IBIS Macromodel Task Group Meeting date: 25 February 2014 Members (asterisk for those attending): Agilent: Fangyi Rao * Radek Biernacki Altera: David Banas ANSYS: * Dan Dvorscak * Curtis Clark Luis Armenta Cadence Design Systems: * Ambrish Varma * Brad Brim * Kumar Keshavan Ken Willis Ericsson: Anders Ekholm Intel: * Michael Mirmak 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 Teraspeed Consulting Group: Scott McMorrow * Bob Ross The meeting was led by Arpad Muranyi ------------------------------------------------------------------------ Opens: - Arpad: I will travel next week, unable to meet. - Mike: I most likely can't attend either. - Arpad will try to find someone else to chair. -------------------------- Call for patent disclosure: - No one declared a patent. - Arpad: Do we still have to ask this under SAE rules? - Michael M: Yes we do. ------------- Review of ARs: - Arpad duplicate Walter's example using BIRDs 163-165. - Done, but it is a work in progress. - Walter advised reviewing it in ad-hoc group meetings, not here. - Walter update new package model example with corrections and parameterized length. - Done. Parser bugs 150 and 151: - Walter: bug 151 is about negative matrix values. - Bob: The negative matrix value one would be an enhancement. - Walter motioned to move these items into the tabled section. - Michael M seconded. - No one objected. Arpad showed back-channel BIRD 147 draft 7 - Ambrish: Arpad and I exchanged emails last week. - We made some changes, which are highlighted. - A bit sequence example was moved to after the explanation of "random". - Rules for LFSR seed bits were clarified. - The source of the BCI string was clarified. - The EDA tool facilitates communication between TX and RX. - Various comments have been added. - Clarification was added for Training_Done. - Taps in examples were changed to Type Tap so names can have "-". - Walter: - On page 2 I object to adding a new Bits type. - It only changes what the delimiter character is. - Random bit patterns will not work, the pattern needs to be well defined. - On page 7 it says the EDA tool can initiate Training_Done. - The tool can only modify the string. - On page 8 Type Tap is actually a float. - Ambrish: This is in Protocol_Specific, that would not apply. - Arpad: Would each number require ".0"? - Walter: It should use List instead of Range if they are really integers. - On page 11 the use model is not clear. The user might ask: - What is the optimum solution? - Will this RX find the solution? - Ambrish: Are there ways to optimize without backchannel? - Walter: Yes there are. - The standard for KR says the RX "shall find a solution". - That becomes the designers responsibility. - The user only wants to know what the BER will be. - It can be assumed a solution can be found. - Who will use the backchannel in simulations? - Only designers may want it, not users. - Kumar: That assumes there is only a single solution - In reality multiple solutions are possible - Performance will depend on which is found. - Walter: True, but the standard says it will find a solution better than some value. - Kumar: It is important to explore the outcomes. - Bob: - On page 1 the BIRD number should be added, and modified date. - On page 2 is Training a DLL input? - Ambrish: Yes. - Bob: - On page 4 maybe "None" could be "Off". - On page 7 what does Default False mean for Training_Done? - Ambrish: Training is by default not done until it is True. - Mike L: Format List is to be used by users to choose values. - Kumar: Because Type is Boolean the Value can only be True or False. - Ambrish changed it to Format Value. - Mike: Format Value can not have a Default. - Bob: I think Default now has to have "[illegal]". AR: Bob work with Ambrish on BIRD 147 editorial work. Package Modeling: - Walter showed an outline of package model status items. - Walter: Are there other scenarios we need examples for? - John had mentioned mixing legacy and new model styles. - We need to decide which tradeoffs to adopt regarding what the industry needs. - Brad: Two issues here: - How to take a submodel and make a more complete model. - What will we do for people who have a full model today? - Walter: Today's proposal covers that. - Brad: The time to define swathing might hold up the simple format. - Walter showed a syntax example. - Walter: Vendors say the coupled models are for pre-layout only. - For post-layout they supply an s4p, no ability to generate a full matrix. - There is no intention to use smaller models to create larger. - An s12p would be for pre-layout. - They might say to use 4 ports of the s12p for post-layout. - Or they might supply the s4p. - There can be cascading of models. - Brad: Many people analyze multiple channels. - They need full package models. - Can they build that from 12 port models? - Walter: No. - Arpad: You are saying post-layout with swathing would not have coupling? - Walter: Coupling is only available with explicit pins, not with swathing. - Arpad: Length can be parameterized? - Walter: Yes, I have seen a 10 port example of that. - It is a half-swathing pre-layout example. - Walter: We need to make sure we have cover all usage scenarios. AR: Walter send package issues outline and example to email list. ------------- New Discussion: ------------- Next meeting: 04 March 2014 12:00pm PT ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives