IBIS Macromodel Task Group Meeting date: 24 Jan 2011 Members (asterisk for those attending): Agilent: * Fangyi Rao * Radek Biernacki Altera: * David Banas Ansys: Samuel Mertens Dan Dvorscak * Curtis Clark Arrow Electronics: Ian Dodd Cadence Design Systems: Terry Jernberg * Ambrish Varma * Feras Al-Hawari Celsionix: Kellee Crisafulli Cisco Systems: Ashwin Vasudevan Syed Huq Ericsson: Anders Ekholm IBM: Greg Edlund Intel: Michael Mirmak LSI Logic: Wenyi Jin Maxim Integrated Products: Mahbubul Bari Mentor Graphics: * John Angulo Zhen Mu * Arpad Muranyi Vladimir Dmitriev-Zdorov Micron Technology: Randy Wolff NetLogic Microsystems: Ryan Couts Nokia-Siemens Networks: Eckhard Lenski QLogic Corp. * James Zhou Sigrity: Brad Brim * Kumar Keshavan Ken Willis SiSoft: * Walter Katz Todd Westerhoff Doug Burns * Mike LaBonte Snowbush IP: Marcus Van Ierssel ST Micro: Syed Sadeghi Teraspeed Consulting Group: Scott McMorrow * Bob Ross TI: Casey Morrison Alfred Chong Vitesse Semiconductor: Eric Sweetman Xilinx: Mustansir Fanaswalla The meeting was lead by Arpad Muranyi ------------------------------------------------------------------------ Opens: - Arpad: Minutes from last week will be sent to Mike for posting - Arpad: There will be no meeting next week due to DesignCON - Arpad: Can we remove the last agenda item? - Walter: Yes it can be removed -------------------------- Call for patent disclosure: - None ------------- Review of ARs: - None (may find some after review of minutes from last week) ------------- New Discussion: IBIS 5.1 parser plans: - Bob: The editorial committee has been discussing this - Use_Init_Output is deprecated - Not going to document executable model flow changes in 5.1 - 5.1 rules will apply to 5.0 models - 5.0 models can have the relaxed syntax allowed in 5.1 - We can discuss this at the DesignCON summit - Arpad: If Version is absent 5.0 rules will apply? - Maybe it should default to 5.1 - Bob: A BIRD would be needed for that - BIRD 126 already covers this well - It mostly deprecates 5.0, with exceptions - Arpad: I like the idea but if 5.0 rules are not used we should not default to 5.0 - Radek: It is not 100% clear what to do if Use_Init_Output is present - Walter: If AMI_Version is 5.1 Use_Init_Output will be flagged as an error - In most cases we can add AMI_Version and delete Use_Init_Output Walter showed an email about Analog Buffer Modeling: - Walter: No one has objected to these statements Walter showed another email with conclusions: - Walter: Point 2 is not 100% true - With just [Ramp] the V-T can't be represented - External Model could have a D-A with a PWL to give the K(t) function - This will represent the behavior of any LTI driver - Walter: Point 4 is about it being easy to accurately describe analog behavior - Every swing setting has a different Voh and impedance - An IBIS file can be generated with many [Model Selector]s, many AMI files - Simulators often do quadratic rounding of ramp corners - Would like to have four internal reference models with reserved subckts - Feras: We don't need the new keywords - IBIS-ISS subckts can model it - Walter: The reference subckt names are not new keywords - Arpad: Is this about standard ISS files or hardcoded circuits? - Feras: You proposed 4 keywords in an email - Walter showed a previous email - Walter: We can ignore the Reserved_Model keyword - I would only like these subcircuit names and files to be reserved - They would not have to be supplied with each model - Feras: Would it be inserted in the IBIS file, not external? - We should be able to include an analog model with or without AMI - Walter: The BIRD allows that - Arpad: BIRD 118 can be extended for this purpose - Ambrish: It should allow linking just any file to the AMI section of IBIS - Walter: Adge Hawes will have related presentation at DesignCON - It will show what vendors need - Arpad: If we allow just any file we have to spell out the allowed syntax - Those changes can be made though - Feras: You have to have an AMI model present - The [Algorithmic Model] section should not be required - Walter: It's OK if this is used only where AMI is present - Ambrish: How would we handle it when there is no AMI? - Arpad: We could have a file pointer without relying on the AMI keyword - Feras: Users must be able to use models with or without DLLs - Walter: There will be a presentation about a model with 128 Touchstone files in it - Arpad: We just need a keyword other than AMIfile - The file format is tree format just like AMI though - Kumar: AMI does not own the tree format - Walters: We should listen to what IC vendors want - Ambrish: How would corners be chosen? - Arpad: The GUI will have to let users choose - Walter: It could be done as an Include statement - Radek: So this is not a reserved parameter, but a reserved name - Walter: One problem is having two different models with the same name - There will be only one of each of these reserved models, no collisions - Ambrish: That is for the user and vendors to deal with, not the spec - If it's hardcoded in the tool we can't change it - Walter: Then a file can be created for anything new - Fangyi: Why are there 4 reserved file names? - Feras: We can publish templates and those can be copied - How will the user find the actual subcircuit? - Walter: It would be in the spec - It would also be supplied as files - Tools might handle it efficiently through internal means - Fangyi: Why can't vendors use a different file name? - Walter: Then we will have file management problem - Feras: There are many other circuit types needed - Walter: Nothing prevents them from doing that - Ambrish: Templates that only work for AMI defeat the purpose - Walter: IBIS-ISS was created for AMI analog models and interconnect - The only application right now is AMI, until we have BSS - Arpad: The 6 points are all correct - Why do we need a shorthand for these circuits? - Walter: At this point I only want them as reference circuits - Radek: A circuit file with no reserved name causes no problem - Walter: I have a model that has 260,000 combinations - Arpad: What benefits do the 4 circuit bring us? - Walter: They cover all known SerDes standards and contemplated for 4 years - Arpad: We can just supply the files - Kumar: These circuit models are an approximation - The parameters are not cast in stone - Fangyi: Why do we still need Touchstone if these are accurate? - Walter: That is needed for broadband - Complicated subckts would be needed, but they are proprietary - Fangyi: So these do not cover everything - Feras: What if someone wants non-linear models? - Walter: ISS only allows linear - Feras: [External Model] allows other non-linear languages - Walter: I have not seen a channel where a non-LTI model makes a difference - Industry tools all use LTI models like MatLab, HSS-CDR - Kumar: We can't restrict ourselves to what those tools do - Fangyi: This analog model is not just for AMI - Rambus did experiments showing that - Kumar: We can't assume superposition will apply into the future - Walter: LTI is more true as speeds go higher - Fangyi: Can parameter names be changed for the reserved models? - Walter: No - Ambrish: That makes them reserved parameters - Walter: Not by the AMI definition - They could be called pseudo-reserved - We need to remember that there are 3 groups: users, EDA vendors, IC vendors - Ambrish: The IC vendors tend to not be well versed in these issues Meeting ended. ------------- Next meeting: 7 Feb 2011 12:00pm PT Next agenda: 1) Task list item discussions ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives