LRM Committee meeting minutes - 31 Jan / 01 Feb

Attendees:
Geoffrey Coram, Analog Devices
Marq Kole, Philips
Patrick O'Halloran, Tiburon
Jim Barby, Univ. of Waterloo
Jon Sanders, Cadence
Prasanna Tamhankar, Freescale
Kevin Cameron, Altera
Sri Chandra, Freescale
Martin O'leary, Cadence

* IEEE 2001 syntax merger
  - Graham has integrated the changes/updates made to LRM 2.2 with the syntax merger work done previously. The current version is available from accellera webpage.
  - A (initial) list of changes that will be required for the individual chapters has been sent for the first 3 chapters. The list is reasonably big (some of the changes are minor), but still there is fair amount of work that would be required. Some of the later chapters might need more work.
  - One of the major changes is an effort to document the semantic restrictions in the chapters more clearly.
  - From the current effort estimate it looks like this task will take about 6-9 months of effort.

* Should the AMS LRM be merged with IEEE 2005??
  - IEEE 2005 version is going for balloting in march. This means the language edits are mostly finalized and stable and should be a standard within the next 3-6 months
  - Its estimated that the AMS merge with 2001 syntax will take atleast about 6 months of work; which will result in AMS language again needing to catchup.
  - Most of the committee members felt that it will be good to merge it directly with IEEE 2005 and try to release an AMS version very soon following the IEEE version.
  - Need to analyse the changes done between 2001 vs 2005 to analyse the effort and impact. Its possible that most of the pure digital features will just be imported. There might be some changes in functionality which might impact AMS language.
  - There is general consensus to do that. Its planned to discuss the changes in the two version at the next committee meeting.

* $table_model proposal submitted by Patrick/Martin
  - The proposal addresses splitting of the table_model functionality into two parts - 1. building the table_model and 2. interpolation on the table.
  - There will be arguments defining the source of the data and also addressing the interpolation control strings used.
  - Also looking at the extending the format and analysing the capability of using XML for the input data.
  - The usage of XML for the data allows the user to add attributes to the data, and also support more than one dependent variable for the table_model. It will also allow putting sections into the data file as the format will allow this.
  - Be able to document the format of the data used within XML in the language

* Some concerns were raised with the table_model proposal
  - There was a concern that LRM2.2 has introduced a flat file format and this is being changed in the next revision of the LRM. However, some of the user requested changes are not trivial to address using the current flat file format.
  - Also there was a concern if the language supports vendors to use other formats (within a given set of guidelines defined by LRM). This would not make the table models portable. However the AMS language has lot of precedences where the language supports vendor extensions (analysis type etc)
 - Is it possible to use VPI where the tools provide an interface to access the information from the table_model. The language will define the standardized interface that needs to be supported. Currently AMS doesn't have any support for VPI for analog constructs and also there was a concern about the performance.

  - A list of requirements is planned to be sent to the reflector with feedback from the vendors/groups requesting this feature. Once there is a complete list, this list is planned to be discussed and prioritized. Based on the requirements identified to extend table_models the syntax and the implementation details will be worked out, with regards to the extensions required to current language. An initial list is planned to be posted by Sri, given the current inputs.

Next Call - 14 Feb 2005, 1:30pm Pacific Time.
Proposed discussion items
* IBIS presentation
* Discussion on 2001 vs 2005 syntax changes/additions
(there may not be time to discuss table_models, if not will be taken up in the next call)

--
Srikanth Chandrasekaran
Freescale Semiconductors, Australia
Ph: +61-8-8168 3592 Fax: 3501