| Dallas | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
| ||||||||||||||
| Sunnyvale | |||||||||||||||
|
| ||||||||||||||
| Eindhoven | |||||||||||||||
| |||||||||||||||
| Audio | |||||||||||||||
|
| ||||||||||||||
Rohit - CTL Syntax & semantics can be frozen but they are still open to change. He need 2 months to check completeness of definitions. He needs 2 months to do documents and examples (these examples are stripped down and just to check modes of operation). He also needs an example from CTAG for testing.
(in response to Ron)
BSDL content can be translated to CTL. Timing relationships (edge to edge,
period, etc) can be specified. However completeness with respect to CTAG
written requirements has to be checked.
(in response to Tom)
CTL can handle overlapping protocols. E.g. the example of the Jetta paper.
Yervant responded that CTAG needs to finish its spec first.
Lee indicated that CTAG is deciding if the standard will be the Scan-Pil + SIL. Sequential time estimate is 4 months to finish SIL + 4 months to finish scan Pil + docs
Ron expressed that the standard is taking too long and CTAG is running the danger of being made defunct by an in-term ad-hoc standard. Ctag needs to cut topics and work faster.
Yervant, Lee agree. Grady expressed concern that people are already working hard on a volunteer basis.
Yervant continued to suggest that the perfect standard is hard. Limit the scope but the SIL is not enough. Scan-Pil will make it widely used. So limit work to what can be done in 8 months and stop. Also milestones are needed to track and align progress
Lee pointed out that it is not a technical challenge but committee work tends to require consensus (thus 4 months)
Mike, Ron, Fidel, Yervant agree that there should be cuts.
Grady expressed that a good standard is important even if it takes more time than estimated.
Mike did not want a repeat/redefinition of rules already in the draft. Instead, just reference them. (all agreed)
He wanted clarification of the mandate to the compliancy task force. DATE audience suggested that it should be a reporting structure only.
Fidel left. Jason takes over minutes at 1:20pm CST
Grady is giving PIL proposal presentation.
Grady: Erik Jan objects to Model #1 because not able to execute test the same way regardless of access mechanism.
Model #1 is merely compliant, not our example of "Best Practice". Uses the user-provided coretest instruction.
Mike: questions need to review this at this meeting.
Grady: just looking for feedback from WG
Lee: This presents info relevant to determining schedule as Model #1 shows how parallel access is achieved without a PIL, only a SIL.
Rohit: Have we considered hierarchy.
Lee: Yes, we flatten hierarchy.
Lee: TAM is not part of the standard.
Rohit: Given Model #1, how do you differentiate between compliant and non-compliant cores.
Jason: We have a precise definition of what must be and what might not be wrapped. We only allow test only terminals to be un-wrapped.
Mike: Un-wrapped does not mean un-touched.
Jason: Un-touched does mean un-wrapped.
Rohit: CTL has no problem describing Model #1
Yervant: What bandwidth restrictions are there with Model #1
Lee: Model #1 has a 1 to 1 mapping through the wrapper, but no restriction on bandwidth on either side of the wrapper.
Lee: "Model #1 is an easier PIL to swallow."
Lee: We could stop here, with Model #1 and go to ballot with SIL only. The next phase would be to describe one or more standard PILs.
GG: TAMs have variable bandwidth therefore need bandwidth conversion.
Ron: Bandwidth conversion can be done inside TAM.
MikeR: Wants to have either Model 1 or Model 2, but not both. Just select one and we don't have anything to work on later.
GG: We have different definitions of PILs.
MikeR: We should cut that out of the standard.
GG: Then we would not have the benefit of tool automation. PIL indicates sequentiality. If we take it outside (to TAM) people can not derive benefit from the standard.
Lee: Agrees with Grady. This is the significant difference, no sequential element vs. sequential element in Model 1 vs. Model 2.
MikeR: Have seen PILs without sequential elements.
GG: Then call it some other dang thang (Texan, may not pass IEEE) instead of PIL
MikeR: Call it a SPIL.
Lee: Reduce task of WBR team to Model 1, which is supported by CTL. Then we could wrap this up. ;-)
Lee makes motion:
Grady: concerned that we will not proceed to another release that does include bandwidth conversion.
Lee: we have always stated that we will include connecting to TAMs in our standard.
Vote:
Lee yes
Grady no
Luis yes
Jason yes
Ron yes
Alan yes
Mike yes
Tom left the building
Yervant yes
Bulent left the building
Rohit yes
Douglas abstain
Tony abstain
Fidel left the building
Motion passes.
Yervant: 4 months to the end?
Lee: that seems reasonable.
Lee: we need to better document Model #1 for CTL
Desire for rising edge behavior is valid for efficiency reasons.
GG: Do we have an access problem to these pins.
Alan: People may want to run the device on a low pin count tester.
Lee: Let's take this back to CTAG. I was confused this has to do with boundary scan cells.
Rohit: What characteristics of a core pin force it to go to chip pin?
Ron: It is an integral part of a hard core and cannot be separated.
Jason pointed out that it neglected the Transfer event.
Discussion ensued about the value of the transfer event that ultimately led to:
Lee provided a solution to using Transfer with the 1149.1 TAP controller.
Discussion as to the relevance of this solution commenced with the basic disagreement being if it should be included in the standard as example material. Mike R. being the foremost detractor.
Meeting adjourned