Minutes of Weekly Meeting, 2011-10-03

Meeting called to order: 11:02 AM EDT

1. Roll Call

Eric Cormack
Richard Foster
Peter Horwood
Ian McIntosh
Patrick Au
Carl Walker
Adam Ley
Heiko Ehrenberg
Brian Erickson (joined 11:05)
Harrison Miles (joined 11:22)

Brad Van Treuren
Tim Pender

2. Review and approve previous minutes:

08/29/2011 minutes:

09/26/2011 minutes:

3. Review old action items

4. Discussion Topics

  1. Report from ITC
    • [Ian] General Impressions: Attendance was a little over 700 and maybe up to 750, which is slightly up on 2010. For the exhibition space, I think the number of available booths was very slightly reduced, but the level of occupancy was higher; it looked busier, but maybe Heiko or Adam can comment on that.
    • [Heiko] I'd say it was a little busier.
    • [Adam] I would say the same. I have no access to any measure like lead count but I feel it was probably better than last year.
    • [Ian] There's probably not too much to comment on the technical program. It had much the same mix as last year. I found it interesting to compare against the proceedings on the ITC website from 2005 where there were around 49 sessions and 10 panels compared to the 17 sessions and 3 panels this year. Only having attended two ITCs, I've no idea what those figures really mean for the size of ITC.
    • [Heiko] The poster session didn't seem to be as busy. It was a little bit wider spread. It was still worthwhile though.
    • [Ian] Yes, it was slow in the first hour, and busier in the second hour. I think that being over lunchtime meant that people were getting lunch out of the way first. It may not have helped that the plan of the posters in the Program had the numbering running the wrong way.
    • [Brian] I think there were two things: There was a lack of beer and the posters were too spread out.
    • [Adam] At the poster I manned on 1149.7, it was initially very slow, as lunch run down, but then I had people there constantly and spent an hour and a half talking nonstop.
    • [Ian] There were a few questions that came along on the SJTAG poster, but often it was someone who'd already formed an opinion based on the poster and just wanted to confirm they'd picked it up correctly.
    • [Ian] A few people expressed enough interest to suggest they'd come to the fringe meeting if they found time, but none did. I think that's a problem for fringe meetings - if you don't have a vested interest, the technical program is likely to look more attractive.
  2. Reestablish basic objectives of the group
    • [Ian] So, moving on to the Fringe Meeting. It was lightly attended, being essentially all SJTAG members in attendance. As you will see from the notes, the discussion mainly centered around trying to define a frame of reference for the core standard. There were two questions I took away from there: The first was from Adam, where he asked who was expected to be the beneficiary of an SJTAG standard, and the second was from Anthony who suggested finding where the 'pain' was.
    • {Harrison joined}
    • [Harrison] What do you mean by the pain?
    • [Ian] Simply what are the things that people find difficult about attempting SJTAG. But that led me thinking about what was in the minds of the people who originally formed this group back in 2005. That was before my involvement and I realized I wasn't exactly sure, and that's why I suggested going back to that initial meeting. Of the original group, only Peter seems to be left!
    • [Adam] That's not strictly true...
    • [Ian] Oh, I know.
    • [Adam] Brad was a contributor in absentia to the meeting in Tallinn, and although the group officially formed at that meeting there had been prediscussions going on for sometime that I was aware of or involved in, though I would not claim to be one of the SJTAG 'founders'.
    • [Harrison] Is there anything critical here?
    • [Ian] I guess it's maybe more of a public perception management issue. There is some sense that SJTAG is a group that is not going to achieve anything, and progress does seem to be slow.
    • [Harrison] OK, but I tend to treat SJTAG like P1687 and differently to 1149.1 and its extensions: There, all the hard work was done in 1149.1, but P1687 is marking a new watershed, and see SJTAG in the same way.
    • [Ian] Yes. It's been commented that maybe IJTAG went into the PAR too soon, and are now uncovering a number of issues to resolve, while we've maybe held off on the PAR so we can understand our issues better first.
    • [Harrison] That's maybe a good way to look at it. I think with P1687 some folks thought it would be a lot easier than it is. SJTAG is at the next watershed, and it's going to need P1687.
    • [Adam] That's all dependant on how we identify our scope. In P1687, although there are details and attendant complexities, their scope was fairly narrowly focussed. On the other hand, we're still exploring our scope, and to achieve an outcome, a standard, we need to talk seriously about narrowing the scope to something that will fit in one volume.
