================================================================================ 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 November 8 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 Mike LaBonte convened the meeting. No patents were declared. Justin Butterfield took minutes. Review of Minutes: - Mike called for review of the minutes from the November 3 meeting. Mike commented that the date for the next meeting should say November 8th. Bob Ross moved to approve the minutes with the correction as noted. Randy Wolff seconded. The minutes were approved without objection. Review of ARs: - Arpad Muranyi to work with Bob and Walter on Sets and Groups language. - Bob reported that they have been discussing it, but no resolution has been settled. Opens: - None File_TS0: Bob stated that he has some comments on the email from Radek about the File_TS0 restrictions. Radek seems to be opposed to including the File_TS0 option. Bob proposed to add an actionable statement in the BIRD about any restrictions. The issue is mixing and matching interconnect models with node 0 as the reference. Walter's position is that model makers know what they are doing, and we do not need to add restrictions to the specification. Bob noted that there could be issues with connecting the ground rails for mismatched reference cases. Mike commented that if you have are using File_TS0, you are bypassing the PDN. Bob commented that Radek's statement about current not passing through the reference needs to be clarified. Groups and Sets: Bob noted that we need to come to an understanding on the disagreement about Groups and Sets. All of the boundaries between the models must be considered. Mike asked about the status of this issue. Arpad noted that from email discussions Bob and Walter have a disagreement about restrictions placed on model makers. Arpad suggested that the technical issues should be resolved first. Bob commented that the issue is the boundaries of the group. Groups should contain complete paths for the desired group of pins. Randy asked in the case the Group is not a complete path if we need a shortcut. Bob stated that we need to be explicit in the paths and eliminate the implicit paths, as this opens the door to incorrect models. Mike noted that Requirement 12 in the BIRD189.5 document states that the interconnect model should contain the complete path. Bob stated that this is his concern, and he would like to remove the implicit paths. Mike asked about Requirements 14 and 15. Bob stated that Requirement 14 is true. For Requirement 15, Bob thought we should not mix model format types. We should not use one style of model for buffer to pad then use the [Pin] list RLC for the pin to pad. Mike suggested that we should go through requirements 12, 14 and 15 and look to see if we have text in the BIRD to support these. Bob noted that groups should be self contained and not affect other groups. Mike asked Randy if he would want to mix and match models between formats. Randy thought that for slow speed signals, he would use the old style model, and for high speed signals, he would use the new style model. Randy asked about the stacked die case where you only want to model the buffer to pad connection inside the IBIS file, while the package pad to pin model would be in an EMD model. Bob noted that we do not support buffer to pad currently. Arpad stated that we should support this use case. Bob commented that currently for bare die models the package RLC values are set to 0. He suggested to have a standard Group consisting of buffer to pin, but we could add separate Groups for partial paths from buffer to pad or pad to pin. Randy stated that everything uses [Component] as the top level. In the case where we use the [Component] for the bare die, we could create a special case that the group would be buffer to pad only. Mike agreed that this would be a good solution. Bob suggested that this should be a separate keyword from the normal Group. Bob commented that different die at different levels could have different pad names in the stacked die case. Randy stated that this could be modeled with EMD. Bob noted that we can delete the full path recommendation for Sets that is currently in the BIRD. He thought we should continue the discussion for full paths in Groups. IBIS-ISS Referencing: Mike asked about the word reference in the IBIS-ISS examples. Arpad had proposed to remove the word reference for the IBIS-ISS examples, as he does not want the user to get confused by the term "reference". He noted that this is not yet done. Bob thought that we should come to agreement on this and update it for BIRD189.5_draft11. The next meeting is tentatively scheduled for Wednesday November 15. Randy moved to adjourn. Arpad seconded. The meeting adjourned without objection.