====================================================================== IBIS FUTURES/COOKBOOK SUBCOMMITTEE MEETING MINUTES Date: July 14, 2005 Attendees: ---------- Cisco - Syed Huq Intel - Michael Mirmak, Arpad Muranyi Mentor Graphics - John Angulo Micron Technology - Randy Wolff Teraspeed - Bob Ross ====================================================================== Next Meeting: Thursday, August 4 9 AM - 11 AM US Pacific Time Telephone Bridge Passcode 916-356-2663 4 446-8920 Agenda: 1 - 1:05 PM Opens 1:05 - 2:20 PM Futures Discussion - BIRD100 Draft Review and Key Questions - Pending BIRD Review 2:20 - 3:00 PM Cookbook Discussion - Review of new clamp language ====================================================================== Futures Minutes Agenda ---------- - BIRD95 - BIRD 97/98 - Cookbook - ICM Parser/ - ICM vote tomorrow - At the end, summit updates. BIRD95 ---------- Syed Huq distributed a call for comments at end of June; no comments have yet been received. Bob Ross stated that Cadence may have remaining concerns, but no major objections have been noted to the BIRD. Arpad Muranyi raised the issue that BIRD95 is "old IBIS" rather than "AMS+IBIS" -- the BIRD continues to use the existing infrastructure rather than enabling the future use of AMS. Michael Mirmak responded that the algorithm and general modeling approach needs to be understood regardless of the specific syntax used. Michael asked whether the BIRD, if approved, should be associated with a particular version of IBIS. No answer was agreed. Arpad inquired whether the proposed Cadence macromodeling approach could be useful for BIRD95-type modeling. BIRD 97/98 ---------- On BIRD97 and BIRD98, Bob noted that Antonio Girardi's documented Vgs approach uses transistor-level data. Behavioral modeling at a highly abstract level is not used. Arpad agreed to write to Katya Koller and also to Antonio regarding providing an additional table for buffer scaling in linear (non-saturation) regions. Bob expressed skepticism about the entire approach -- do the tables and equations actually work to accurately express transistor behavior? BIRD94 ---------- Bob provided comment on the meaning of Vdiff within IBIS, in response to Arpad's inquiries. He agreed that the meaning of Vdiff needs to be written up into the specification. Bob observed that Vdiff is traditionally where inverting and non- inverting inputs cross, beginning flight time computation. Arpad asked, if Vdiff is NA, what is its actual value to tools? The specification implies BOTH 0 mV AND 200 mV, but why? Bob noted that the meaning changes with the Model_type context. Bob stated that the problem in "fixing" Vdiff is its integration with [Model Selector]: what happens when we switch in inputs for outputs through the selector mechanism? Arpad agreed to rewrite the BIRD, to include explanation and a possible [Model Selector] fix. BIRD99 ---------- Bob noted that no comments had been received to Arpad's message; perhaps the wider IBIS community doesn't care about the issue. Arpad noted that he hasn't heard from Ian Dodd privately or publicly. The team agreed to schedule a call for vote at next meeting, but Bob asked that the SPICE reference in the BIRD be removed. Cookbook --------- Michael called for comments on the current revision of the Cookbook, and expressed hope that the Cookbook could be introduced and approved quicky. He noted that the language of differential buffers -- "true, half and pseudo" -- had been changed to "fully, half and pseudo." Bob stated that "pseudo-differential" from an IBIS perspective can apply to designs which are fully differential. Similarly, "pseudo-differential" designs can be modeled in IBIS using the same methodology as can apply to a half-differential design. Michael agreed to clarify notions of true, half and pseudo-differential buffers in terms of design versus IBIS language in a revision to the Cookbook.