IBIS Macromodel Task Group Meeting date: 25 March 2014 Members (asterisk for those attending): Agilent: * Fangyi Rao * Radek Biernacki Altera: * David Banas ANSYS: Dan Dvorscak * Curtis Clark Cadence Design Systems: * Ambrish Varma Brad Brim * Kumar Keshavan Ken Willis Scott Huss Ericsson: Anders Ekholm Intel: * Michael Mirmak LSI * Amaresh Malipatil * Dai Xingdong Maxim Integrated Products: Hassan Rafat Mentor Graphics: * John Angulo * Arpad Muranyi Andrey Matvienko Micron Technology: * Randy Wolff Justin Butterfield QLogic Corp. James Zhou Andy Joy SiSoft: * Walter Katz * Todd Westerhoff * Mike LaBonte Teraspeed Consulting Group: Scott McMorrow * Bob Ross The meeting was led by Arpad Muranyi ------------------------------------------------------------------------ Opens: - Arpad: Interconnect task group meetings resume tomorrow. - Mike L: Michael Mirmak owns the email list for that. -------------------------- Call for patent disclosure: - None ------------- Review of ARs: - Arpad produce examples for BIRD 163, 164, 165 - In progress. - Can be dropped from this agenda. ------------- New Discussion: BIRD 147 draft 7: - Walter: Summary of issues: - Not general enough to satisfy user community. - It does not satisfy system designers evaluating channels. - It does not support statistical flow. - It doesn't deal with presets. - It doesn't tell how how to program the device. - Can't see how to change it incrementally. - I am asking Ambrish to remove my name as author. - Arpad: What changes do you propose? - Walter: I don't have the resources to work on it now. - It might model what IC designers want, but not system designers. - It can't find optimum solutions. - This has been mentioned several times. - Ambrish: By definition a back-channel is time domain. - We might look at statistical later. - Todd: It isn't worth it, a new approach is needed. - Ambrish: Two IC vendors have asked for this. - They are happy with the implementation. - Arpad: Walter said system designers couldn't use it. - Ambrish: They need a model of the actual device. - Kumar: That is an artificial distinction between system and IC designers. - The model just has to do what the device is doing. - Walter: Scott Huss said he had a model that finds an optimal solution. - System designers just want to know what that solution is. - It's asking a different question, doesn't report the solution. - Ambrish: The models can report whatever they want. - Kumar: They only have to model what the actual device does. - Michael M: Is there any plan to propose a solution that does address statistical? - Kumar: We have no plans. - Todd: There seems to be no consensus in this group that statistical analysis is important. - Michael M: It's important. - Arpad: Are we saying there is or is not a need to support statistical? - Todd: We absolutely need it, to support simulation of many channels. - Michael M: I use statistical analysis. - We also need time domain too. - Kumar: These should not be mixed up in the same BIRD. - Radek: The problem is that this BIRD can not be passed if it will not allow extension to statistical. - Kumar: I can't support that. - Radek: We should expect the results of optimization to be applied to statistical analysis. - This BIRD does not even discuss statistical. - Ambrish: Nothing stops anyone from extending it to statistical. - Michael M: It must not be required to use a bit stream for optimization. - Arpad: We don't want to corner ourselves with this. - Michael M: We probably could add to this BIRD to open the door for statistical. - We must carefully define what the silicon actually does. - The model might not do what the silicon does. - There needs to be a parameter stating whether the model matches silicon. - Kumar: Customers will not know what to trust. - There should be a separate model for statistical. - Mike: This approach is an abstraction. - This is parameter passing, not voltages at the physical level. - Fangyi: True, that is a concern. - Kumar: There are many problems in adaptation, abstraction is needed. - Todd: All modeling is an abstraction, otherwise we would be using HSPICE. - My customers want performance, some abstraction is needed. - We need a strategy to support time domain and statistical together. - Arpad: It would be best done together even if in separate BIRDs. - Otherwise we might corner ourselves. - Michael M: Who can do this? - Todd: We can't put time into this until as a group we decide to address the issue. - We might have a vote on this. - Arpad: We can have a vote next week, giving us time to think about it. - Michael M: This is a "sense of the committee" vote. BUG 150 - ambiguous directionality: - Michael M: Nothing is new, other than reflector discussion. - We can discuss this next week. BIRD 128 - AMI_parameters_out passing AMI_parameters_in data - Arpad: It seems BIRD 147 affects BIRD 128, there is a dependency. - Walter: There is a dependency. - We had an objection from Radek. - Radek: Arpad raised concern too. - It is fixable but I don't have time right now. - Arpad: Yes I am concerned. - We should wait to see what happens with BIRD 147. Walter: I would like to discuss a new BIRD on redriver flow next week. ------------- Next meeting: 01 April 2014 12:00pm PT ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives