======================================================================
IBIS INTERCONNECT MODELING AD HOC TASK GROUP MEETING MINUTES AND AGENDA

http://www.eda.org/ibis/adhoc/interconnect/

Mailing list: ibis-interconn@freelists.org

======================================================================

Minutes from June 10

Attendees (incomplete):
----------
(* denotes present)
Agilent                    - Radek Biernacki, John Moore, Ken Wong
Ansoft                     - Denis Soldo
Cadence Design Systems     - Terry Jernberg, Brad Griffin
Cisco Systems              - Mike LaBonte
Green Streak Programs      - Lynne Green
Hewlett-Packard            - Rob Elliott
Intel                      - Michael Mirmak*
Mentor Graphics Corp.      - John Angulo, Vladimir Dmitriev-Zdorov*
Micron Technology          - Randy Wolff
Sigrity                    - Sam Chitwood, Raymond Y. Chen, Tao Su, Brad Brim*
SiSoft                     - Walter Katz
Teraspeed Consulting Group - Bob Ross*

========================================================================
No patents were declared. 

Michael reviewed several proposals for sparse matrix representation, including some from Vladimir and Walter, and some that he himself had developed.

Bob observed that the sparse matrix format would assume upper-/lower-half symmetry.  The format would also assume port reciprocity (S12 = S21).  He asked the team whether the proposals apply to active devices or not.  

Brad suggested that Mirmak case #1 was only good for linear connectors.  A non-linear (non-symmetrical) "blob" can't be reduced.

In reviewing Walter Katz #1, Vladimir suggested putting all data into a single file, and asked why the data was being packed into a smaller square matrix.  He also inquired why names rather than numbers were used in Walter's proposals.  On Mirmak #1, Vladimir stated that a "strip" matrix is a special case of a sparse matrix, where matrix values with identical componhents become zeroes.  

Michael added that he did not believe extra files were necessary or convenient.

A strip has a main diagonal and subdiagonals.  We therefore cannot assume only zeroes exist in other sections of the matrices.  Do we therefore need to address a general case, or just the special case of a strip matrix?

Brad responded that the beauty of vectors is that undefined data is assumed to be zero.

Bob commented that he doesn't like the use of names for the vectors.  Brad noted that this focuses on EDA tools that rely on connectivity by name.

Michael asked whether the objective was to get rid of zeroes or to establish regularity.  Vladimir repsonded that we are covering both in the case of block-dagonal matrices.  Mirmak #1 is a special case of a special case.

Compression was discussed.  Brad noted that sizes of 100 ports or more are the key cases to address with compression, adding that Walter's connector example was representative.  Vladimir suggested that, in terms of procedures, mixed-mode mapping should be conducted before sparse matrix mapping.

\