================================================================================ IBIS INTERCONNECT TASK GROUP http://www.ibis.org/interconnect_wip/ Mailing list: ibis-interconnect@freelists.org Archives at http://www.freelists.org/archive/ibis-interconn/ ================================================================================ Attendees from October 25 Meeting (* means attended at least using audio) ANSYS Curtis Clark Cadence Design Systems Bradley Brim Cisco David Siadat Intel Corp. Michael Mirmak* Keysight Technologies Radek Biernacki, Ming Yan Mentor, A Siemens Business Arpad Muranyi* Micron Technology Justin Butterfield*, Randy Wolff* SAE ITC Maureen Lemankiewicz, Logen Johnson Signal Integrity Software Walter Katz*, Mike LaBonte* Teraspeed Labs Bob Ross* University of Aveiro in Portugal Wael Dghais Michael Mirmak convened the meeting. No patents were declared. Justin Butterfield took minutes. Review of Minutes: - Michael called for review of the minutes from the October 11 meeting. Mike LaBonte moved to approve the minutes. Bob Ross seconded. The minutes were approved without objection. Review of ARs: - Arpad Muranyi to write text on the selection mechanism. - Arpad reported that there were some email discussions, but that no text has been written, as we need to have some discussion and make a decision on this issue. - Bob to add the Unused_port_termination text to draft10. - Bob reported this is done in bird189.5_draft10_v1, but the version Radek Biernacki sent out has the same name and different edits. Opens: - Michael added a line item to the agenda to go through an update from what was discussed at the ATM meeting. Summary of IBIS-ATM Discussion: Arpad stated in this week's ATM meeting there was some discussion on the Unused_port_termination, and Radek's changes were discussed. Walter Katz had indicated that as long as the parameter is required he would be okay with Radek's suggestions. Radek had sent out a version of the document with his suggestions. Arpad noted in the previous week's ATM meeting, they discussed the selection mechanism. Although, there were no documents generated from this discussion. Resolve multiple pins, one Group treatment: Mike suggested to discuss the multiple pins in one Group later on. Resolve selection mechanism for Model Sets vs. other keywords: Arpad stated he would like to have the Groups be the only selection mechanism. The issue is that in the Sets there could be multiple models overlapping. Arpad noted that Walter is proposing that victims pins must be unique while aggressor pins may be duplicated. Arpad suggested to have an example to show the issue. He commented that the issue is that there are no restrictions on multiple models covering the same pins. Walter reiterated that victim pins must be unique, while aggressor can be duplicated, and he commented that the issue is when a pin is only an aggressor and never a victim. Bob commented that only one Group is brought into a simulation and different Groups can have the same pins. There is no selection within the Group, and it defines the passive structure in the simulation. There should not be any duplication within the Group. Bob's point of reference is that everything should be contained within a Group, and there should be no collisions between the pins. Walter commented that companies generate their models by selecting the victim pin, then finding its aggressors. Model makers would want to put these models in Groups. Michael asked if we are in agreement of what we mean by Groups. Mike agreed that this is a good question and asked if the names "Sets" and "Groups" are appropriate. Arpad asked if we need the two levels of distinction between Sets and Groups. Mike stated that we do need this, and he noted that Sets can be delivered in an external file. Bob thought the Group is a powerful concept. He noted the DRAM example Randy Wolff had sent out a long time back with different partitions. Bob commented that his view of the issue is what Sets can be brought into the Group. Arpad gave an example of a 10 pin model with one victim and 4 aggressors. He asked how to handle all the signals for this example as well as power. Walter commented that the model maker would have to make multiple models to cover all cases for the victims. Arpad asked about the edge of the board. Walter commented that you can scan the board and every pin could be an aggressor. Aggressor means that in this model the net is not a suitable victim and to find another model. Randy commented that signals on the edge of the board will simply have fewer aggressors and less coupling. Arpad commented that there are no rules for what pins and pads can be in Sets and Groups. Walter stated his rules addressed some of these concerns. Bob noted that power delivery Sets should be different from signal only Sets. Mike thought the rules should be the same between Sets and Groups. Arpad asked if we need two selection mechanism for Sets and Groups. Walter suggested that the Groups should be the only selection mechanism. Walter asked how we plan to support the many pins with few models, as this is how models are currently generated. Walter noted we should support all cases of pin to buffer, pin to pad, and pin to pad to buffer. Bob stated he would like to limit the models from pin to buffer. If the model make does not provide connections for pad to buffer, then they should provide an explicit short. Bob commented that we are opening the door to models that are incomplete. Randy commented that models from pin to pad only would be likely and would be missing the pad to buffer connection. In some cases, there may be no need to model the pad to buffer interconnect, and this is a valid use case. He added that another use case is a bare die model, where you are only considering the buffer to pad interconnect and there is no package. Bob commented that we have limited our selves to pad names to match the pin names, and there could be naming issues. Arpad noted we have still have the one to one pin name to pad name restriction. Michael suggested that Arpad, Bob and Walter prepare a bullet point list of rules that need to be added for the Sets and Groups. Arpad noted that Walter and Bob both sent out the rules already. Arpad asked if Walter would be okay to start with Bob's rules. Walter stated he would like to make Bob's rules more specific. Arpad will take an AR to start some email discussion on the set and Group rules. Mike stated he would like to address the Unused_port_termination as the top issue for next time. The next meeting is scheduled for Wednesday Nov 1st. Mike moved to adjourn. Randy seconded. The meeting adjourned without objection.