IBIS Macromodel Task Group Meeting date: 22 August 2017 Members (asterisk for those attending): ANSYS: Dan Dvorscak * Curtis Clark Broadcom (Avago): Xingdong Dai Bob Miller Cadence Design Systems: * Ambrish Varma Brad Brim Kumar Keshavan Ken Willis eASIC: David Banas Marc Kowalski Ericsson: Anders Ekholm GlobalFoundries: Steve Parker IBM Luis Armenta Trevor Timpane Intel: * Michael Mirmak Keysight Technologies: Fangyi Rao * Radek Biernacki Ming Yan Maxim Integrated Products: Hassan Rafat Mentor, A Siemens Business: John Angulo * Arpad Muranyi Micron Technology: * Randy Wolff Justin Butterfield SiSoft: * Walter Katz Todd Westerhoff * Mike LaBonte Synopsys: Rita Horner Kevin Li Teraspeed Consulting Group: Scott McMorrow Teraspeed Labs: * Bob Ross TI: Alfred Chong The meeting was led by Arpad Muranyi. -------------------------------------------------------------------------------- Opens: - Arpad noted that Curtis had informed him that he would not be able to attend the meeting August 29th. Arpad said he would need a volunteer to take minutes at that meetings. ------------- Review of ARs: - Radek to send out an updated BIRD 158.6 draft 3 with changes as discussed. - Done. - Arpad noted that Radek's new version contained more changes than he had expected to see based on the discussions at the August 8th meeting. - Radek noted that at the August 8th meeting he had said he would feel free to add new parameters to give the model maker more ways to describe the content/usage of the package models. He said the version he sent out after the August 15th meeting was consistent with his statement on August 8th. - Arpad agreed, and Arpad and Radek noted that Walter had sent out a follow up proposal after Radek had emailed draft 3. - Randy to send C_comp_model_Using_IBIS_ISS_BIRD_rev1_WIP.docx to Mike L for posting. - Done. -------------------------- Call for patent disclosure: - None. ------------------------- Review of Meeting Minutes: - Arpad: Does anyone have any comments or corrections? [none] - Radek: Motion to approve the minutes. - Randy: Second. - Arpad: Anyone opposed? [none] ------------- New Discussion: BIRD158.6: - Arpad: After Radek posted his updated version, then Walter proposed one that removed everything related to Ts4file_Boundary and moving the connection points from buffer terminal to pad, or pin. - I'd like to get Radek and Walter's comments and then open up for discussion. - [sharing side-by-side Walter's and Radek's versions] - Radek: The right side is what I posted and we discussed last week. - My version contains additional parameters to provide package model description information that was suggested by Bob Miller. We need to get his opinion. - Walter then decided to remove the pin and pad locations and all the packaging options. - I think it's best to give the model maker a way to specify this package info for the EDA tool and the user. - We need to figure out the best way to fulfill the needs of the model maker. - Walter: The original concept of BIRD158 was a shortcut for [External Model]. - [External Model] is for the buffer (it is a replacement for [Model]). - It's what people do for their on-die S-parameters. - Then some combination of Radek and others decided to allow the Touchstone file information to go to the pin. - As a practical matter, no one does this, and it just complicates things. - I didn't object at the time because I was going to just have it go to the buffer by default. Everyone makes their models this way, and it wouldn't be a problem. - But then we got into all these involved discussions about packaging and interactions with the Touchstone file. - The bottom line is if the Ts4file includes the package then you should put a zero package model in the IBIS file. Then you're done. I think we should go with simplicity. There's no need to complicate this BIRD and the parser with additional keywords that no one will use. - On-die S-parameters have been around almost since the beginning of IBIS. - People have been doing this for ten years. Let's keep it simple. - Let's not make EDA tools worry about dealing with odd combination when the model maker can simply make the package model compatible with what's in the Ts4file. - Bob R.: I support Walter's approach of simplicity. - This BIRD has spun out of control. Deleting all packaging related parameters is the best solution for a number of reasons. - You can always cascade the Ts4file with the existing IBIS package model and make the package model zero. - EDA tools can give options. - I think Walter's version should become BIRD158.6 draft4. - Arpad: I too prefer Walter's version. - I feel similarly that Ts4file was simply supposed to replace [Model]. - In light of BIRD189, it would simplify things to consider this BIRD 158 to just