X86 Tools Group Hillsboro, Oregon Date: July 13, 1993 From: Will Hobbs (503) 696-4369, fax (503) 696-4210 Subject: Minutes from IBIS Open Forum 7/2/93 To: Attendees, list Attendees: AnSoft - Henri Maramis Cadence Design - Sandeep Khanna, Chris Reed, Pawel Chadzynski Contec - Maah Sango HyperLynx - Kellee Crisafulli Integrity Engineering - Wayne Olhoft Intergraph - Ian Dodd Intel Corporation - Will Hobbs, Stephen Peters, Gary Saunders, Doug Boyce Logic Modeling Corp. Siuki Chan, Randy Harr MicroSim - Arthur Wong, Meeling Wei PC Ware Paul Munsey Performance Signal Integrity Eric Bracken Quantic Labs - Mike Ventham Zeelan Technology Hiro Moriyasu List: High Design Technology Michael Smith Intel - Randy Wilhelm, Jerry Budelman, Intel IBIS team IntuSoft - Charles Hymowitz Mentor Graphics Greg Seltzer, Ravender Goyal Meta-Software - Stephen Fenstermaker North Carolina State University Paul Franzon Quad-Design - Jon Powell Thomson-CSF/SCTF Jean Lebrun Next Meeting: July 30, 1993, 9:00 AM to 11:00 PDT. Phone number, (916) 356-9999. Reservation number is 93794 (this meeting only). Note: In future meetings, please identify yourself and your company each time you speak. There are a lot of people attending this forum, and hopefully this procedure will keep confusion to a minimum. Overall Summary: This document contains the minutes of the seventh IBIS Open Forum, held on July 2, 1993. Open Issues: Issues that are raised in Open Forum meetings will be tracked here, along with the owner and time when they'll most likely be addressed. As promised, I have deleted earlier issues that have been closed for more than a month. This list will form the basis of the issues portion of the agenda for each meeting. Other opens, of course, are welcome. # | Description | Owner | On Agenda When? | Done 1 | IBIS 1.0 to stds committee? CFI? IEEE? | Doyle | 5/21, 6/4, 6/17, 7/2 | tabled to 2.0 2 | golden parser progress | Munsey | 7/2 | 7/2 3 | thresholds for timing | Powell | 7/30 | 4 | buffer delay to ramp | Powell | please document | drop 5 | ECL Model Extensions | GD/CM | 7/2, 7/30 | 6 | measurement guidelines | Crisafulli | postpone until more semi vendors on board | 7 | expand ramp definition for open-drain | Powell | 5/21 | 5/21 8 | sim. tech. for model vendors | Ventham | please clarify | 9 | IBIS file name conventions | Hymowi | 7/2, 7/30 | 10 | Phased turn-on/turn-off of multiple transistors | Powell | 7/2, 7/30 | 11 | Expansion of membership, EDA, SCV | Hobbs | 7/2, 7/30 | 12 | Structure of Forum: Treasurer, subgroups? | JP/WH | 7/2, 7/30 | 13 | Golden parser update | Telian | 7/2, 7/30 | 14 | VHDL International BBS | Harr | 7/2 | 15 | Central e-mail alias, mail groups | Harr | 7/2 | 7/2 16 | IBIS Open Forum Summit | Hobbs | 7/2, Fall | 17 | Joint IBIS Press Release | Harr, Hobbs | 7/2, 7/30 | 18 | IBIS V1.1, clarification release | Hobbs | 7/30 | 19 | N/A in columns | | | Agenda: 9:00 Dial-in time, Quorum count 9:10 6/17 Minutes review, Open time for new issues 9:15 Golden Parser Update 9:20 Open Issues from 6/17 minutes 10:45 Wrap-up, action items Minutes: I. 6/17 Minutes review, Open time for new issues: We asked for corrections or comments on the 6/17 IBIS minutes (none). II. IBIS Golden Parser Update I read the update that Don Telian sent me (Don was not present at the meeting.). In addition, Paul Munsey, the parser's design engineer attended to augment the report. The summary appears below: 1. Paul is shipping the parser to all 10 paying companies on July 2. It will be at each company by Tuesday or Wednesday of the week of July 5. (Update: Intel has received its copy; I assume others have also received theirs.) 2. It is an alpha version, for everyone to try out and comment on. Issues/critiques/comments from each company are encouraged, and due to Paul in writing no later than July 22. Comments will be accepted either by mail to Paul,, or by fax to him through Don Telian: (916) 356-6710. 3. During the alpha version review period, Paul will be doing more testing and refining the code further. PLEASE UNDERSTAND THAT THIS IS NOT THE FINAL RELEASE OF THE GOLDEN PARSER. 