Subject: Minutes: April 21st Future's Meeting From: Syed Huq Date: Tue, 26 Apr 2005 16:58:17 -0700 To: Arpad Muranyi , Bob Ross , Ian Dodd , "John' 'Angulo" , shuq@cisco.com, michael@mirmak.com, Lynne Green , Lance Wang , Randy Wolf , Otto Bennig ====================================================================== IBIS FUTURES/COOKBOOK SUBCOMMITTEE MEETING MINUTES Date: April 21, 2005 Previous Meeting Attendees: ---------- Intel - Arpad Muranyi Mentor - John Angulo, Ian Dodd Teraspeed - Bob Ross Greenstreak Programs - Lynne Green ====================================================================== 1 PM - 3 PM US Pacific Time Telephone Bridge Passcode 916-356-2663 2 474-4324 Agenda: 1 - 1:05 PM Opens - Bob 1:10 - 1:15 Parser License $1,000 put for vote at following meeting. OTHer parser 2500? 1:15 - 2:00 Cookbook - Arpad 2:00 - 2:15 BIRD94.2 2:15 - 2:20 BIRD95.4 2:30 - 2:45 BIRD97 Other. ------------------------------------------------------------------------- Upcoming Futures Meeting: Date: Thursday, May 5, 2005 Bridge: 3 Passcode: 122-6310 ====================================================================== Futures Minutes (abbreviated discussion) Bob Ross ran the since Syed Huq and Michael Mirmak were out. PARSER LICENCE ISSUES Bob Ross summarized that we agreed to the $1,000 fee for ICM parser source code license. We currently have no method to support practical commercial usage. We may wait until Michael returns before creating a license. Ian Dodd indicated that he did not want the process to take more than a few months. A simpler one would be adequate. Bob stated that he would bring it up the subject at the April 22, 2005 teleconference meeting and propose a vote at the following meeting. Bob also suggested raising the regular ibischk4.1 source license price to $2,500 versus $2,084. This makes a nice round number and also rewards those companies who have already paid for the parser at the lower rate. Bob will also bring up the discussion at the next teleconference meeting. COOKBOOK Arpad Muranyi opened the discussion of the latest Cookbook (Ver 0.7b) now uploaded. He did not get any written feedback. However, we had a long discussion on the latest draft. A few major points are documented. The purpose of the discussion was to get people to look at the Cookbook again. John Angulo questioned the time correlation statements on page 11, Section 3.2.2. He noted that the Specification states that not only do the the rising waveforms all have to be time correlated, and the falling waveforms all have to be time correlated, but the rising and falling waveforms have to be time correlated with each other. This is required for differential buffer operation and also for obtaining better results for minimum clock widths. Bob also mentioned that if leading delays are taken out, the same amount needs to be removed from all waveforms. John Angulo commented about the amount of discussion. Some simpler but valid approaches may be lost. John suggested some sort of introductory section to help the reader and perhaps moving some other material to an appendix. However, a lot of the material relates to technology currently being used and is quite valuable. Bob Ross felt that it was not worth a major rewrite. As noted later, John may look at the document further and provide comments. The terms "true" and "full" differential buffers are used. Also IBIS uses "psuedo" and "true" differential. The "half" terminology relates to certain current stearing topologies where one current source exist, but the outputs are terminated by pullup or pulldown resistors of lower impedance. Arpad MUranyi defines "full" or "true" as current stearing having two current sources and a high impedance output. All can be modeled reasonalbly well with "psuedo" buffers - meaning that the buffers are independent of each other. However the extraction needs to be done under differential simulation conditions. Bob likes the distinction psuedo versus true, and that half and full are variations of differential topology. Currently, the document reads psuedo, half and true. Aprad plans write up pre-emphasis and also differential C_comp. Aprad will send John and Bob a Word version to upload/work on directly with comments and for Bob to upload in the futures/cookbook site. The expectation remains to finish the Cookbook when Michael returns. BIRD94 No discussion, still needs some review. BIRD97 No discussion, Arpad still plans an updated BIRD97.1 per comments received. BIRD95.4 Bob summarized the changes to be presented at the IBIS telecomference meeting: (removal of the [*_Equiv] keywords in favor of the existing Series keywords and general connectivity, adding the other power rails to the reference model and emphasizing that the current is flowing into the [Pullup Reference] pin and anything else connected to that pin, and removal of the Input or Termination issue. The simplified BIRD95.4 now requires transforming any buffer specific decoupling impedance into all impedances in parallel across the power rails. =============================================================================