============================================================================== IBIS INTERCONNECT TASK GROUP Mailing list: ibis-interconnect@freelists.org ================================================================================ Attendees from June 11, 2025 Meeting (* means attended at least using audio) ANSYS Curtis Clark, Wei-hsing Huang, Juliano Mologni Arista Networks Jim Antonellis Broadcom James Church Chipletz Stephen Newberry* Intel Corp. Michael Mirmak*, Xiaoning Ye Keysight Technologies Ming Yan Marvell Steve Parker MathWorks Walter Katz* Micron Technology Justin Butterfield Siemens EDA Weston Beal, Arpad Muranyi*, Randy Wolff* Simberian Yuriy Shlepnev ST Microelectronics Aurora Sanna Synopsys Ted Mido, Edna Moreno University of Illinois Jose Schutt-Aine Michael Mirmak convened the meeting. No patents were declared. He noted that the minutes from the June 11 meeting had not yet been completed. The team discussed the changes in draft 14 of the port-mapping proposal. Regarding units, Arpad Muranyi asked about the use of "angstrom" as a unit, noting that it was unusual and shouldn't be needed with exponents. Michael took the AR to check on the Touchstone specification's exponential notation syntax rules, as well as the base units assumed by Touchstone. [AR] Michael took the AR to send out the latest draft to the reflector. [AR] Randy Wolff asked about the Bus_Label language, specifically whether cross-checking is implied against IBIS. Should the Touchstone parser look in IBIS files? The team agreed to remove related "should" language. Arpad expressed some concern about how "grouping" was used in the current draft, though several changes to clarify this had already been made in the draft. Arpad asked about data usage and data measurement in the supplied examples. The distinction between these concepts is not always clear. In IBIS, there is a fixed definition as to which port connects to which pair of notes; how does this related to the port-mapping usage of "Logical" and "Physical"? For instance, in Example 2, "Logical" is used, but Example 1 is also implied to use Logical associations. Michael suggested that Logical was intended for use with schematics, while Physical was for lab measurements or layout. Walter Katz replied that a schematic can have both. This can include pin number, refdes, and pin location. A Logical association doesn't indicate place in space. Randy referred to the example of a BGA ball; Michael noted the definitions for PCIe slot connectors, USB-C connectors and the like. Arpad stated that his main motivation for port mapping support was to automate the generation of EMD models. Touchstone files containing port information, including lists of pads, balls, etc. could be connected automatically. The current format doesn't tell you what this component will be connected to. Walter replied that the format should not do this; unless you are working with a PCIe connector where the functions are defined as associated with specific pins and locations, a connector is just that with nodes that can be used for any purpose. Arpad invoked the inside of the package, where ball names must be associated with die functions, but each ball name physical definition and connection could be different than in the full system. Walter noted that the specifications should keep the IBIS distinction between pin names and pin numbers. Arpad noted that refdes assignments can change depending on context at the system level. Michael asked whether, if a model makes physical and logical distinctions, a parser reading it should throw errors based on mismatches. Walter replied that connectors are mated by pin number usually, but sometimes by logical connection. Walter noted the Physical description of "A.1" would not have a functional or Logical meaning. Michael and Randy pointed out that this may or may not be true, depending on context. For instance, a standard PCIe connector could be used for a non-PCIe application. Arpad suggested that the pin-specific functions for generic connector aren't known, but the functions for a package of a device will be known. If someone sends an EMD file that includes all this information, that's preferable but separate files may not be easily associated for analysis. He referred to the distinction between what Touchstone does and how IBIS defines connections. Michael noted that the connections between separate components is "one step up" in the hierarchy. In other words, there is usually a separate netlist which makes the connections between individual components; the components themselves do not carry this information. Randy added that the data is useful, but net names will be stated one level up where they will be more meaningful. Arpad replied that he often receives Touchstone files with commented headers defining mapping locations including board or die connections. He took the AR to send out such a file to serve as an example. [AR] Arpad moved to adjourn the meeting; both Walter and Randy seconded. The meeting adjourned without objection.