IBIS Macromodel Task Group Meeting date: 17 November 2015 Members (asterisk for those attending): ANSYS: Dan Dvorscak * Curtis Clark Avago (LSI): Xingdong Dai Bob Miller Cadence Design Systems: Ambrish Varma Brad Brim Kumar Keshavan Ken Willis Cisco: Seungyong (Brian) Baek eASIC: David Banas Marc Kowalski Ericsson: Anders Ekholm GlobalFoundries: 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 SiSoft: * Walter Katz Todd Westerhoff Mike LaBonte Synopsys: Rita Horner Teraspeed Consulting Group: Scott McMorrow Teraspeed Labs: * Bob Ross TI: Alfred Chong The meeting was led by Arpad Muranyi. ------------------------------------------------------------------------ Opens: - None. -------------------------- Call for patent disclosure: - None. ------------- Review of ARs: - None. ------------------------- Review of Meeting Minutes: - Arpad: Does anyone have any comments or corrections? [none] - Walter: Motion to approve the minutes. - Arpad: Second. - Arpad: Anyone opposed? [none] ------------- New Discussion: New Redriver Flow BIRD: - Arpad: Fangyi and Radek mentioned last week that they could not attend today. - Therefore, we probably can't have much of a discussion on this today. - I would just like to get a quick update from Walter, as I was not quite sure where things stood with the off line email exchange. - I believe there were two suggestions, one involved deconvolution and the other involved an additional IR being added to the IR matrix. - Last week I asked the same question of Fangyi [Walter could not attend], and Fangyi indicated that he didn't think the issues could be resolved without the additional IR matrix data. - However, I saw a response from Walter that seemed to indicate that the deconvolution could work. - I just wanted to understand the current thoughts on whether deconvolution would work. - Walter: Deconvolution will work in the same way that it works in the current standard. - In the current standard, one of the nine combinations listed in our current flows relies on deconvolution. [Tx uses Getwave(), Rx uses Init()]. - Deconvolution is already required for this combination. - The combination Fangyi is bringing up is essentially the same. - Deconvolution is one answer, there are two others. - Arpad: Total of three options? - Walter: Yes. - Second option is to pass in two IRs (in addition to the crosstalk IRs), one IR determines the equalization, and that equalization is then applied to the other one. - Note that there are certain advantages to not using deconvolution. - Deconvolution works -almost- all of the time, but there are cases where it's problematic. - Third option: - If you think about what Fangyi is suggesting, he's saying that there's one IR that determines the equalization. When it's determining the equaliz- ation, it's determining two parts. One is the linear part, and the other is the full equalization, including a DFE, for example. - We could provide one input for the channel IR, and room to return two IRs. One is the IR of the full equalization of the Rx, and the other is the IR of just the linear part. - We could also do that if we took Fangyi's approach of passing in two IRs as inputs. But those two IRs could be set up by the EDA tool as unit impulse responses when passed in. The output of one would be the full equalization of the channel, and the other would be the linear equal- ization. Then the EDA tool could apply those two as it wishes. - I suggested that in the most recent email to Fangyi, and I haven't heard back yet. - Arpad - In the emails I hadn't realized that there was a third option. - Since I hadn't seen a response from Fangyi, I was wondering if he might have changed his mind about whether deconvolution would work. - I asked Vladimir here at Mentor, and he thought deconvolution would work. - Walter - I think Vladimir is correct, except... - We have seen some pretty awful IRs, including some with singularities, returned by some receiver models. Discussion of language corrections regarding "ground": - Walter: I think I've done the initial work on this, and now Radek and the rest of the group have to tackle it. - I've made specific proposals and people have suggested they'd like something different. - Arpad: Yes, it's now ready for a group effort. - Arpad: Does anyone have anything else to discuss today? [none] - What should we do about next week's meeting? Are many people traveling? - Walter: I move that we hold the next meeting on December 1st (in two weeks). - Curtis: Second. - Arpad: Anyone opposed? [none] - Walter: I move that we adjourn. - Arpad: Second. Anyone opposed? [none] - Thank you all for joining. ------------- Next meeting: 01 December 2015 12:00pm PT ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives