Minutes of Jan 15th AMS Committee Conference Call


Subject: Minutes of Jan 15th AMS Committee Conference Call
From: Srikanth Chandrasekaran (schandra@asc.corp.mot.com)
Date: Mon Jan 14 2002 - 19:46:39 PST


Attendees: Jon, Martin (Cadence), Kevin (NSC), Sri, Graham (Motorola)
Date: 14th Jan, 3:30pm, US PST.

* Status on "Accessing of discrete nets in continous context" proposal
  - The proposal would be sent by the end of the week. Just required some
    clarifications as to what was discussed when this issue was taken up about
    6 weeks back.

Action (Sri): To complete write up and post it by end of week.

* Issue with using port names for instantiating digital primitives
  - Currently LRM does not support digital primitives to be instantiated from
    an AMS module.
  - This would require having to define port names for digital primitive ports
    since the "split" case in discipline resolution where a connect module is
    instantiated for each port is created with instance name
    "SigName__InstName__PortName"
  - However digital primitives do not have any concept of port names
  - The LRM would limit the use of these port names
    1. The port names shall only be used for creating unique instance names
    and for accessing a specific instance (say for example overiding parameter
    value using 'defparam' statement).
    2. Named Port Connections shall be disallowed while instantiating a
    digital primitive. Digital primitives shall have to be instantiated using
    ordered connections

Action (Sri): The appropriate LRM section would be modified to capture the
              above two points on digital primitive instantiation

* Disussion on Section 8.7.2, which states that the discipline_list specified
  in connect-resolveTo statement have to be compatible.
  connect <discipline_list> resolveTo <discipline>
  - There was a lot of debate on this with regards to "Port vs Process" bound
    approaches in relation to discipline resolution.
  - It was agreed that there may not be a need for having the restriction.
    Infact it would be useful in cases where a top level module instantiates a
    digital block with an implicit net at top level connected to digital port,
    and the same top level instantiates an analog block where the implicit net
    is connected to a continous port.

Action (Sri/Graham): To send a list of examples and a proposal to remove the
                     restriction on the "connect-resolveTo" syntax

* Inherited Connection Proposal used by Cadence
  - Jon has discussed that with Cadence management with regards to presenting
    this proposal used within cadence with respect to patent etc
  - Looks like there may not be any issues with this, and Jon would be posting
    the Inherited Connections Proposal to the AMS committee

Action (Jon): To send this the AMS committee mailing list

* Proposal on disciplines of analog primitives using compiler directives
  DEFAULT_ANALOG_PRIMITIVE & DEFAULT_DIGITAL_PRIMITIVE
  - Martin & Jon are working on a proposal that gives the ability to override
    the default_discipline on an instance-by-instance basis. However this may
    not use the compiler directives as was discussed earlier
  - Also Kevin discussed the possibility of using similar capability like C++
    override methods using Module overriding.

Action (Martin/Kevin): To post their respective proposals which would be
                       discussed further in the next committee call

* Discussions on some ambigous & incomplete specifications on the LRM
  - Currently it is not clear with connect-resolveTo statement whether the
    resolved disciplines have to be one of the disciplines in the list
    specified
    connect x1 x2 x3 resolveTo y; Should y be one of x1, x2, x3??

  - Also does it imply subsets also fulfil the same rule statements.
    connect cmos1 cmos2 cmos3 resolveTo cmos1;
    Is it implied by the above statement that a connection between cmos2 and
    cmos3 (subset of above) would resolve to cmos1?

  - If there are two rules
    connect cmos1 cmos2 cmos3 resolveto cmos1
    connect cmos1 cmos3 resolveto cmos3
    Which rule applies for a connection between cmos1 & cmos3? The LRM is
    ambigous on this.

Action (Sri): To send a posting on the above issues to the committee and also
              any possible suggestions on resolving the same. This will be
              discussed in the next committee call.

* What happened to Derived Disciplines?
  - There is an example in Section 3.8 which gives an example with derived
    disciplines. However there is no BNF to support this.
  - It was discussed that this part of some legacy verilog-a std definition
    pre LRM1.0 version. The example would be modified in the LRM to remove
    this ambiguity.

Next AMS Committee Call: 28th Jan, 3:30pm, US PST.
-----------------------

cheers,
Sri

--
Srikanth Chandrasekaran
Global Software Group, EDA SBU
Motorola Australia.
Phone: +61-8-8168 3592 Fax: x3501
email: schandra@asc.corp.mot.com



This archive was generated by hypermail 2b28 : Mon Jan 14 2002 - 19:55:00 PST