4. Delivered with the code will be Paul's issue list of about 15 assumptions and clarifications that he made in developing the code. We discussed these assumptions briefly. Chief among these was the large group of characters that cannot be defined as comment characters, including the entire upper and lower case alphabet.. Upon finishing the review of the Parser, the IBIS Specification should be revised to incorporate these assumptions, changes and interpretations. III. New Open, N/A in Typical Column: Kellee Crisafulli of HyperLynx pointed out a potential problem in the 74F244 IBIS file provided with the spec. He proposes we add a comment to the next revision of the spec that says that if there is an NA in a column, such as the typical pull-up value, the entire column should be marked NA. Chris Reid of Cadence agreed that if some measurement techniques do not allow some points to be filled in, then the file should be consistent: either all points should be filled in or none should be. Paul Munsey described how he left the issue in the parser: it is OK to have NA in linear segments of curves, but the parser will return a warning and continue the parsing to completion. The parser will not flag it as an error. A/R All: If you have an opinion on this topic, feed it back to Paul in your evaluation of the Version 0.1 Golden Parser, due 7/22. IV. Structure of IBIS forum, Treasurer, Central e-mail alias, mail groups Randy Harr has volunteered to set up a mail alias for the group. This will allow all members to communicate with the whole group by sending mail to ibis@lmc.com. A/R, All: send mail to Randy, randyh@lmc.com. He will add you to the "IBIS reflector." We later discussed the form of the group, and determined that the subcommittee approach that formed from the ECL discussion (see below) points the way. We still have a question about how to handle the $500 new member fee. (This was a part of the 6/17 discussion; a new member can receive the golden parser source for $500, the money will go into a fund to be used for various IBIS activities, such as advertisements and a small DAC booth.) The group nominated Don Telian to act as treasurer. A/R Arpad Muranyi: Since Don was not present to comment on his nomination, see if Don would be willing to act as treasurer. Randy Harr recommends that we do not incorporate IBIS unless we have at least $10,000. He suggested instead opening a bank account in the IBIS name and doing nothing more formal than that. In the VHDL International Users' Group, they took in up to $100,000 in this informal manner with no problem. Randy offered to help sort out the right form of financial organization. We then moved into the issues listed in the Issues table. V. Thresholds For Timing. Buffer Delay To Ramp, Phased turn-on/turn-off of multiple transistors These items were briefly discussed, though not in this order. The threshold and phased turn-on discussions were tabled to 7/30 due to Jon Powell not attending, and the Buffer Delay to Ramp was dropped. VI. ECL Model Extensions Significant interest exists to expand IBIS to support ECL. Chris Reid would like the voltage range to independently specify the power supply and reference voltages. After some discussion, most felt that the current spec was adequate. If a user is simulating ECL and wants to pulse the input, it must be in the range of -1.9 and +0.8 V. The threshold would be inside this range. There was enough interest in ECL among attendees to form an ECL sub-committee. Interested individuals were to contact each other by e-mail by 7/9. Randy Harr offered to give the ECL group its own alias. For the time being, I will own the e-mail list, which I will forward to Randy when there are changes to it. Randy will then have is system administrator update the alias. The alias will be called: ibis_ecl@lmc.com When the group convenes, I would like a leader to be chosen, both to give reports to the IBIS forum, and to assume ownership of the mail list. The leader can then maintain the list and update Randy when changes are made. Randy would like to initiate the updates to mail aliases only once a week, so if there is significant activity that can't wait, use ibis@lmc.com and thus copy everyone. A/R: The following individuals, and any others with the interest, should attempt to link up via e-mail. I have sent Randy this list, and ibis_ecl should become active by Friday, 7/16. Arpad Muranyi, Intel, arpad_muranyi@ccm.hf.intel.com Stephen Peters, Intel, speters@ichips.intel.com Chris Reid, Cadence, cer@cadence.com Kellee Crisafulli, HyperLynx, 71436.1314@compuserve.com Maah Sango, Contec, e-mail address unknown Hiro Moriyasu, Zeelan, e-mail address unknown Mike Ventham, Quantic, ventham@quantic.mb.ca Zhen Mu, Quantic, mu@quantic.mb.ca I apologize for not being on top of this. These minutes would have been more helpful if issued by 7/9. VII. Measurement