IBIS Macromodel Task Group Meeting date: 26 October 2010 Members (asterisk for those attending): Agilent: * Fangyi Rao * Radek Biernacki Ansoft: Chris Herrick * Danil Kirsanov Cadence Design Systems: Terry Jernberg * Ambrish Varma Celsionix: Kellee Crisafulli Cisco Systems: * Mike LaBonte Stephen Scearce Ericsson: Anders Ekholm Intel: Michael Mirmak LSI Logic: Wenyi Jin Mentor Graphics: * John Angulo Vladimir Dmitriev-Zdorov Zhen Mu * Arpad Muranyi Micron Technology: * Randy Wolff Nokia-Siemens Networks: * Eckhard Lenski Sigrity: Brad Brim Kumar Keshavan * Ken Willis SiSoft: * Walter Katz Mike Steinberger * Todd Westerhoff ST Micro: Syed Sadeghi Teraspeed Consulting Group: * Scott McMorrow * Bob Ross Vitesse Semiconductor: Eric Sweetman Xilinx: Mustansir Fanaswalla The meeting was lead by Arpad Muranyi ------------------------------------------------------------------------ Opens: Arpad: Our task list gives us a lot of agenda items - Would like to reserve 20 minutes at end for BIRD -------------------------- Call for patent disclosure: - none ------------- Review of ARs: - Walter divide BIRD 119 into 4 separate BIRDs - Done BIRDs 121-124 have been submitted to the Open Forum - Arpad change Typos BIRD to say the Typ is the implicit default for all types - Done - Arpad and Walter discuss parameter Default/Value examples - Done - Arpad address BIRD 114.2 escape sequences in strings - Done - Arpad: Write parameter passing syntax proposal (BIRD draft) for -AMS models in IBIS that is consistent with the parameter passing syntax of the AMI models - Done, BIRD 117 & 118 submitted - TBD: Propose a parameter passing syntax for the SPICE - [External ...] also? - Done, BIRD 117 & 118 submitted ------------- New Discussion: Arpad showed the changed Typos BIRD: - New note explains Usage Out/InOut - Phrase about Model Specific parameters may not be clear - Walter: The model maker may want to report back some internal data - The EDA tool should report it - Arpad: That is already clear - Arpad deleted the phrase - New text explains tree data structure - Optional keywords like Format now explained with one general example - Other examples follow same notation style Task list item 4a: ambiguity between Format and Text - No comment Task list item 4b: Format = Table is invalid - Walter: We should defer the discussion on Table - The Jitter BIRD may want to use Table - We will understand Table better at that time Task list item 4c: Leaf syntax - "-50" as example of leaf name is confusing because it looks like a number - Walter: This is correct AMI_Version BIRD discussion: - Walter: An existing 5.0 IBIS file can convert to 5.1 without requiring any changes to AMI files that it references - Radek: Agree - Walter: Sections 6 and 10 will be a maintenance nightmare if we keep them in IBIS - Fangyi: Why not integrate this with IBIS version? - Todd: The IBIS and AMI files are a matched set - Ken: It will be hard to remember what works together - Radek: There is no reason to have a separate version field if it is part of the IBIS spec - Arpad: In 5.1 we would only be able to reference new AMI files - Radek: It could say "same or lower version" - Todd: We are making this more complicated than it needs to be - Arpad: So IBIS 5.1 does not need to use older AMI models? - Ken: We could have IBIS_Ver in the AMI file - Walter: Then changing it in the IBIS file means changing all referenced files Walter showed BIRD 121: - Supporting_Files is a simple list of files - DLLPath gives places where DLLs might be found - Fangyi: It should be stated what EDA tools do with it - Ken: Do they have to be in the same directory? - Walter: It can be relative, below the current directory - Mike: Are forward slashes specified? - Walter: Yes - DLLid is needed when the same DLL is used more than once - Mike: DLLPath and DLLid should be DLL_Path and DLL_Id - Walter: These spellings are already in use in Opal - Samples_Per_Bit may be needed - Walter: Some models only work at one rate - Scott: That is out of place in this BIRD - This BIRD is about how to get the simulator working with files - Todd: We can rename the BIRD Walter showed BIRD 122: - This is about Equivalent-Circuit Analog Model parameters - Single ended output has to be converted to differential - Step response to input of TX can be differentiated to get impulse response - Vendors do not want to disclose their circuits, not even the interconnect - The partition between analog and algorithmic portions is important - This does not handle common mode, but that is a general AMI problem Scott showed Circuit and Algorithmic Modeling: - SerDes can be separated into circuit and algorithmic models - BIRD 119 clouds the boundary - We need flexibility to model next generations of serdes - IP vendors want to hide their IP - Most vendors use a Matlab type of environment - Some of the approximations are not correct - There are inappropriate uses of AMI for analog modeling - A long list of things need to be discussed - What is the appropriate model for each type of architecture? - A model maker can easily oversimplify - Would rather have a generalized model than a template model - Maybe a SPICE-like circuit language would be best - Fangyi: The analog model in BIRD 119 belongs in IBIS, not AMI - Walter: The presentation is correct, with good issues - BIRD 119 reflects how IC vendors are currently partitioning - Except that Noise/Jitter is in BIRD 119 - Todd: The boundary was never well defined - We are trying to get to 10Gb/s - Fangyi: Some models give different results when terminations change - Touchstone can't be renormalized to any impedance, for example - Scott: The termination is usually in the s-param - Todd: This is working today with touchstone - Ken: Why would we go back to rigid templates? - Todd: Vendors will not want to give clear text circuits - Scott: Even the touchstone part of BIRD 119 is not well defined - Arpad: We need to see what we can achieve with BIRD 119 - Scott: Peaking coils are in many shipping devices - We need to get that right - Walter: Can they be represented with touchstone? - Scott: As long as the boundary is defined correctly - It's not clear what the s-param interface represents - Arpad: Is this a new BIRD? - Todd: We need more technical discussion - Scott: The template codified in the spec is what people use - Walter: People need to ship models today - Arpad: The BIRDs already submitted should help with that - Walter: The parameters belong in the AMI file - The question is where the template belongs - Fangyi: Can we extend corners to the dependency table? - Arpad: Even with my BIRDs we would have to write about boundaries ------------- Next meeting: 02 November 2010 12:00pm PT Next agenda: - BIRD 119 - Any other new BIRDs ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives