IBIS Macromodel Task Group Meeting date: 20 January 2015 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 IBM Steve Parker 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 Teraspeed Labs: * Bob Ross (Note: Agilent has changed to Keysight) The meeting was led by Arpad Muranyi. ------------------------------------------------------------------------ Opens: - Arpad: Hoping Michael Mirmak will be able to give the ATM summit presentation. - I will not be there. -------------------------- Call for patent disclosure: - None ------------- Review of ARs: - Arpad to review IBIS spec for min max issues. - In progress. ------------- New Discussion: Co-optimization: - Todd showed IBIS-AMI and Co-Optimization. - Todd: I'll just explain the changed slides. - slide 5: - Todd: This describes all 3 scenarios on one slide. - Ambrish: No simulator involvement except messaging for 1 and 2? - Todd: Correct, but the simulator has to know when training ends. - Ambrish: It is the same for 1 and 2. - Todd: 1 and 2 are functionally the same. - Arpad: There may be a big difference in what is simulated. - Todd: But 1 and 2 use the same plumbing. - Ambrish: Should it be 1a and 1b? - Todd: I have no preference. - Mike: 1 and 2 differ regarding GetWave and Init. - Walter: We are mixing functional and user requirements. - One BIRD can be agnostic whether we are doing 1 or 2. - Todd: We should move on and come back to this. - slide 6: - Todd: The point about local and global optima has been removed. - slide 7: - Todd: Need to be clear it has to support industry protocols. - slide 10: - Todd: Calling this Simulator Based Co-Optimization now. - Important to note that meta-data must be supplied. - slide 17: - Todd: An algorithm outside the RX knows how to explore TX settings. - The blue boxes around things are gone. - Those functions have to exist but we don't care how and where. - slide 19: - Todd: Scenario 3 has different combinations of models that don't know how to communicate optimization - slide 20: - Todd: In 3a The RX thinks it is participating. - The simulator is taking over for the TX, communicating with the RX. - The meta-data helps the simulator do that. - slide 21: - Todd: In 3b The TX is participating , simulator takes over for RX. - slide 22: - Todd: In 3c The simulator takes over for both. - Todd: I propose presenting these slides at the summit. - slide 21: - Fangyi: Please explain the 3 levels. - Todd explained the Simulate, Self-Optimize, and Co-Optimize functions. - Slide 15 explains the case where models do not change settings during simulation. - In slide 16 the RX changes its settings automatically. - Fangyi: We support that today. - The blue box shows what is in the RX, right? - Note the TX can also do self-optimization. - Todd: On slide 18 we have the co-optimization function added in. - The 3 bands show that these functions must exist no matter where they are. - Fangyi: There may be too many colors here. - slide 20: - Fangyi: Yellow color should be used for the parts that the simulator does. - Radek: It would be good to have the requirements listed independent of scenarios. - There may be 10 to 20 requirements. - The scenarios are affected by what the models have. - Todd: We may start looping back on ourselves. - Arpad: Could we add slides for what our meeting goals are? - You were hoping for consensus by the time of the summit. - We may want feedback on this at the summit. - Todd: We do want that. - Ambrish: Are these requirements presented as what we agree to? - We may not agree on scenario 3, it can wait. - What do we expect of the audience? - Walter: The TX configurator is a requirement from David Banas. - Radek asked for the EDA tool to participate in co-optimization. - Who does not support scenario 3c? - Ambrish: We believe that can wait. - David: If and RX and EDA tool can do this, the pressure is put on the TX to support it. - A legacy TX model should not obstruct the process. - Todd: This suggests 3b is important. - Arpad: If we do 3a, 3b and 3c fall out of it. - Radek: Agree. - Ambrish: Can we have a debate that this is not as important? - Arpad: We seem to have support for scenario 3 now. - Ambrish: Six months ago we had support for a proposal without this. - Walter motioned to vote on supporting 3a, 3b, and 3c. - SiSoft would propose a BIRD if it passes. - Todd: We should bring this to a wider audience at the summit first. - Arpad: Is the BIRD for all scenarios? - Walter: Yes, all requirements in this presentation. - Arpad: Feedback from the audience would be best. - Walter: Objection: we have the right to propose what we want. - Arpad: The question is if we are asking the audience for agreement. - Walter: Ambrish has complained about this process. - Ambrish: I'm not complaining. - Todd: This formulation is two weeks old. - We need to reduce acrimony. - This could be a status report or it could be a request for input. - Arpad: We have the right to put what we want in a spec, but we need to solicit feedback to make sure it is useful to the industry. - Todd: Agree. If they want it, the need is validated. - Arpad: Could we have a slide asking for feedback? - Todd: Yes. - We debate here because we are authorized. - It will have more meaning if we bring it to a wider audience. - If they don't care about it, that is useful input. - The first slide can say where we are in the process. - The last slide can be a call for feedback. - Bob: The presentation should be submitted by tomorrow. - Bob: We can vote after DesignCon. - Walter withdrew his motion. - Arpad: There is no need to post this presentation to the ATM archive, it will be posted for the summit. ------------- Next meeting: 03 Feb 2015 12:00pm PT ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives