Meeting Minutes - 2015-10-12

Minutes of Weekly Meeting, 2015-10-12

Meeting called to order: 11:06 AM EDT

1. Roll Call

Ian McIntosh
Eric Cormack
Carl Walker
Michele Portolan
Brad Van Treuren
Heiko Ehrenberg
Peter Horwood (joined 11:08)
Tim Pender (joined 11:18)

By Proxy:
---

Excused:
Adam Ley

2. Review and approve previous minutes:

  • Approval of September 28 minutes (draft circulated 9/28/2015):
    • No corrections noted.
    • Eric moved to approve as amended, seconded by Brad. No objections or abstentions.
  • {Peter joined}

3. Review old action items

  • All: do we feel SJTAG is requiring a new test language to obtain the information needed for diagnostics or is STAPL/SVF sufficient? See also Gunnar's presentation, in particular the new information he'd be looking for in a test language (http://files.sjtag.org/Ericsson-Nov2006/STAPL-Ideas.pdf)
  • Ian: Add the previously discussed lists to the 'master' template. Ongoing.
    • Some sections need further expansion that may take time to develop.

4. Reminders

  • Consider Adam's three points (from the action from the first weekly meeting) and suggest what is preventing us from answering those questions:
    • Establish consensus on goals and constraints
    • What are we trying to achieve?
    • What restrictions are we faced with?
  • Forum thread for discussion: http://forums.sjtag.org/viewtopic.php?f=3&t=172
  • Move persistent AOB items to Reminders in future:
    • The Newsletter was due at the end of July.
    • Try to get Al Crouch on call re 1687.1.

5. Discussion Topics

a. Feedback from ITC

  • Heiko reported that ITC seemed "slow" again this year. Attendance was near identical to last year. Heiko talked to a couple of people who indicated interest in perhaps joining one of our conference calls or to get added to the email-reflector; one person was an engineer from Google (don’t have name or email info) and the other was Jason Doege (who was also very active in IEEE 1687, formerly from AMD and now with Centaur). Due to booth personnel shortage Heiko wasn’t able to go to many papers and also missed Michele’s presentation.
  • Brad commented that both Google and Facebook looking into developing their own ASICs and FPGAs to improve server performance, so it was likely that they may both have an interest, possibly Apple too. Heiko observed that Apple were not represented on any of the Standards.
  • Michele found people interested in the SJTAG poster but were put off when they learned it was not under an IEEE PAR. The was lots of talk in the papers about 1687 and Michele was able to point out the added value from SJTAG.
  • Heiko added the ITC seemed to be "very IEEE 1687 heavy", even the panel was all 1687 people, and board and system test remained a small topic.
  • Peter asked if the were any customers asking for 1687 support.  Heiko said he had none but understood some other vendors were reporting it.
  • {Tim joined}
  • In regards to the poster session it was noted that unfortunately the posters where really only available to be viewed for the couple of hours of the poster session. There were two poster sessions this year, our poster was scheduled for the second day (although it was not even listed in the poster directory and we didn’t have a poster number assigned).
  • Michele remarked that talk was mainly of the SIB and the instruments or their usage, and EDA vendors seem uninterested in possible functional access to instruments, although Brad noted that some CAD tools were now giving some access to 1687 in FPGAs.
  • Michele added that although 1687 was a very new standard, people were finding ways to use it that had not been predicted. Ian was surprised as adoption of a new standard usually took a few years.  Michele felt that many had been doing something similar on their own, but wanted to avoid the internal tool chain maintenance and move to 1687.
  • Ian commented that he had originally assumed that there would be a "Standards" block of posters as in past years but that seemed not to be the case, and Bill had fitted us in at the last minute. Heiko and Michele thought that maybe no other standard had offered a poster this year.
  • Heiko pointed out that the our poster did not explain that the 'S' in 'SJTAG' was for 'System'.
  • Michele sensed that people were really interested in the kinds of things in his presentation, and particularly how information was exchanged. When explained, a comment was "This is a lot of software and not a lot of people are doing this kind of software". Michele observed that we should be careful about object orientation and abstraction as these may be difficult concepts to pass on.  It is also necessary to have some "hard proof" to show.
  • Brad agreed that similar things were evident in 1687; it can be difficult to explain the software impact of hardware proposals.
  • Heiko reported on the TTSC meeting: A study group has been approved to work for 6 months on a proposal for a working group project on the subject of extending IEEE 1687 to support package level interfaces other than JTAG; Al Crouch is leading this effort.  Heiko will forward the slides Al showed on this topic.
  • Michele also commented on the 1687.1 discussions, noting that he felt there was a realisation that some functional element might be needed inside the language and people were starting to think about some things at a higher level.
  • Brad remarked that a comparison had shown that TCL was slow, possibly the slowest of interpreted languages and was at the mercy of the OS for scheduling.  Michele thought that some people didn't appreciate the difference between interpreted and compiled languages.
  • Ian noted that TCLs main use was originally in scripting together sets of actions within tool chains where each step may be time consuming itself so execution speed was not of prime importance.  Michele considered it a "glue" language and not really intended as a computational language.
  • Brad noted it was a high level language and could be great if you had libraries in, e.g., 'C', and 1687 realised that in providing formatting to align with different languages.
  • Michele mentioned a chat with Jeff Rearick in which Michele had pointed out that SJTAG was considering a minimal information set rather than a language per se, and thought it might be worth pushing that point in future communications.

 b. Review Michele's proposition of September 24.

  • {Proposal shared}
  • Further to this, Michele had made a forum post following ITC (http://forums.sjtag.org/viewtopic.php?f=13&p=1082#p1079).
  • Brad commented that while he hadn't had a lot of time to look at the post, he thought that the some of the diagrams might be using 'TAP' incorrectly - it's not really the TAP that is being connected to, the main point being that it's the capture-shift-update cycle they're being brought into. It's a protocol conversion.
  • Michele agreed but noted that one of Al's slides had shown a board where I2C was driving the TAP via an unspecified protocol conversion.  Brad mentioned other cases where this had been done by simple 'bit-banging' and was very slow.
  • Brad added that 1687 is really coming in below the TAP Controller.
  • Michele went on to talk through his post and the proposal.
  • Brad commented that there are cases where the Access Link is defined by an Access Link cycle followed by a Data Link Cycle. An example is a scan path linker where there is both an IR scan as well as a DR scan, but the end result is that Access Link is connected.
  • Michele responded on how his model handles that, with Brad noting that this was now beginning to define the software interfaces needed for the different aspects; access side and data side.
  • Brad generally agreed with Michele's ideas, but cautioned that the specific commands mentioned may not prove to be the correct ones.
  • Ian felt that the forum post, once the ideas are developed a little, might forma Green Paper or Newsletter item.
  • Brad asked if we needed to vote on Michele's proposal, but Michele felt it was much too early to treat it as a formal proposal - it was really for discussion.  Brad felt we would need something like that proposal else we run the risk of being incompatible with those standards.
  • Voting on the proposal is deferred to a later date.
  • Michele agrees to copy his proposal from September 24 to the forums where it can be commented on {ACTION}.

c. Preclusion: What gets precluded, what are the locking mechanisms?.

  • {Not discussed due to lack of time}

6. Topic for next meeting

  • Preclusion: What gets precluded, what are the locking mechanisms?

7. Key Takeaway for today's meeting

  • None.

8. Glossary terms from this meeting

  • 'Scheduler' (from Aug 31) is TBD.

9. Schedule next meeting

  • October 19.
  • October schedule:
    • 19, 26.

10. Any other business

  • None.

11. Review new action items

  • Michele - Copy proposal from Sep 24 email to forums.

12. Adjourn

  • Eric moved to adjourn seconded by Peter. Meeting adjourned at 12:01 PM EDT.

Thanks to Heiko for additional notes.

Respectfully submitted,
Ian McIntosh