    • [Harrison] I agree, that's why I thought my primitives helped us focus.
    • [Adam] I'm not sure what you mean by primitives there, but a key question is do we want to enable a JTAG application in today's environment, or the future?
    • [Harrison] It has to be the future: The environment is changing too fast. Look at P1687, 1149.7 and what's happening in 1149.1-2011 that is soon to become 1149.1-2012; taking the case there then we have to look to the future.
    • [Adam] I would ask the meeting, whether they see, as equipment manufacturers, or as tool vendors, or as device vendors, that they are undergoing or supporting some form of SJTAG development today?
    • [Heiko] Goepel do quite a bit of system development. For closed systems, where the design is all from the same organisation, it's not too difficult. It gets harder when 3rd party boards are used.
    • [Ian] I agree. We're largely in control of our entire system design so we can achieve a level of SJTAG implementation, but COTS processor boards can be awkward. Basic operations aren't too tricky, simple tests or programming but emulation/debugging using vendor tools is a problem and even some device programming operations can become complicated.
    • [Harrison] I don't think it's only 3rd party boards: Even silicon vendors can be disaggregated, and even some design centers not all designs are from the same region.
    • [Ian] I don't see those issues. I'm not saying they don't exist, but I don't see them in our particular business.
    • [Harrison] No, it's in some segments where there's fastest growth, in their supply chain. Not in places like Department of Defense, but that's not a growth area.
    • [Ian] That leads me to a question I have. I'm tempted to see uptake of SJTAG as mainly in high value, low volume sectors, where designing it in might give a return. Is it viable in low value, high volume product where it might be cheaper and faster to produce dedicated, custom test and diagnostic stations than to use JTAG at all?
    • [Harrison] I think of a matrix of the segments: There's the high volume, high mix segment where maybe 500 designs are produced each year. Then high volume, low mix, where maybe iPod fits in, etc. You have to map into that matrix and see where your uptake will come from.
    • [Harrison] It all gets to the notion of cost, and where work is done. We're now seeing China trying to slow down its economy to allow wages to rise. That will change the equation.
    • [Harrison] I think SJTAG will map into high value high mix as where it fits best, but does that map into something that will get adoption?
    • [Ian] I'm always careful to note that my perception is limited by my own experiences, and I'm keen to get opinions from a wider end user base. We've run a couple of surveys to try to get that wider opinion, although I have to say that sometime the results haven't been as clear as I'd have hoped.
    • [Harrison] iNEMI ran a survey on board BIST. That got a good distribution. Once the Telecom markets came in it underscored why P1687 is going to be important. A lot of people want to know how they are going to test between devices. There were hints towards P1687 and hints towards SJTAG.
    • [Harrison] There is a question of what is a system? For the IC designer the floorplan of the chip may be the system.
    • [Ian] Indeed, and an SOC is a system by definition. I think for most people here we're assuming a system to be a multi-board assembly, but ultimately I don't think it really matters.
    • [Harrison] 3D test is becoming important because SOCs are becoming more important - Frost and Sullivan are now tracking SOCs explicitly.
    • [Ian] Harrison, can you put this matrix you have in mind down on a slide for us?
    • [Harrison] Yes, I can. {ACTION}
    • [Harrison] Look at dot6, how many silicon vendors are doing anything yet?
    • {Adam, Carl and Richard left}
    • [Ian] Some of these references you make to dot6, dot7 and P1687 are all things that driving change into silicon. There's an argument to be made that SJTAG may be more of software and tooling issue and not constrained to changes to devices.
    • [Harrison] Yes and no. SJTAG needs dot7 and IJTAG as primitives. Can you do this with Ken Parker's 'lobotomized board'?
    • [Ian] OK we're losing people off the call now that we've passed 12:00. I guess that's a clue that we need to close.

5. Key Takeaway for today's meeting

6. Schedule next meeting

Next Meeting:
October 10th (11:00 AM EDT, 4:00 PM BST)

Schedule for October 2011:
10th, 17th, 24th, 31st.
Brad will not be able to attend on 17th.
Eric may have difficulties joining.

7. Any other business


8. Review new action items

9. Adjourn

Eric moved to adjourn at 12:07 AM EDT, seconded by Peter.

Respectfully submitted,
Ian McIntosh