VHPIMmJul2010 < P10761

P10761 Web>ProjectsArea>ProjectAlignment107608>MeetingMinutesVHPI>VHPIMmJul2010 (2016-01-22, ErnstChristen) (raw view)EditAttach
---++ IEEE P1076.1 Subcommittee on VHPI-AMS - July 22, 2010 8am PDT

<br />%TOC%
---++ Attendees
   * David Smith, champion, note taker
   * Peter Ashenton
   * Yaseen Zaidi - absent
   * Ernst Christen
---++ Agenda
   * Call to order
   * Approval of agenda
   * Review and approve minutes from June 24 meeting
   * Review action items and any content provided from them
   * Discussion
   * Next Meeting
   * AOB
   * Adjourn
---+++ 1. Call to order

David opened the meeting and started presenting the [[%ATTACHURL%/VHPI_SC_-24_June_2010.pptx][slides]]
---+++ 2. Approval of agenda

The agenda was discussed and there were no changes.
---+++ 3. Review and approve minutes from June 24 meeting

Minutes were approved with no objections.
---+++ 4. Review action items and any content provided from them
   * David - Write Peter about a time line for object analysis. 
      * Contacted Peter and we have the object analysis
   * David - Do analysis of VHPI-AMS functions. 
      * No progress
   * David - Keep in touch with Cadence on possible donation. 
      * They are working on passing it through legal.
   * Peter - develop initial analysis of objects that need to be added to existing VHPI. 
      * Material is on the TWiki site and reviewed in this meeting.
   * Ernst - Check with IEEE about the donation rights form and requirements involved in the donation process. Also, continue pursuing a possible donation. 
      * If material published (with copyright notice) need a donation and copyright permission form. Ernst needs to write a letter to use the material - including as is or with modifications. They need to give letter allowing it. There is no way to review without permission to use it but need permission to review it. Catch 22. Ernst working through the issue.
      * Need all copyright permissions when ask for PAR. Need to get permissions for everything from 1076 since 1076.1 is a different standard. Work with Jim Lewis to get handle on donations and permissions required.
   * Ernst - Investigate technical feasibility of writing to quantities and requirement to related use cases - look at !PutValue. 
      * No progress
   * Ernst - Move all presentations at an easy access location in TWiki - break presentations from minutes. 
      * Done
   * Yaseen - Do an analysis of VPI for Verilog-AMS specific tasks. 
      * Presentation provided but not reviewed in this meeting (ran out of time).
---+++ 5. Discussion

Majority of meeting was spent on reviewing Peter's material at: [[P10761/VHPIInfoModelAndAPIAdditions][VHPIInfoModelAndAPIAdditions]].

The following items were discussed:
<ol type="1"> <li>Relatively small addition to vhdl model</li> <li>Natures - similar to types</li> <li>Objects - extend existing for terminal/quantity</li> <li>Ports - information model may make assumptions that ports are only signals - may need to extend for terminal/quantity</li> <li>Will have to deal with callbacks for simultaneous statements. All simulators need partial derivatives - what impact does this have?</li> <li>What does it mean to have access to the "characteristic expression"? It might be to look at all the contributions to the through summation at a node and the across values at a node. What is "participating" at the net. This could be static or dynamic. This supports observation. How to support transformation - adding to a nodes set of contributions. Wait until requirements surface.</li> <li>Do we have callbacks on solver execution? What level of detail of access to the solver? AMS solver is more abstract than digital due to different solvers used. Can we identify common steps that would lead to a solution? At least trigger on before/after solver invocation to get access to quantities. Can be used for limiting by not going into the details of the solver agent. Can this be done without requiring callbacks in a solution that does not actually have these points.</li> <li>How to represent quantity value between 2 solution points. In digital the value is constant. In analog it is a continuous value between the points. It does have a value between the points. Most waveform graphers assume linear. The actual value is dependent on the integration algorithms. We could provide a function that provides a value at any time (not just at a solution).This is related to how a process reads a quantity that does not have a solution by the analog solver at the point in time where it is being read.</li> <li>The question of providing a specification of the waveform between two solutions is one of scope. This is not typically required.</li> <li>Callback on quantity value change - what is a change? Do you provide threshold crossing? What is noise? Range specification?</li> <li>Statement callbacks - the question of when this is done due to iterative solver nature - i.e. declarative.</li> <li>vhpiTimeT vs vhpiRealTimeT - !RealTimeT has much finer resolution (and range) than discrete time. Conversion between the two looses data. Universal time is used to handle this.</li> <li>How do we maintain compatibility with existing functions that use discrete time? Do we have to create a new set of "mirror" functions that handle real time? For example: !GetCurrentTime and !GetCurrentRealTime may be required.</li> <li>Move foreign models to second round of scope.</li> </ol>

After the discussion the question of what next was raised. The are two concurrent tracks being pursued:
   1 Waiting for donation from Cadence to see if it meets our requirements.
   1 Develop our understanding of what is required and how to approach it assuming no donation from Cadence or to use to review donation from Cadence. Within this track we have the following: 
      * Answer questions raised in the above review. David has an action item to enumerate the questions.
      * Review Verilog-AMS approach.
      * Analysis functions in light of object review.
      * Address the !PutValue question or can we modify analog values during simulation time and why do we want to.

---+++ 5. Next Meeting

Thursday August 19, 2010, 7am PDT.

NOTE: The time has been moved 1 hour earlier to make it easier for Peter to attend.
---+++ 6. Action Items:
   * David - Do analysis of VHPI-AMS functions.
   * David - Keep in touch with Cadence on possible donation.
   * David - Develop open questions from notes on today's meeting.
   * Ernst - Continue working through how to accept donation.
   * Ernst - Investigate technical feasibility of writing to quantities and requirement to related use cases - look at !PutValue.
   * Yaseen - Give VPI presentation in next meeting.
---+++ 7. AOB
---+++ 8. Adjourn

Meeting was adjourned at 9:30am PDT.

-- Main.DavidSmith - 2010-07-22
Attachments
I	Attachment	Action	Size	Date	Who	Comment
pptx	VHPI_SC_-24_June_2010.pptx	manage	42.9 K	2010-07-22 - 17:36	DavidSmith
Topic revision: r2 - 2016-01-22 - 22:17:33 - ErnstChristen
P10761