====================================================================== IBIS INTERCONNECT MODELING AD HOC TASK GROUP MEETING MINUTES AND AGENDA http://www.eda.org/ibis/adhoc/interconnect/ Mailing list: ibis-interconn@freelists.org ====================================================================== Next Meeting Wednesday, May 20, 2009 8 AM US Pacific Time Telephone Bridge Passcode 916-356-2663 4 471-3668 (for international and alternate US numbers, contact Michael Mirmak) LiveMeeting: https://webjoin.intel.com/?passcode=4713668 Agenda: - Call for patents - Opens - Sparse matrices and port/node mapping (Walter Katz) ====================================================================== Minutes from May 13: Attendees: ---------- (* 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 the latest parser requirements document draft. Mike suggested the command name "tschk2" is suitable. Bob agreed. The only command argument is the filename with no other options except possibly a help menu. Brad suggested using "check" as a suffix, unless this was used elsewhere. Michael will update the document to use "tschk2" He will also add a separate document requirement for errors/warnings. Mike noted that the lack of a comprehensive list causes issues with IBISCHK today. The parser plans will be discussed in the Friday IBIS teleconference, with an end-of-June deadline for bids. A caveat should be noted to bidders regarding finances: the contract may not be awarded if funds are not available. The bid committee is looking for knowledge and competency in programming. Mike suggested that the team supply test files to the developer. Mentor can do this, as can others. Bob added that the committee will want to see a test suite created as the parser itself is completed. Mike suggested putting a large file requirement in requirements doc, to ensure that large input files can be processed. The team moved to a new topic. Walter reviewed his proposals for sparse matrices and port-to-node mapping. On the latter, industry needs connection/orientation information to prevent misuse of data files. He observed that many in the industry have a need to keep Touchstone "pure" as a data-only file format. He proposed creating a separate file, to specify wrapping of nodes vs. ports. Mike suggested this was like an encrypted SPICE buffer. Walter also outlined the need for sparse matrices to save file space when large connector or package structures were described. Coupling terms in file transfer functions, if below a certain threshold, would be assumed zero. Bob suggested that the sparse matrix approach is equivalent to ICM swathing. Walter responded that swathing still doesn't work, but it's the right analogy. Walter noted he is willing to accept either changing the data format or adding a wrapper file; pointing to a wrapper file that remaps a smaller data set to a larger structure is the intent. He cited the example of mapping an S600p file to an s8p data set. This results in a 6400x reduction in file size. Mike agreed, suggesting this was like compression by repetitive substring reduction Radek stated that what's arguable is the theoretical validity of the approach. Walter objected, noting that an existing file with all zeroed terms removed could be accurately reduced. Radek agreed, but asked whether are zeroes in the file are really zeroes. The files contain inherently-dense matrices, where data may be useful even if very small. Brad observed that RLC files are passive and causal by their nature, even if reduced. The team agreed that, for S-parameters, if the program throws away terms that are too small, the results may not be un-passive, but could be non-causal. The discussion will continue in the next meeting.