====================================================================== IBIS INTERCONNECT TASK GROUP MEETING MINUTES AND AGENDA http://www.eda.org/ibis/interconnect_wip/ (note new URL) Mailing list: ibis-interconn@xxxxxxxxxxxxx ====================================================================== Next Meeting Wednesday, October 27, 2010 9 AM US Pacific Time Telephone Bridge Passcode 916-356-2663 5 761-1665 (for international and alternate US numbers, contact Michael Mirmak) LiveMeeting: https://webjoin.intel.com/?passcode=7611665 Agenda: - Attendees - Call for patents - Review of Minutes * Oct. 20, 2010 - Opens - MCP Comments Review and live editing * Appeal for volunteers - Agenda for Next Meeting - Approve Next Meeting Dates: * Nov. 3, 2010 * Nov. 10, 2010 ====================================================================== Minutes from October 20: Attendees: ---------- (* denotes present) Agilent - Radek Biernacki*, John Moore, Ken Wong Ansoft - Denis Soldo Cadence Design Systems - Brad Griffin, Terry Jernberg, Taranjit Kukal, Dennis Nagle Cisco Systems - Mike LaBonte Green Streak Programs - Lynne Green Hewlett-Packard - Rob Elliott IBM - Greg Edlund ICT-Lanto - Steven Wong Intel - Michael Mirmak* Mentor Graphics Corp. - John Angulo*, Arpad Muranyi*, Vladimir Dmitriev-Zdorov Micron Technology - Justin Butterfield, Randy Wolff* Sigrity - Brad Brim, Raymond Y. Chen, Sam Chitwood, Tao Su SiSoft - Walter Katz* Teraspeed Consulting Group - Bob Ross* ======================================================================== No patents were declared. No opens were raised. Michael Mirmak reviewed ARs given in the last ISS discussion. He reported that the latest Synopsys HSPICE* documentation and parser implementation does not disallow either: - the $ character starting a comment line - the * character preceding an in-line comment Arpad Muranyi asked whether IBIS-ISS will document the same behavior. Michael responded with a suggestion for a caution or warning. Bob Ross Suggested that the document note the behavior, as using cautions/warnings means tabulating what the document considers caution-able or warn-able. Michael reported on what is considered a token in IBIS-ISS, where token really groups element names, nodes, expressions and arguments into a common concept. Walter Katz added that the = character is a token delimiter and that sometimes + and /, etc. can be delimiters. The original source documentation makes this clear. Michael added "parameter" in some contexts is the text "DC" or "Z0" which ISS calls arguments. Walter suggested using the definition of token used for other languages. Radek noted that a string is split into tokens by delimiters, including within expressions. Current definition in source documents is as follows: "- An input token is any item in the input file that [the tool] recognizes. Input token delimiters are: tab, blank (whitespace), comma (,), and parentheses ( ). - Single (') or double quotes (") delimit expressions and filenames. ... Note: The equal sign (=) it is a token delimiter in the sense only that when you define a parameter, both the parameter name and parameter value are considered tokens, so the '=' is a token." The team discussed the rules for single- vs. double-quotes. Arpad requested Specific rules for quotation usage. The source documentation suggests that single-quotes are only permitted format for str(...) parameters. Michael noted that not much work would be left for the document once these issues and some formatting problems were resolved. The team reviewed Arpad's comments on IBIS-ISS. In Arpad's comments, statements are noted as being up to 1024 characters long; Is this *lines* or really *statements*? Michael responded that lines was the correct limitation. Arpad noted a statement requiring values to be floating point doubles. This should be removed according to Radek and Bob as an unnecessary limitation on the engine processing IBIS-ISS input. Walter added that the engine can also alternately process the values as strings (only using numbers in calculations). On parameters and expressions, the team noted that the document needs to mention explicitly that these are only evaluated once, at parsing time, not iteratively or dynamically. The document should not allow non-linear element definitions for R, for example, based on voltages elsewhere. Similarly, examples using the HERTZ parameter definition and V(abs(1)) should be removed. Walter noted that this means only supporting constants and parameters in expressions. Revision 0.7 will be issued with the above changes. ARs Michael Mirmak: The phrase "values may be numeric" in parameters should mention strings. Michael Mirmak: are expressions tokenized within single- and/or double-quotes? Michael Mirmak: Shouldn't the T/U element not allow ZO vs. Z0? Michael Mirmak: Change lines to statements in 1024 character limit Michael Mirmak: no nodes, no elements shall be allowed in parameter expressions ------ Arpad's original notes are enclosed below, for the record. pg. 10 "...statement may be 1024 characters long" Does this mean that any element can't have more than 1024 characters in it, even if it is on multiple lines, or does it mean that each line of an element description that spans multiple lines can be no longer than 1024 characters? IBIS may be haunting me here with the line length limitation, but I thought that 1024 was a line length limit, not statement limit. pg. 12 Shouldn't the T/U element not allow ZO Z0? pg. 16 "In addition, single (') or double quotes (") delimit tokens used as expressions and filenames." I am a little confused about when single and double quotes are used. Is that described anywhere? The table on pg. 15 says the same for both " and '... pg. 16 If instance names can be 1024 long, how is it that a statement can also be (only) 1024 character long? Doesn't this contradict each other? pg. 18 on the bottom doesn't mention that parameters may also have string values... pg. 19 if parameter names are allowed to be 1024 long, how does it work with the rule that a statement can't be more than 1024 long? Same for node name later on the page, and element name on the bottom of the page. pg. 22 "All parameter values in IBIS-ISS are IEEE double floating point numbers" This is not completely true since we can have strings too. I think it should say "All numeric parameters..." pg. 23 "In IBIS-ISS, an algebraic expression, with quoted strings, may replace any parameter." We need to mention when these are evaluated. As far as I remember, we said that they are evaluated once at parsing time, and not dynamically at each iteration. pg. 23 Why do we have that sentence there for the "\\" and "\"? We already explained that somewhere else... pg. 30 "An include file can contain nested .INCLUDE calls to itself or"... Really? Isn't' that going to be an infinite parsing loop? pg. 30 "If the path starts from the current working directory, IBIS-ISS may also find the .INCLUDE file, but prints a warning." I don't understand this. Is this a relative path? Why is a warning necessary for this case? pg. 30 "The file path, plus the file name, may be up to 16 characters long." That sounds an awful short limitation... pg. 32 If .model is used for W and S elements only, why do we show it for a Resistor on pg. 28? pg. 33 Subcircuit definition: "or in a calling subcircuit at any level above" kind of violates the locality rules we established for global nodes are parameters. If a subcircuit is defined in the top level netlist, is it visible inside an ISS subcircuit? pg. 39 Need to check Zo... ------------------------------------------------------------------ The IBIS Ad Hoc Interconnect Task Group Mailing List Archives are available at: //www.freelists.org/archives/ibis-interconn TO UNSUBSCRIBE: Send a message to "ibis-interconn-request@xxxxxxxxxxxxx" with a subject of "unsubscribe" To administer your subscription status from the web, visit: //www.freelists.org/list/ibis-interconn Meeting minutes and files are available; visit: http://www.eda.org/ibis/interconnect_wip/