connect to the buffer terminal. For legacy IBIS package models that's the same as the pad (since there was no on-die interconnect). But considering the BIRD189 interconnect model that may specify on-die interconnect, it simplifies things to consider this BIRD to connect to the buffer terminal. - Radek: I would like to hear from Bob Miller. - He's the one who wanted the pad as an option and the ability to provide a separate Touchstone file for packaging information. I think we have to do what's best for the model makers. If we don't accommodate the way he does his measurements then we are in trouble. - Arpad: If we have BIRD189 style models then I don't think he will be prevented from doing what he wants to do. - Walter: I don't like to be the one to announce this, but Bob Miller has taken early retirement from Broadcom and is no longer their representative. - I move to vote on whether I should prepare my version as BIRD158.6 draft4 and send it to the ATM. Next week we can vote on whether to recommend it to the Open Forum. - Bob R.: I second. I would like a roll call vote. - Radek: I can work with either proposal. I will abstain until we hear from Bob Miller. - Arpad: We have a motion and a second. - Roll Call Vote: ANSYS - abstain Cadence - abstain Intel - abstain Keysight - abstain Mentor - Yes Micron - abstain SiSoft - Yes Teraspeed Labs - Yes - Arpad: What does this vote mean? - Radek: I think the motion passes. We have a record of the vote. - Walter: I'll prepare BIRD 158.6 draft 4 before the next ATM meeting. - Bob R.: Could Mike L. go back and add cumulative revision history to draft 4? - It's missing from the recent previous versions. - Arpad: Walter needs to start from an earlier version that only has "buffer terminal" in the figures. - Radek: I had an earlier version that included the various figure changes, like Vp and Vn in the Tx figure, but did not yet include anything other than "buffer terminals" in the figures. Walter can start from there. BIRD 192: Clarification of List Default Rules. - Michael M.: This BIRD resulted from a brief email thread I started that related to Format List. - As currently described, the default value to use from a List may be ambiguous. - If the first item in the list is considered the default value (as in the initial description of List itself, List ...) it's not clear if this "default value" should be a duplicated member of the existing list or if it should be a separate member. - If the (Default xxx) specification is used, it's not clear how that interacts with the List statement that the first value is the default. - We have found that different tools interpret this differently. - We introduce new text to clarify the intent: - You can use a duplicated value from the list as the first entry (). - But the need not be a duplicate of another entry in the list. - If (Default xxx) is used, it overrides the first entry and provides the default value. - Ambrish: The first entry () is simply an entry in the list. - Arpad: I think the is separately listing something already in the list. Otherwise, we could have just said for the first entry, and stated that the first value is the default value. - Radek: I thought it was clear that if (Default xxx) were specified then it would override the first value. - I think we should clarify whether duplicate entries are still allowed in the example in the last sentence with the (Default xxx) value: (List 0 1 2 3 4)(Default 2) - Walter: That example is exactly why we left the optional Default specification in for List. Sometimes you don't want the first entry in an ordered list to be the default. That last sentence example properly demonstrates the original intent. - Michael M.: Perhaps it would be clearer to just say: List ... - We can then say that the first value in the list is the default (if (Default xxx) is not specified). - Arpad/Curtis: If you change the first entry to , what do we do with the possibility of duplicates. - Ambrish/Bob R./Walter: Unlimited duplication of the same value in a list is always permitted. - Bob R.: The parser will complain if a (Default xxx) specifies a value that is not a member of the list. - Walter: I suggest Michael M. can make a BIRD192.1 with this change and submit it to the Open Forum. - Walter: Motion to adjourn. - Curtis: Second. - Arpad: Thank you all for joining. AR: Walter to create BIRD158.6 draft 4 with his version that removes the Ts4file_Boundary and related parameters. AR: Mike L. to add cumulative revision history to Walter's BIRD158.6 draft 4 and post it. AR: Michael M. to create BIRD192.1 with the discussed change of the first item in the List description from to . ------------- Next meeting: 29 August 2017 12:00pm PT ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives