IBIS Macromodel Task Group Meeting date: 21 May 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: The last day to submit BIRDs for 6.0 is this Friday -------------------------- Call for patent disclosure: - None ------------- Review of ARs: - Fangyi update BIRD 156 per task group feedback - Done, BIRD 156.2 - Walter to write new BIRD for [Repeater/Retimer Pin] keywords. - Done, combined into BIRD 156.3 - This has been emailed. - SiSoft should be added to the author list. ------------- New Discussion: Interconnect Task group report - Michael M: At the meeting last Wednesday we discussed: - BIRD 161.1 - A new format and structure for EMD - A ports presentation from Walter Redriver BIRD 156.3: - Fangyi and Walter expressed a desire to work out final details soon. - Todd: We have a lot of models out there, it would be good to finalize this - Fangyi showed the BIRD and described the changes - It was clarified that "repeater" is the generic term for either redriver or retimer - Arpad: Some important information in the analysis section needs to be in the main section so it will be in the spec - Fangyi: A retimer requires RX AMI_GetWave() outputting clock_times - What if the RX has no GetWave? - Walter: Statistical analysis is still possible without GetWave using jitter parameters - We could require GetWave - Arpad: Can a retimer not have clock ticks? - Walter: The simulator can derive clocking - Todd: Without a CDR there can be no accurate time domain - Fangyi: We should require GetWave - Ambrish: Init_Returns_Impulse would not be required? - Fangyi: That makes it easier for the model developer too - Fangyi: The redriver RX output drives the TX input continuously - Ambrish: But the TX has Pullup and PullDown, how is the stimulus generated? - Arpad: Only [External Circuit] can have a pure analog input like that - Todd: We derive an impulse response for each channel in advance - A digital input is used for that part - The result of that is used for the algorithmic simulation - Arpad: It is confusing that it says the "TX analog model" is driven - Todd: It should say "algorithmic" - Fangyi: We are talking about how the TX algorithmic drives the TX analog - The order is given in the paragraph - Arpad: In simulation the order is not the same - Fangyi: It is the same - Michael M: We are back to the disagreement on whether TX input is analog or digital - Fangyi: Do we agree on the flow order? - Todd: In the reference flows the impulse response is passed to TX Init - So for Init TX analog comes before algorithmic - For GetWave it is reversed - Walter: There are 3 ways to derive impulse response for the downstream channel - BIRD 158 describes Tstonefiles - The analog channel includes the TX driver - Arpad: How can the sentence be changed to make it clear? - Todd: At first I would have changed "analog" to "algorithmic" - Do we agree the upstream and downstream channels are evaluated independently? - Fangyi: Yes - Todd: The repeater TX Init will receive repeater RX algorithmic Init output - Fangyi: No, it will get the analog - Todd: You are right - The repeater RX GetWave gets the TX GetWave output - Once the impulse response is derived, the analog models can be thrown away - Ambrish: So the sentence is not needed? - Fangyi: Yes, it could be deleted - Arpad: This should be handled by email, but it has to be done by Friday - Fangyi: The issue is between Arpad and Ambrish, and should be discussed here - Todd: The picture is useful, but the sentence only complicates things - Fangyi: The details are still given in the next paragraph - Fangyi removed the "Note that" sentence - Fangyi: Step 8 is split into a and b for redriver and retimer - Walter: Agree that we should add the statement about half UI offset - Ambrish: The new signal is produced by the simulator, not the model? - Michael M: In a real device there would be a bit pattern - Walter: After resampling it should have the same bits - Todd: Fangyi is right, passing a waveform forward is not helpful - Letting the redriver RX recover the signal lets us do upstream diagnostics - The retimer RX and TX do not know they are part of a retimer - Ambrish: What do the physical devices do? - Walter: They do actually recover the waveform - Todd: The RX signal may be 80mV and it is converted to +/- 0.5V - Arpad: I thought the TX input was not digital - Todd: For a redriver - Ambrish: Where does it say the digital signal is not created? - It would be nice to avoid it if not necessary - Todd: The current proposal just allows for better analysis - Fangyi: I sent an email asking about a noise issue - Walter: RX noise affects the BER, not the waveform - Fangyi: Walter made changes for step 7a and 7b - Walter: Both RXs have a BER, and the repeater itself can have jitter - Ambrish: We decided to use numbered channels, not "upstream" and "downstream" - Fangyi made this change - Arpad: It should stated that the Repeater AMI param is in Reserved_Parameters - Bob: It also needs usage rules - Walter: It should "require" GetWave_Exists, not have an "if" - Fangyi: Is this in the TX or RX AMI file? - Walter: It could be in both - Michael M: If in one only does that mean a non-repeater model could be used? - Ambrish: The RX model for this is generic, any RX would do - Todd: It's not a big deal to add this keyword to it and use it - Arpad: It will not be clear why this has to be added - Todd: It really should be handled at component level, but still better to handle in AMI - Ambrish: Do we need it in both places? - Todd: No - Bob: This is global, it will apply to all RX pins - Todd: Most users don't look at AMI files closely - It is better to make sure the tool gets it right - Arpad: If not in the AMI file can the tool figure it out? - Todd: It would not know if it is a redriver or retimer - It might figure it out from clock ticks, but sometimes those are delayed - Michael M: Can we mix and match TX and RX? - Todd: There is a way to take a full model and turn things off to use it - Fangyi: I will update and send to Walter - Then it will go to the reflector - Arpad: It has to be submitted to Open Forum by Friday - Todd: We should have an edited version for all to review by tomorrow - Michael M: It needs to be receive by Open Forum before Saturday - Minor editorial changes can be approved at voting time - Fangyi changed "Repeater" to "Repeater_Type", with approval - Walter: I will work with Bob on the Labels BIRD - Radek: I have comments on BIRD 158 - Ambrish: I don't think we need it - Walter: It serves a need that has been discussed AR: Fangyi submit updated BIRD 156 to Open Forum AR: Mike post update BIRD 156 to ATM web ------------- Next meeting: 28 May 2013 12:00pm PT ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives