====================================================================== IBIS INTERCONNECT TASK GROUP MEETING MINUTES http://www.eda.org/ibis/interconnect_wip/ Mailing list: ibis-interconnect@freelists.org Archives at http://www.freelists.org/archive/ibis-interconnect/ ====================================================================== Wednesday, August 21, 2013 8 AM US Pacific Time Agenda: - Attendees - Call for patents - Agenda and Opens - What is the meaning of the Si2 specifications recently released to the IBIS Interconnect community (Brad Brim is willing to present on this)? - Resolving the path of IBIS between EMD and BIRD 125 and related BIRDs, and how IBIS-ISS will interact with all of these - Next Meeting Dates/Agenda ====================================================================== Attendees: ---------- (* denotes present) Agilent - Radek Biernacki Ansoft - Denis Soldo Cadence Design Systems - Terry Jernberg, Brad Brim*, Dennis Nagle I/O Methodology - Lance Wang Intel - Michael Mirmak Mentor Graphics Corp. - John Angulo, Arpad Muranyi* Micron Technology - Justin Butterfield*, Randy Wolff SiSoft - Walter Katz*, Mike LaBonte* Teraspeed Consulting Group - Bob Ross* ======================================================================== No patents were declared. ======================================================================== Opens: None ======================================================================== BIRD 125: Si2 Chip Package Interface Protocol: Brad described the origin of the Si2 CPIP. The Apache Chip Package Protocol (CPP) describes in SPICE comments the connection between chip and package. MCP (Model Connection Protocol) is a Sigrity protocol to accomplish the same thing. It describes physical pin names/locations and electrical nodes. Brad previously could not reveal parts of this due to a non-disclosure agreement for the Apache specification and the less open (relative to IBIS) process of Si2. All three formats are in SPICE comments. The strengths of both MCP and CPP are incorporated in the Si2 CPIP format. CPP pin list is flattened, MCP and CPIP pin lists are hierarchical. If present in all subcircuits from buffers to packages, MCP can be applied to enable an EDA tool (or manual user editing of netlists) to concatenate the models to produce the entire description. This makes it very valuable. Arpad suggested that we need a way to deliver canned models for chips. Brad said IBIS makes assumptions about topology, and maintaining that legacy is hard. Combining on-die, packages and modules, we have something like a system netlist. Walter agreed about the complexity. He said CPIP would be useful for vendors to produce model files that would use IBIS-ISS. He said that while MCP/CPIP could handle complex cases, this approach could solve simpler case. Brad agreed. Arpad asked if IBIS-ISS files could have MCP comment headers in them. Brad said this could work to define the system both using something like EMD or by directly using the SPICE with MCP/CPP/CPIP headers. Bob asked if the flavor of SPICE mattered. Brad said it should be agnostic about that. Arpad asked why no special characters were added to flag the special comments. Brad said Si2 called for BEGIN and END CPIP comments, as did CPP and MCP. Bob noted that there is no connection to Touchstone files. Brad said that a common reference [electrical] node exists for ports but the [physical] pins of the ports are not always defined in that manner by most tools. A mechanism to define ports more precisely as they are defined for the extraction needs to be agreed upon to extend CPP/MCP/CPIP to convert S-parameters. Touchstone was not addressed at this time for that reason. Arpad asked how this might impact the EMD vs. BIRD 125 decision. Brad said it was assumed EMD would be used. Walter suggested that CPIP should not have any direct bearing on our work. Arpad noted that this was an important decision for our group. Brad mentioned that a patent acquisition company filed an exception against CPIP claiming that a set of patents filed by R3 Logic may have been violated. Si2 has pursued legal evaluation of this claim. Walter asked if there are pointers to the patents. Brad said Si2 notes the patents that are cited. Arpad asked if any other information about the case is available. Brad knows of no legal action that has yet been initiated by the patent acquisition related to Si2 specifications. Arpad asked why some information is not public. Brad said the work was initially for Si2 members only; which is how Si2 operates, in contrast to IBIS. AR: Brad forward Si2 document on patents Bob asked what companies are using CPIP. Brad knew of none making a formal announcement. Next steps: Arpad showed an August 15 email from Walter listing survey question to ask vendors and users, sent to the ibis-interconn list. Walter showed a response to this from Randy Wolff. Arpad noted that email server problems prevent these emails from being received by people at Micron. Walter said that people at Micron seem to receive replies from others successfully. Walter described parts of Randy's email. He proposed evaluating each part to determine if BIRD 125 and/or EMD can handle it, to make a decision. Walter clarified that "s2p" means a two port circuit, not necessarily Touchstone. Arpad said BIRD 125 could be updated to handle all needs, but the bigger question is whether we like the style. Walter noted that IBIS keywords can potentially do anything parameter trees can. Once we define functionality using parameter trees but the final implementation is not restricted by that. Arpad said we could create a new keyword, but we should not touch existing keywords. Walter agreed and noted that we must define functionality first. ======================================================================== Next Meeting: Wednesday, August 28, 2013