IBIS Macromodel Task Group Meeting date: 19 May 2015 Members (asterisk for those attending): ANSYS: * Dan Dvorscak * Curtis Clark Avago (LSI) Xingdong Dai * Bob Miller Cadence Design Systems: Ambrish Varma Brad Brim Kumar Keshavan Ken Willis eASIC * David Banas Marc Kowalski Ericsson: Anders Ekholm IBM Steve Parker Intel: * Michael Mirmak Keysight Technologies: * Fangyi Rao * Radek Biernacki * Nicholas Tzou Maxim Integrated Products: Hassan Rafat Mentor Graphics: * John Angulo * Arpad Muranyi Micron Technology: Randy Wolff Justin Butterfield QLogic Corp. James Zhou Andy Joy SiSoft: * Walter Katz Todd Westerhoff * Mike LaBonte Synopsys Rita Horner Teraspeed Consulting Group: Scott McMorrow Teraspeed Labs: * Bob Ross TI: Alfred Chong (Note: Agilent has changed to Keysight) The meeting was led by Arpad Muranyi. ------------------------------------------------------------------------ Opens: - Arpad: Should we discuss [Initial Delay]? - Michael M: Not yet, there will be a BIRD. - Radek: We tabled [Define Package Model] but could discuss the changes made. - What are the differences between the one we approved and what Open Forum will vote on? -------------------------- Call for patent disclosure: - None ------------- Review of ARs: - Arpad to review IBIS specification for min max issues. - In progress. ------------- New Discussion: Directionality BIRD draft - Michael M: Have all issues been resolved? - Bob M: I had asked some questions but they may be academic. - The problem we are trying to solve is not clear. - Arpad: AMI models do not say if they are TX or RX. - There is no way to check if they are assigned incorrectly. - Michael M: It's not much of an issue if they work only in one direction. - AMI parameters can be used to check them in that case. - But AMI can be used for bidirectional DDR4 signals, for example. - Bob M: A bidirectional SerDes that does turnaround makes no sense. - Michael M: Only the current state would be given, no dynamic turnaround. - Bob M: In EDA tools I assign models explicitly to TX or RX. - A single executable that can be TX or RX should be possible. - Michael M: Users don't want to copy a topology just to model the reverse direction. - The BIRD supports single executables, but separate AMI files are required. - Walter: IBIS describes parts and must describe the behavior of each pin. - If a pin is I/O then IBIS must be able to describe both behaviors. - Arpad: We were going to check direction by Reserved_Parameters. - There are few required parameters, we may not have enough for checking. - Walter: The model writer can add an In Model_Specific parameter that the model checks. - John: A Reserved_parameter would not require that kind of user intervention. - Arpad: We could require a Reserved_Parameter for model direction. - Walter: We would define TX AMI files and RX AMI files. - Radek: A Model_Specific parameter would clarify this. - Walter: It would be of type Value. - Bob R: For I/O models it's unclear. - Walter: There are separate AMI files. - John: Each model would look for a parameter of a certain name. - Bob R: The EDA tool doesn't need to know it, only the model. - Arpad: Is this BIRD ready for submission? - Bob R: We left the Reserved_Parameters in it. - Walter: Michael M should issue a clean final draft and we can vote on it next week. - Bob R: How would we identify the TX and RX Executable lines? - Radek: There is a parameter for that in the IBIS file. - Michael M showed the BIRD. - Michael M: It is a subparameter of the [Algorithmic Model] keyword. - There are no Reserved_Parameters defined in the BIRD at all. - There is however a table listing known TX and RX Reserved_Parameters. - David: How do we use one DLL for I and O? - Michael M: It is called from one [Algorithmic Model] with Direction TX and one with Direction RX. - Bob R: Can a single DLL be used for multiple models? - Radek: We removed that restriction. - Michael M showed the removed phrase. - Michael M: This was done so existing models don't have to change much. - Direction is not required except for I/O Model_types. - David: Can there be multiple Directions in one section? - Walter: There could be Executable_TX and Executable_RX instead of the Direction subparameter. - David: A field right on the line would be consistent with information. - Arpad: We still need to resolve if multiple TXs and RXs are allowed for one [Algorithmic Model]. - Walter: We already do, for platforms. - Bob R: Only one Direction subparameter should be in one [Algorithmic Model]. - John: There could be a risk with having multiple [Algorithmic Model] sections. - Michael M: Each [Algorithmic Model] should point to one AMI file. - Doing it this way we can check for that. - Cadence liked this syntax better. - Arpad: Can we have two [Algorithmic Model] section with the same Direction? - David: Could there be subparameter hierarchy? - Michael M: I don't think we have any examples of that so far. - Arpad: We have hierarchical keywords. - Mike L: This will require one AMI file per [Algorithmic Model] block? - We have not had that rule so far. - Michael M: Yes. - Mike L: I know of no models that use different AMI files so far. - Software would be easier to write with this assumption. - Bob R: The example in this BIRD has different AMI file names for TX and RX. - It's not clear. - David: It's clear, the question is what we want. - Mike L: Some models have platform dependent Supporting_Files. - They might want to use separate AMI files for each platform. - David: So there have to be unique TX and RX AMI files, but there could be more than two files. - Radek: Do we currently disallow multiple [Algorithmic Model] keywords? - Michael M: Yes, that changes in this BIRD. AR: Michael M email list to start discussion of new Directionality issues. Problems with Global Ground in IBIS: - Arpad: We only have time for a short introduction. - Walter showed a presentation "Terminal Names in IBIS". - Walter: There is confusion over what a terminal name is. - It is about voltage references. - C_comp is involved. - Some of our figures have ground symbols. - We made an editorial error while going from IBIS 5.0 to 5.1. - We should replace some terminal names in diagrams with standard names like A_pcref. - The references for IV curves must be discussed, some use a mythical ground. - It would be best to use C_comp_power_clamp and C_comp_ground_clamp instead of C_comp. - Arpad: Will this be discussed in the interconn meeting tomorrow? - Walter: It should be. AR: Walter send "Terminal Names in IBIS" presentation to Mike L for posting. ------------- Next meeting: 26 May 2015 12:00pm PT ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives