IBIS Macromodel Task Group Meeting date: 16 September 2014 Members (asterisk for those attending): Altera: * David Banas ANSYS: * Dan Dvorscak * Curtis Clark Avago (LSI) Xingdong Dai Cadence Design Systems: * Ambrish Varma Brad Brim * Kumar Keshavan Ken Willis Ericsson: Anders Ekholm Intel: Michael Mirmak Keysight Technologies: Fangyi Rao * Radek Biernacki Maxim Integrated Products: Hassan Rafat Mentor Graphics: * John Angulo * Arpad Muranyi Micron Technology: * Randy Wolff Justin Butterfield QLogic Corp. James Zhou Andy Joy eASIC Marc Kowalski SiSoft: Walter Katz * Todd Westerhoff * Mike LaBonte Synopsys Rita Horner Teraspeed Consulting Group: Scott McMorrow * Bob Ross (Note: Agilent has changed to Keysight) The meeting was led by Arpad Muranyi. ------------------------------------------------------------------------ Opens: - None -------------------------- Call for patent disclosure: - None ------------- Review of ARs: - Bob: Send BIRD 147 comments to Ambrish - Done - Ambrish - Investigate other ways to communicate tap coefficients in BIRD 147. - In progress. - Todd asked for the new version to be posted. - Arpad to review IBIS spec for min max issues. - In progress. ------------- New Discussion: BIRD 128: - Radek: This is needed for BIRD 147 but is otherwise independent. - The usage details are not spelled out enough to avoid inconsistencies. - Walter had agreed to that. - Walter, Ambrish and Radek should meet to work it out. - No specific required changes are known yet. - Arpad: We can table this until there is an update. - Todd: RX ami_parameters_out would be used to feed data to TX ami_parameters_out in Init. - BIRD 128 does not support any parameter passing in Init. - Ambrish: That is already handled in IBIS. - Todd: It does not support using into ami_parameters_out as an input, but this BIRD calls for that. - Ambrish: That needs to be addressed. - Radek and Arpad agreed. - Bob: It could be voted if updated by this Friday, but we might not want to push that hard. BIRD 147: - Ambrish showed BIRD 147 draft 13. - Bob: Numerical training modes were moved into the BIRD text, they are important. - There were no definitions of coefficients, that was addressed. - The in/out changes we discussed are needed too. - Ambrish: I will change that and send it to Mike. - Ambrish: Will this also be submitted to the open forum? - Radek: Points from the last meeting have not been addressed yet. - Walter expressed strong opinions by email after the meeting. - It might bounce back to us anyway if submitted. - Arpad: There will be no other BIRD changes, it is ready. - Ambrish: The new discussions are more philosophical. - Arpad: We need to decide what recommendation to make to the open forum. - Todd: We have talked with semiconductor vendors about their requirements. - This is not done yet. - Backchannel training modeling is very important to them. - Not having to do that is equally important for performance reasons. - It can be done quickly in Init without calling GetWave. - They have algorithms to do it. - They expect customers to skip GetWave most of the time. - They are concerned that the BCI file will constrain TX to RX data passing. - The allowed BCI keywords might constrain them. - It's hard to tell what is allowed. - The Init and GetWave training descriptions are loose. - Ambrish: When the TX thinks the RX can't take any more settings it sends a range. - Todd: The messaging has two types of values. - The vendor needs to address things like granularity. - Ambrish: That is in the Protocol_Specific section. - Todd: The RX has to generate tap coefficients with some granularity. - Ambrish: The TX sends its allowable tap values. - Todd: We had not heard it could be done this way. - This is based on the AMI syntax because we know it. - The AMI file sets possible values for parameters, just a leaf and values. - In Init mode tap has min/max but in GetWave it has mode/adjustment. - This is the only place where behavior would change. - Bob: This is a bit uncomfortable. - Ambrish: It has to be understood that if there are two values they are min and max. - Todd: Page 6 says the BCI file sets the range but the model can go beyond. - Ambrish: It could modify less taps or more taps than specified. - Todd: Do we spell out that an RX must ignore anything it doesn't understand? - Ambrish: That should be understood. The RX may ignore or accept. - Todd: It says a model can support more than the BCI has yet still point to the BCI. - The TX will send taps info the RX might know nothing about. - Ambrish: There is no confusion because with type tap it is known that 0 is the main tap. - Todd: Models today are expected to ignore parameters they don't understand. AR: Ambrish send updated BIRD 147 draft to Mike for posting. - Arpad: We still need to decide our recommendation to the open forum. - Radek: We still need to settle some issues. - Ambrish needs to clarify some things. - The TX and RX know some things about each other. - Ambrish: They might not know if the other supports GetWave for example. - Radek: This puts information where it doesn't need to be. - Arpad: Walter's flow related questions are not addressed in this BIRD. - It has a specific order of events. - We need to decide if this BIRD is sufficient. - It might be rejected. - Ambrish: That will add much complication for something that may not be required. - We should wait for someone to say it is needed instead of speculating. - Arpad: We should not put limits in front of us. - Radek: Do we pass this and start a new BIRD? - Arpad: Either that or we say this BIRD is not sufficient. - Todd: We should take a meeting or two to decide what the requirements are. - Only then should we look at if this BIRD meets it. - Ambrish: That would be going back to the start. - We have already had discussions about requirements. - Arpad: We could pull those slides and look at it next week. - Todd: That would be good. - Ambrish: Does the BIRD meet your expectations or not? - Todd: Possibly but we need to look at it. - Much editorial improvement is possible. - Arpad: We should not do that. - Todd: If it passes there is room for ambiguity. - Ambrish: Walter felt this BIRD needed no changes. - Todd and Walter should coordinate on this. AR: Todd produce slides for co-optimization requirements discussion next week. - Ambrish: I would like to see Todd's BIRD 147 notes. AR: Todd send BIRD 147 notes to Ambrish. ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives