(r6) TDMMtgMinutesJun710 < P10761

P10761 Web>ProjectsArea>ProjectTableDrivenModeling>TDMMtgMinutesJun710 (revision 6) (raw view)~~EditAttach~~
---++ IEEE P1076.1 Working Group - Table-Driven Modeling Project - Telco Metting Minutes of June 7, 2010

<strong>Status: </strong>approved on July 5, 2010.

%TOC%
---+++ Attendees
   * Ernst Christen
   * Joachim Haase (chair, meeting convener)
   * Arpad Muranyi
   * Alain Vachoux (note taker)
---+++ Agenda
   * Call to order
   * Volunteer for minutes
   * Approval for agenda
   * Review plan and refinement
   * Discuss requirements analysis steps and solicit volunteers
   * Next meeting
   * AOB
   * Adjourn
---+++ Call to order on June 7, 2010 at 8am PST (5pm CEST)

Joachim welcomes the participants and starts presenting the [[%PUBURL%/%WEB%/ProjectTableDrivenModeling/1076.1_table_driven_modeling_2010_06_07.ppt][meeting slides]].
---+++ Volunteer for minutes

Alain volunteers to take notes.
---+++ Approval for agenda

Agenda unanimously approved.
---+++ Review plan and refinement

The scope shall include the support of both time and frequency domains. The main difference is that the former uses real values while the latter uses complex values. The DC domain is naturally implied.

Supporting more than one dependent variable in a single lookup function call would be needed for, e.g., transistor models to improve simulation performance. This is however not absolutely necessary as this can be replaced by multiple function calls. Ernst proposes to look at the issue in more details to clearly define the requirements.

The proposed plan is OK. Project telcos once a month. Most of the work through emails.
---+++ Discuss requirements analysis steps and solicit volunteers

Starting points:
   * IBIS web site fine, but no clear information on information that could contribute to requirement definition. Arpad mentioned two areas: buffer modeling and waveform representation. The latter is related to the Touchstone file format (see below). Arpad agrees to provide more details.
   * Compatibility with Verilog-AMS is important.
   * Difficult to find documentation on Modelica web site. Joachim agrees to find more details.
   * The Touchstone file format is a good starting point for frequency domain support. It seems not recommended to define yet another file format. Rather, a limited list of (de facto) standard file formats (e.g., Verilog-AMS, IBIS, CSV, Touchstone) should be supported. It is better to focus on the semantics of data that can be imported through files. Internal data structures for representing tables need anyway to be optimized for access and interpolation/extrapolation, which is not the case for files.
   * The existing VHDL-AMS implementation from VDA FAT AK30 will be considered when developing the reference packages.
   * Initial requirements from March 2006 have to be revisited and reformulated. Alain started to write a new requirement document ( [[%PUBURL%/%WEB%/ProjectTableDrivenModeling/TDM_Requirements.txt][initial draft version]]) to be discussed and refined.

Alain presents the new requirement document. A number of points are discussed:
   * The scope must be reformulated to say that the development of an optimized implementation is not considered.
   * The description of the multidimensional lookup function with a single dependent variable y is formally covering the needs. The case of multiple dependent variables will be anyway investigated by Ernst.
   * The number of dimensions <em><strong>should </strong></em>not be limited. Another <em><strong>shall </strong></em>requirement that guarantees the support of at least 4 dimensions needs to be added (practical usage). Frequency-domain tables only require to be linear (1D).
   * Dependent and independent variables shall represent real __or complex__ values.
   * The use of _may not_ is not appropriate as it is not formally the opposite of _may_. The proposal is to use _is not required_ instead.
   * Rationales must be added to explain the intent of each requirement.
   * Add support of piecewise constant (ZOH) interpolation method.
   * Control of data lookup as supported in Verilog-AMS should be supported, but the actual mechanism behind must be investigated further. Ernst agrees to look at this.
   * The file format section must be reformulated as the intent is not to define another format, but rather to support standard ones.
---+++ Next meeting

The next telco is planned on July 5, 2010 at 8:00am PDT or 5:00pm CEST.
---+++ AOB

None.
---+++ Adjourn

Joachim adjourns the meeting at 6:25 CEST.
---+++ Action items

Alain:
   * Register Arpad to the Twiki site.
   * Update requirement document from the meeting discussion.

Arpad:
   * Provide links to places in IBIS web site related to table-driven modeling.

Joachim:
   * Provide information related to table-driven modeling from the Modelica environment.

Ernst:
   * Look at the requirements and consequences of supporting multiple dependent variables.
   * Look at how Verilog-AMS control strings are actually working.
Topic revision: r1 - 2016-01-22 - 22:50:16 - TWikiGuest
P10761