IBIS Macromodel Task Group Meeting date: 03 February 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: We need to speed up work on an actual BIRD. -------------------------- Call for patent disclosure: - None ------------- Review of ARs: - Arpad to review IBIS spec for min max issues. - In progress. ------------- New Discussion: DesignCon Summit Review: - Walter: Todd gave the same presentation as the last ATM meeting. - Ken Willis gave the Cadence presentation. - There was push-back from Cadence on including scenario 3 in BIRD 147. - They have models out there now. - One objection was the stimulus pattern features in the BIRD. - Ken seemed to be OK with dropping that. - I plan to rework BIRD 147 and share with Cadence. - Michael M: There was good agreement on some points. - Rita Horner raised concerns about structure. - Walter: It's the difference between 802.3 and PCIe3g. - In 802.3 training happens at the PHY level. - In PCIeg3 the MAC does training, a level above the PHY, an outlier. - There can be cases where the PHY and MAC come from different vendors. - Todd: The MAC would be looking at eye quality. - Ambrish: The MAC is a hardware layer? - Todd: Two pieces of IP work together to send commands to the TX. - Walter: The MAC may be software too. - Todd: At the near end of the network the packets could come from either entity. - For simulation this may not matter. - For plug and play operation it may matter. - Ambrish: Is the MAC another model? - Todd: The RX has to produce something for the TX exploration algorithm to look at. - Fangyi: Is the MAC standardized? - Walter: It can be proprietary. - Fangyi: It is another black box we need to add? - Michael M: The issue is that parts of this might come from separate companies. - Michael M: Is it possible for both sides of the IP to interpret the data differently. - We can't see who is listening to the backchannel. - Walter: This is a PIPE. - Only the MAC talks to the TX. - Todd: ... If we define things correctly. - Walter: We need to talk to the people who generate these. - Todd: Cadence agreed with scenarios 1 & 2 - They would prefer to put off scenario 3. - This is scenario 4 which we need to understand. - SiSoft and Cadence need to find an approach we can sign off on. - Arpad; David Banas requested support for legacy models. - Radek: There is no difference between 2 & 3. - Ambrish: They are different. - In 2 the intelligence is in the AMI model,not the EDA tool. - Walter: Scenario 1 is GetWave and 2 is Init. - Cadence and SiSoft agree both are needed. - Arpad: Did David comment on this? - Walter: IBIS is a container of data, not how tools operate. - Nothing prevents a tool from doing 3. - Arpad: For 3 we need a model interface. - Radek: Something has to happen to enable it. - Walter: Agree. - Agree with Cadence too. - An independent BIRD can be worked on in parallel. - Arpad: When will the BIRD 147 rework be ready? - Walter: I will be on vacation for 2 weeks. - Arpad: We should discuss C_comp next week. - Randy: I have nothing new, but could work with Walter. - Michael M: We could go over the summit presentations. - It might tie into Bob's K-table presentation. - Bob:We should not resurrect BIRD 79 - IBIS-ISS gives us a general solution. - Randy: We need more detail on series elements. - Arpad: We should discuss redriver flow. - Walter: I gave all-Init and all-GetWave flows. - Everything else is up for grabs. - Michael M: Is this in addition to BIRD 166? - Walter: It was an email. - Fangyi: We need to consider crosstalk and multiple redrivers. - Arpad: I could prepare something on min/max issues. - Michael M: I can propose some reserved parameters. - Michael M: Should we invite Rita Horner to talk about layers? - We should make sure our solution meets criteria. - Walter: She could join the Open Forum meeting Friday. - Walter: There might have been at least 20 papers involving IBIS-AMI at DesignCon. - Some were about PAM4. - Some were using flows other than the reference flows. - Todd: Xilinx compared 6 EDA simulators. - It was unknown which were correct. - Michael M: The impulse response difference was interesting. - Ambrish: Were there many attendees? - Michael M: There may have been slightly fewer than last year. - Arpad: Do we have any options for multi-level signaling? - Walter: I had proposed a simple "this is PAM4" keyword. - The RX needs to report 3 signals and clock skew. - Fangyi and I have agreed on this. - Arpad: Can it use legacy IBIS? - Fangyi: Some IC vendors see no urgent need to change the legacy IBIS language. - That is not formal. - Arpad: It might be handled algorithmically. - Fangyi: Did Walter mention something about the RX returning slicer info? - Walter: I did not. We need to resolve that. - Ambrish: We would like to be involved. ------------- Next meeting: 10 Feb 2015 12:00pm PT ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives