
of
        variables. Can use enumerated types for declaring FSMs.

        Issue (Jay): One of things that should be done is create a real
pragma
        mechanism.

    $Root Issue (Karen)
        Question (Arturo): Is there define before use?
        Answer: Should consider

        Question (David): Are interface and class supported?
        Answer: No reason why not. (Jay) Should include program.

    SV-BC (Karen)
            Change keyword is gone
        Removal of char is ok but not byte
        Assignment as operator and ++/-- stays

        Question (Dave): Use of extern for both defined elsewhere and
        export/import on C interface.
        Answer: Use VPI in quotes as another technique. (David) The
interface
        is a C compatible interface and not a fortran or C++ interface
(except
        in those cases where they are C compatible - not many).

    SV-AC (Surrendra)
        Question (Karen): Is this SV-EC or SV-BC?
        Answer: Talk to Stefen and put into BNF.

        Question (David): Is the problem that you are reluctant to make
        changes in the rest of the language that are not in the
Assertions
        chapter?
        Suggestion (David): Go ahead and make the extension and inform
the
        other committees.

        Example on page 137 was presented to go through sequence syntax.
        Example on page 156 was presented to go through sequence syntax.

        Issue: The ; and () interact to make problematic syntax.

________________________________________________________________
Sign Up for Juno Platinum Internet Access Today
Only $9.95 per month!
Visit www.juno.com



This archive was generated by hypermail 2b28 : Sun Mar 02 2003 - 23:31:05 PST