================================================================================ 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 January 24, 2018 Meeting (* means attended at least using audio) ANSYS Curtis Clark Cadence Design Systems Bradley Brim Intel Corp. Michael Mirmak* Keysight Technologies Radek Biernacki Mentor, A Siemens Business Arpad Muranyi* Micron Technology Justin Butterfield*, Randy Wolff* SiSoft Walter Katz*, Mike LaBonte* Teraspeed Labs Bob Ross* 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 January 19 meeting. Mike LaBonte moved to approve the minutes. Arpad Muranyi seconded. The minutes were approved without objection. Review of ARs: - Arpad to add text to introduce his version of the Aggressor graphic. - Arpad reported this is done. - Mike to fix missing spaces in the BIRD189.5 draft. - Mike reported this is done in BIRD189.5_draft16_v2. - Mike to send out BIRD189.5_draft16_v1 and BIRD189.5_draft16_v2. - Mike reported this is done. - Arpad to send out BIRD189.5_draft16_v3 with his changes. - Arpad reported this is done. Opens: - Bob Ross mentioned an editorial issue that needs changed. The term "rail_pin_name" should be "rail pin_name". Michael asked if this is a global change. Mike replied that is only an issue in three instances on page 9. Arpad moved to change "rail_pin_name" to "rail pin_name". Mike seconded. There were no objections. - Michael stated he will have to leave the meeting early. Review from ATM: Michael asked what the latest version of the document is now and if there was further feedback from ATM. He noted there are now some comments in the document that need to be resolved. Bob stated there are some technical questions, and he is not ready to vote on the BIRD in the Open Forum. Michael asked if we have a draft that can be uploaded. Arpad stated that draft16_v4 is the latest. Michael suggested we formally upload the document as a baseline draft. Mike clarified that there is a difference between uploading the latest draft versus submitting the BIRD to the Open Forum as the official BIRD189.5. Bob suggested to work on changes today then upload that version as draft16. Bob stated that it is a separate question if we want to submit the BIRD to the Open Forum as a snapshot. Michael suggested that we wait toward the end of the meeting to decide. BIRD189.5_draft16_v4 review: Mike noted he had some suggested edits in an email. Mike asked about the comment on page 10 which states the text has "awkward English". Arpad noted that he added this comment after the ATM meeting based on the discussion. Mike suggested add an "a" to "It is necessary...". Mike noted that this section is stating that the rules apply to Interconnect Model Groups and as a result the subsequent Interconnect Model Sets. Bob suggested to leave this as a question to be edited later. Mike showed his email comments which suggest some changes. In the I/O pin_name rules section at the second to last bullet, he proposed clarifying to say that Aggressor_Only signals may still be used as a victim in crosstalk simulations. Randy Wolff commented that the double negative in the first sentence is confusing. Arpad stated that Mike's changes make the rule no longer accurate. Mike stated that the user wants to use the signal as a victim, but all the tool finds is Aggressor_Only pins. Arpad stated if you have a model as Aggressor_Only it can not be used as victim, since it does not have cross talk. Randy pointed out that the second sentence qualifies the first sentence by warning what the user is missing. Mike commented that we created the Aggressor_Only tag, then we added these rules which create an exception when there is no suitable victim. Arpad stated that the name Aggressor_Only is not quite accurate if the signal can be used as a victim as an exception. Walter Katz noted that originally it was named Aggressor rather than Aggressor_Only, and he thought the name Aggressor_Incomplete might be better. He stated that we have been discussing for many weeks what to do when the model maker does not give good models and have made few changes to the actual keywords. He recommended that we should submit the BIRD to the Open Forum to get more feedback. Mike proposed to remove the last two bullets in the I/O pin_name rules section. This would leave what to do in these cases up to the simulator. Walter moved to remove the last two rules in the I/O pin_name rules. Arpad commented that this discussion started with the decision that we only wanted one selection mechanism, and it would be for the Interconnect Model Groups. These rules were added to remove the need for an additional selection mechanism. Walter stated that we do not need to cover all cases of bad models in the IBIS Specification. Walter noted that the intent of BIRD189 is for model makers to be able to wrap the models they are creating today, and we should keep things simple. Arpad thought we should write rules in the IBIS Specification to make bad models illegal. Mike noted that we have always allowed for incomplete models. Mike asked if Arpad's main concern was about saving the sub-bullets under the last I/O pin_name rule. Arpad replied that his concern is if you look through the Interconnect Model Group and you find the same pin in different models, how do choose which model to simulate. There was no second to the motion and Mike declared the motion dead. Mike suggested to post the current working document we have as draft16. Arpad suggested to resolve some of the comments first. Arpad stated there were questions on the sentence from page 7 which appears repetitive. Mike noted that he made some suggestion in his email. He clarified the term "first" as it could be determined based on the hierarchy of the Interconnect Model Group versus the Interconnect Model Sets in the file. Bob thought the conditions make this a complicated rule, but the wording of the sentence was necessary. Mike asked about the the original issue with the last two rules in the I/O pin_name rules section. Randy stated he is leaning toward deleting these rules. Bob moved to delete the last two rules in the I/O pin_name rules. Randy asked if there is anything we lose by deleting this section. Arpad noted that without these rules there is an ambiguity regarding which model the EDA tool should select. We have a rule that you can have multiple models with the same pin as an aggressor but there is no way to know which model to select. Walter stated that which ever model is selected is wrong. He noted that we can generate a report showing the connections and the content of an Interconnect Model. He thought that we cannot solve this problem with rules in IBIS. Walter seconded the motion. Arpad noted if we were to vote that he would abstain. The roll call vote tally was: Mentor - Abstain Micron - Yes SiSoft - Yes Teraspeed Labs - Yes The roll call vote concluded with a tally of Yes – 3, No – 0, Abstain – 1. The motion passed. Mike will delete the last two rules in the I/O pin_name rules [AR]. Mike will post BIRD189.5_draft16 [AR]. Next Meeting: The next meeting will be February 7. Randy moved to adjourn. Arpad seconded. The meeting adjourned without objection. Task List BIRD189.5 editorial additions/changes to be completed: