Print

Minutes of Weekly Meeting, 2010-01-25

Meeting called to order at 10:35 AM EST

1. Roll Call

Eric Cormack
Ian McIntosh
Patrick Au
Carl Walker
Heiko Ehrenberg
Brad Van Treuren
Brian Erickson
Adam Ley
Michele Portolan
Tim Pender (joined 10:50)

2. Review and approve previous minutes:

01/18/2010 minutes:

3. Review old action items

4. Discussion Topics

  1. Outline Schedule for Q1
    • [Ian] I sent out a draft schedule earlier. I seem to have left the wrong title on it though at Task 1! Should be 2010 Q1.
    • [Ian] One thing I put on here as a reminder was the Newsletter.
    • [Patrick] Should that be the 'Spring' Newsletter?
    • [Ian] I'd thought about that when we went to quarterly issues; the timing is a bit ambiguous. Since it's largely a review of the previous quarter, calling it 'Winter' seemed OK.
    • [Ian] There are a few activities leading up to assessing whether or not we're in a position to submit a PAR. It's maybe not time critical, but I feel that many of our observers are keen to see us move into the formal PAR stage.
    • {Tim joined}
    • [Brad] Task 10 may have dependencies on tasks 4, 5 and 6.
    • [Ian] I expect you're right, and possibly dependencies going the other way too. I wasnt trying to show a detailed project plan with all the links here; I don't think we know them yet. It's more to show a rough time frame for the tasks, so we have some idea where the meetings are going.
    • [Patrick] It's pretty aggressive, but we can try.
    • [Ian] Everybody's time is tight. It's maybe an aspiration, and we've probably slipped already, but we can only do what we can do. It's not a life or death matter.
    • [Ian] It's a guide for our own use. I'm not planning on publishing this.
  2. 2009 Survey - Preliminary review of submissions
    • [Ian] Previously, we'd got as far as Q3.5.
    • Q3.6
    • [Ian] The vast majority agree that standard language is need for System JTAG, but the comments show that people see different uses for a language.
    • [Brad] Yes, I think all different levels are captured in the "Why not" answers. What is encouraging is that people feel that there is a way to do some standardizing.
    • Q3.7/3.8
    • [Ian] We looked at these briefly last week, and were surprised by the number of people claiming to use CAD files to describe the system assembly. It's possible that some people do that, but it doesn't happen in my company.
    • [Brian] Maybe some people are using a combination of CAD files and VHDL?
    • [Brad] Yes, that's possible. Maybe we can get the data out another way so we can see if the people that gave answer h) also chose other options.
    • [Ian] The data is availabile to that. It's different to the extractions I've done up to now, but easy enough. Probably a simple matrix will tell us what we want. {ACTION}
    • [Brad] I'm surprised at the amount of ticks against VHDL. I think in P1687 the view was that VHDL was overkill for instrumentation.
    • [Michele] Well if you include RTL. I think many US users are more used to Verilog and are a bit scared of VHDL.
    • [Ian] I can see why many would claim VHDL is part of the system description. Many boards are really just FPGAs so the whole board function is described by the VHDL for those FPGAs.
    • [Tim] That was why I ticked that option.
    • [Brad] OK, it just seems to be at odds with the 1687 view, and I don't want us to be going in the wrong direction.
    • [Ian] I don't know 1687 well enough to comment, but I could see that VHDL might not map on to the lower level instrumentation level as well as it does at the board or system levels.
    • Q3.9
    • [Ian] Well there's universal desire for reuse.
    • [Brad] That's probably more difficult that many realize.
    • [Ian] We didn't ask about difficulty, just whether it was wanted and it clearly is.
    • [Brad] Yes, and I predict we'll spend most of our time resolving 3.9!
    • Q4.1
    • [Ian] A broad spectrum of previous experience with JTAG in systems is represented.
    • [Brad] It's notable that 50% have been through the pain of using JTAG in systems, so they'll understand many of the issues.
    • Q4.2
    • [Ian] I was surprised that SJTAG was being applied in relatively simple systems. I rather expected it to be confined to the more complex ones.
    • [Brad] But we didn't clarify which use cases are being used, it may be just for updates.
    • [Ian] That's true, but once you have something in place, it could be available for other use cases.
    • [Brad] Yes, the big step is getting the hardware on in the first place.
    • Q4.3
    • [Ian] No-one is expressing any serious doubt about SJTAG being beneficial.
    • Q5.1
    • [Brad] I was surprised at this. I didn't expect there to be as much interest in distributed systems.
    • [Ian] Nor did I. I expected 'desirable' to come up. This maybe goes back to some of the things Harrison was talking about maybe a year ago. I didn't really think it was that important to people, but maybe I need to reconsider.
    • [Brad] Option c) was the high runner I expected too.
    • [Ian] Again, we don't know the use cases.
    • Q5.2
    • [Ian] I'm a bit surprised that so many people are interested in mezzanine to mezzanine test.
    • [Brad] Most modern designs have dynamic switching to route signals to or from daughter boards now. There isn't the hard wired mezzanine-mezzanine connections.
    • Q5.3
    • [Ian] In the end, I'm not sure how different answers a), b) and c) really are.
    • [Brad] Well they'll differ in the control portion of SJTAG; how you manage things. Tha paths through the system are going to differ.
    • [Ian] OK. Most people want to be able to do backplane tests, but aren't so bothered about power and cooling subassemblies.
    • [Ian] Concurrency is something that people are interested in. Concurrency for updates I'd seen argued but that didn't sway me too much as we don't often have multiple identical boards, but I'm now realizing a bigger hit in terms of concurrency for BIST.
    • [Brad] That's especially true for BIST.
    • Q5.4
    • [Ian] Connectivity only, c), get the low response here. People want a lot more standardization than that. It could be difficult to achieve.
    • [Brad] That maybe goes along with 3.7 and 3.8, where HSDL was a low runner.
    • Q5.5
    • [Ian] SJTAG as a redundant test interface wasn't that popular. Single point of access and external control are the high runners, although there's not much between all the answers other than b).
    • [Brad] I'd be interested to know if all the people who answered c) really understood the data management that's involved there.
    • [Ian] I guess some of the language in the survey was difficult. It's likely that some of the questions may have been misinterpreted, or that the implications of some answers may not have been fully thought through.

5. Schedule next meeting

Schedule for February 2010:
Monday February 1, 2010, 10:30 AM EST
Monday February 8, 2010, 10:30 AM EST
Monday February 15, 2010, 10:30 AM EST
Monday February 22, 2010, 10:30 AM EST

6. Any other business

None.

7. Review new action items

8. Adjourn

Eric moved to adjourn at 11:45 AM EST, seconded by Tim.

Respectfully submitted,
Ian McIntosh