IBIS Macromodel Task Group Meeting date: 19 Feb 2013 Members (asterisk for those attending): Agilent: Fangyi Rao * Radek Biernacki Altera: * David Banas Julia Liu Hazlina Ramly Andrew Joy Consulting: Andy Joy ANSYS: Samuel Mertens * Dan Dvorscak * Curtis Clark Steve Pytel Luis Armenta Arrow Electronics: Ian Dodd Cadence Design Systems: Terry Jernberg * Ambrish Varma Feras Al-Hawari Brad Brim Kumar Keshavan Ken Willis Cavium Networks: Johann Nittmann Celsionix: Kellee Crisafulli Cisco Systems: Ashwin Vasudevan Syed Huq Ericsson: Anders Ekholm IBM: Greg Edlund Intel: * Michael Mirmak Maxim Integrated Products: Mahbubul Bari Hassan Rafat Ron Olisar Mentor Graphics: John Angulo Zhen Mu * Arpad Muranyi Vladimir Dmitriev-Zdorov Micron Technology: Randy Wolff Justin Butterfield NetLogic Microsystems: Ryan Couts Nokia-Siemens Networks: Eckhard Lenski QLogic Corp. James Zhou SiSoft: * Walter Katz * Todd Westerhoff Doug Burns * Mike LaBonte Snowbush IP: Marcus Van Ierssel ST Micro: Syed Sadeghi Teraspeed Consulting Group: Scott McMorrow * Bob Ross TI: Casey Morrison Alfred Chong Vitesse Semiconductor: Eric Sweetman Xilinx: Mustansir Fanaswalla Ray Anderson The meeting was led by Arpad Muranyi ------------------------------------------------------------------------ Opens: - Arpad: I may not make it next week - Someone else could run the meeting -------------------------- Call for patent disclosure: - None ------------- Review of ARs: - Walter write new BIRD 122 replacement draft - Done - Ambrish email reflector regarding BIRD 147 status - Ambrish: There is confusion over this, but working on it ------------- New Discussion: Interconnect Task Group report: - There was no meeting Arpad: Walter will introduce a new BIRD proposal and immediately table it - Arpad showed new BIRD draft "Support PAM4 in AMI Modeling" - Walter: - Mainly PAM4 has 4 signal levels - NRZ is also known as PAM2, but that is not commonly used - We should only have to tell the tool it is PAM4 - Arpad: PAM can have various signal level counts, could we just have a number? - Walter: That is possible - Arpad: Should this be in IBIS, not AMI? - Walter: [Model] is very much dedicated to binary buffers - Arpad: Should analog models be in IBIS or in AMI? - Walter: PAM is tied to SerDes, therefore it should be AMI - Bob: Should we wait for 802.3bj to be approved? - Walter: That is well under way, but we can table until then - Michael M: When will that be finalized? - Walter: Probably a year - Arpad: It would be best to spend our time on other BIRDs right now - Arpad moved to table - Walter seconded - No one objected Arpad showed BIRD draft "AMI Touchstonefile Analog Buffer Models" - Walter: This uses the Tstonefile AMI parameter to call for an s4p on-die model - Circuit diagrams show the other elements required for TX and RX - Last week there was a question about whether we should specify how to generate the impulse response - Radek: That was for legacy IBIS - Walter: We could specify reference methods, not normative - An alternative would be to use the arithmetic recommended by James - I do not think either normative or reference methods could be described - Todd: This is not related to BIRDs 116-188 - Arpad: There were some emails about this - Immediately after the summit we had a discussion and agreed to have this proposal - The other proposal is BIRDs 116-118 - This proposal uses a template and solve the issue only for AMI - Bob: The proposal did not mention [External Model] - This seems like it works only with [Model] does not interfere with [External Model] - I don't see conflict - Arpad: At DesignCON I suggested we could have both proposals - They do not conflict - Walter said vendors are already doing this and models exist - Ambrish: Is that good enough? - Can Walter's proposal be done with BIRDs 116-118? - Walter: Yes, but a subckt and other unnecessary work is required - Todd: We wrote the AMI file as though those BIRDs were in effect - Ambrish: Does this replace BIRDs 116-118? - Todd: No, we wrote it as though that ISS subckt existed, and just used the parameters - There may be a disconnect because this is differential and [Model] is single ended - Arpad: There is mention of subckts - Michael M: Examples would help - Todd explained this using the circuit diagrams - Walter: Clarification: the subckt does not include the voltage supplies - Todd: Once the circuit in IBIS declares parameters they can be used from AMI - We simply assumed the circuit existed - Walter: We could wait for BIRD 116 to be approved, then add in the other stuff - That would be IBIS 6.0 compliant - Meanwhile we can do it this way and there is no illegal syntax - Radek: We do not necessarily want it, but the parameters could become reserved - That would take precedence over the legacy IBIS model - Ambrish: That would be mandating BIRD 122 - Todd: I agree with Radek, these parameters technically should be reserved - Arpad: After BIRDs 116-118 are approved we might be required to have the actual circuits - Bob: Bypassing BIRDs 116-188 makes sense - Radek: It is also necessary that we cover how source impedances are treated - Ambrish: That can be done with the circuits - Radek: This shortcut is useful - Todd: We all agreed to have this proposal - This is not to undermine BIRD 116 - Ambrish: We have gone from model specific to reserved, to just for discussion, to a real BIRD - Walter: We can have both solutions - This does not preclude BIRD 116 - The only question is whether to support the shortcut - David: Walter said non-straight-line IV curves could not be used - Walter: They can be used but there will be accuracy problems - David: Do BIRDs 116-188 allow those non-straight-line features? - Walter: Since IBIS-ISS is LTI, we will not have that problem - Todd: You can't model non-linear with it - Walter: With the AMS languages and SPICE you can - Arpad: For this we would want reserved, for BIRDs 116-118 model specific - Other than that they are identical - I don't mind this interim solution - There are a few details we would have to go over - An analog simulator can't have zero time edges, etc. - Todd: The time is "practically zero" - David: Something is wrong if touchstone depends on how it is driven - Todd: The output changes - Driving with a square wave then a ramp will have different outputs - David: Is "close to zero" ramp the right approach? - Todd: It is the best we have right now - Walter: Impulse responses are all die to die s4ps, that is how they are doing it? - Bob: We can: - Accept this - Accept BIRDs 116-118 - Accept both - Arpad: The other BIRDs could also be modified to include this - Bob: The diff circuit might not be centered around zero - Only BIRDs 116-118 are general enough for that - Walter: IBIS-ISS would not handle that either - IBIS-BSS could do non-LTI - BIRD 116 should wait for that - Arpad: That would be an interim solution - Ambrish: In what order should we discuss these? - Radek: We should discuss both approaches - Arpad: We should try adding the shortcut idea to BIRD 116 - Radek: There was a BIRD to add Touchstone as a language - But that would require specifying the termination conditions - There is nothing to be done here - Bob: The press release referred to BIRDs 116-118 as though they are supported - Are those supported yet? - Radek: No, but there would be no difficulty doing that AR: Arpad try to add shortcut language to his BIRDs Bob: About the PAM4 BIRD, Range would allow it to support any number of levels ------------- Next meeting: 19 Feb 2013 12:00pm PT Next agenda: 1) Task list item discussions ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives