====================================================================== 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 May 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 Graphics 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. Randy moved to approve the minutes of the May 11 meeting. Bob seconded the motion. No objections were raised and the motion carried. No other opens were raised. AR: Mike LaBonte to look for 1:1 pin-pad-buffer assumptions in the BIRD text. Mike said this was done and can be discussed. AR: Michael Mirmak to fix diagram vs. Table XX (e.g., Signal Pins) - the diagram may simply be deleted from the draft. Bob Ross noted that the figure is deleted from draft 34. Mike showed a document comparing drafts 33 and 34 of the Interconnect BIRD draft. Referring to his AR he said he found a sentence about 1:1 pin-pad-buf relationships in the first paragraph of the Connecting Pins, Pads and Terminals section on page 13. Arpad noted that we have discussed and agreed to that previously. The list of definitions was found to exist in both the ANALYSIS/PATH section and the actual proposed specification text. Mike deleted it from the first section as we had previously agreed to have it in the specification. Bob noted that the definitions would get more scrutiny as such. Randy suggested adding some text to the ANALYSIS/PATH section so that it would not be empty. Mike said it's not clear whether there can be multiple [Interconnect Model Selector] keywords. Bob said it works like [Model Selector], with multiple interconnect model names following the single [Interconnect Model Selector] keyword. Arpad noted that the second paragraph in the [Interconnect Model Selector] usage rules says there can be none, one, or more than one Interconnect Model for a component. Mike said it was unclear to him if "Interconnect Model" there referred to an instance of an interconnect model or a definition of one. He suggested that an [Interconnect Model Selector] might correspond to an instance and [Begin Interconnect Model] might correspond to a definition. Bob said we need to define that clearly in the [Interconnect Model Selector] section. Arpad asked if we should allow more than one [Interconnect Model Selector] keyword per component. Bob replied that there must be only one, and that everything must be in it. Mike said that if a wrapper must be created to make models complete, we would need to have a sufficiently useful ability for ISS subckts to form wrappers by calling other subckts. He wondered if IBISCHK would be able to find all of the required files to check them if they are not all listed in the .ibs file. Arpad asked if this would be like [Model Selector] where functionally equivalent variants are selected, or if [Interconnect Model Selector] allows choosing a set of models from within it that can be used together. For example if a chip has 3 bus types and separate package models for each, would these be listed under one [Interconnect Model Selector]? Or might [Interconnect Model Selector] be used to select either an uncoupled or fully coupled package model for the same set of pins? Alternatively would there be three [Interconnect Model Selector]s because there are three buses? Bob said all of those could be contained in the [Interconnect Model Selector]. Arpad noted that the BIRD sets the first [Interconnect Model Selector] as the default, implying that only one of its choices can be selected. Therefore an [Interconnect Model Selector] would be needed for each bus to be analyzed. Bob asked if we intend to analyze only one bus at a time. Arpad replied that it would not always be that way, but flexibility would be desirable. Arpad said we probably would need multiple [Interconnect Model Selector]s for cases such as where separate buffer-to-pad and pad-to-pin models must be combined to form a complete path. Bob said those would be under one [Interconnect Model Selector], that one selection could correspond to a group of models. Arpad asked if two models are needed for one simulation and there are two variants of each of those how would the tool choose valid combinations from the four models listed? Bob suggested that we could have named [Interconnect Model Selector]s where the names are informational, and each contains a complete set of models. Mike suggested calling it [Interconnect Model Set] since we would not be selecting from within it. Bob said that might work. Mike asked if we should require [Begin Interconnect Model] to include all pins used for a simulation, allowing a single [Interconnect Model Selector] to select from among what would then be equivalent models. Bob said the user or even the EDA tool is expected to do this. Arpad said that would force model makers to instantiate a full model when it might be more convenient to have partial models, for example calling a subckt four times. He noted that requiring subckt calls to other subckts may be cumbersome. Bob said an ISS wrapper may be needed to call a bunch of sp4s, for example. Walter said that Randy for example would work with EDA tools to use whatever facilities they offer, but IBIS would need the flexibility to do anything. Arpad noted that we still had not settled whether multiple [Interconnect Model Selector]s would be allowed. Walter said there should be no limit. Bob said we need to be able to support any or all combinations of Interconnect Model connections, for example buffer to die, die to pin, buffer to pin, coupled, uncoupled, etc. Bob noted that we need to give the examples in the BIRD draft better names. ARs: Mike LaBonte to post Interconnect BIRD draft 35 with changes as discussed. Arpad moved to adjourn. Bob seconded. The meeting adjourned.