Guidelines A brief discussion of measurement guidelines (1 pF probe, short ground lead) quickly produced the information that different vendors will have different equipment, and that at least Intel uses a variety of techniques to obtain data, including simulation (pre-silicon), measurement of test silicon, and measurement of final silicon. Different lab equipment and different testers have different capabilities, which will factor into any guidelines we produce. The recommendation was to shelve this discussion until more semiconductor vendors have joined the forum, then possibly add recommendations to IBIS 2.0. VIII. Expand ramp definition for open-drain We decided to form another subcommittee to address open drain issues. The initial members of this forum will be: Jon Powell, Quad Design, jonp@qdt.com Will Hobbs, Intel, hobbswil@ccm.hf.intel.com Stephen Peters, Intel, speters@ichips.intel.com Kellee Crisafulli, HyperLynx, 71436.1314@compuserve.com A/R, above individuals: initiate discussions via e-mail. IX. Sim. Tech. For Model Vendors No one remembered what this item meant. It will be dropped unless someone re-activates it. X. IBIS File Name Conventions A/R All, Send me (Will Hobbs) proposed file name conventions. I will include it in the next meeting minutes or meeting announcement. Since that time, I have received two submissions, one from Siuki Chan and one from Randy Harr, LMC. they appear below: Siuki Chan: XX??????.ibs, where XX is the manufacturer identification code. ?????? are up to 6 characters for manufacturer specific naming convention. There are only 6 characters and many possible classifications: cmos, bicmos, ttl, logic, interface, communication. Randy Harr's comments were substantially the same, but with a different number of vendor identifier characters: The convention is to use the first few characters of the file name to -identify the IC producer of the information and let them assign the remaining to whatever files they desire to distribute. This is essentially a registration process much like Internet requires for Internet ID's (you register your network main number and then have 255, 1638x, or 4 MEG machine specific names for the subnet). As for the first few character company name, you could use IHS, Cahners CAPS, or the one mentioned over the forum. Another source would be the CAGE code which all manufacturers have (stands for Commercial and Government Entity and is required of anyone who ever sells anything to the government; therefore just about everybody). I believe CAGE codes are 5 characters. The trailing characters can be limited to numeric or allow any characters. The best convention would be for the company to use a linear progression like aaaa, aaab, aaac, etc. (for alphabetic characters used). You can then require that the company publish a list of the files they have supported: intaaaa.ibs - CHMOS IV input pads intaaab.ibs - CHMOS IV output pads intaaac.ibs - 386SX model .... motaaaa.ibs - PowerPC model .... and so on. This list could be searched by anyone to find what they might be looking for. This scheme allows for the manufacturer to decide how they distribute information and yet makes it easier for users to cut down the search at least to a given manufacturers files. Note that you will need to deal with version control and release liabilities for any information distributed by a semiconductor vendor; but I guess I do not need to remind anyone of this. XI. Expansion of membership, EDA, SCV We re-confirmed that we would expand the membership to include more semiconductor manufacturers when the golden parser was released at the end of July. I read a list of nine vendors who are rumored to already be interested. A/R, EDA vendor Applications Engineers. Urge your customers to request IBIS format models from their semiconductor vendors. Note: The July issue of CAD Report reviewed the DAC highlights and discussed IBIS at length. The author suggested that designers request models in IBIS format to help get more models in the works. XII. IBIS Open Forum Summit Another face-to-face gathering, this one to be for a full day, will be appropriate in the Fall. It will be discussed at the August meeting. Intel is willing to host the summit. XIII. Joint IBIS Press Release A/R, All. Give Randy Harr the names of your appropriate Marketing people to discuss a joint press release. Randy will convene a phone conference within a week. Contact Randy at (408) 957-5241. IBIS Open Forum: Minutes of 7/2/93 Meeting, Page 1