Minutes, IBIS Quality Committee

17 February 2015

11:00-12:00 EST (08:00-09:00 PST)

ROLL CALL

Ericsson:                           Anders Ekholm
Intel:                              Michael Mirmak
IO Methodology                    * Lance Wang
Signal Integrity Software         * Mike LaBonte
Teraspeed Labs:                   * Bob Ross

Everyone in attendance marked by *

NOTE: "AR" = Action Required.

-----------------------MINUTES ---------------------------
Mike LaBonte conducted the meeting.

Call for opens and IBIS related patent disclosures:

- None

Opens:

- Mike: Activity in this group is slowing down, should we meet every other week?
  - The alternative would be to have short meetings weekly.
- Bob: That would be better.
- We agreed to continue meeting weekly.

ARs:

- Mike recover percent format specifications in spreadsheet.
  - Some new progress, but almost done.

- Mike add comments to CMPNT, DLY, MSPEC sections.
  - No new progress.

- Lance add comments to EBD section.
  - No new progress.

- Bob add comments to CIRCUIT section.
  - No new progress.

IBISCHK6 User Guide:

- Mike: I could produce text files for each message that has no comment.
  - These would include some lines of source code before and after the message.
- We agreed to try this.

Golden waveforms:

- Lance: Maxim is creating a model quality group.
  - They would like to put golden waveforms in IBIS files.
  - Anders wanted this too.
  - Maxim first wants more sophisticated test loads.
- Mike: We could invite them to speak to us.
- Lance: They are busy validating older models now.

- Bob showed the IBIS spec.
- Mike: Anders and I had proposed in 2010 a BIRD draft that was never submitted.
- Bob: Sophisticated test loads will be testing mostly the tools, not models.
- Lance: Maxim would like to test correlation.
  - It would be easier if the load and data was inside the model.
  - Customers would be able to validate how well it is working.
- Bob: This is part of the model today.
- Lance: We have ISS now, we can use that for the test load.
  - This should not complicate the IBIS file much.
  - There might be some restrictions on allowable elements.

- Bob: This would create more work for model makers.
- Lance: It would not be mandatory.
- Bob: Some companies have produced quality reports.
- Lance: A reliable source of quality data would be a plus.
  - They are doing the work anyway, but it does not end up in the IBIS model.

- Bob: Would it be measurement or simulation based?
- Lance: Whatever they are comfortable with.
- Mike: Would EDA tools have to implement anything?
- Bob: Some small company implemented this, but they folded.
- Lance: The Cadence Model Integrity product had this, but that is discontinued.
- Mike: There might be about 10 EDA tools that each model should be checked against.
  - Users of those tools would be doing the same tests in parallel.
- Lance: With real test loads there can be bad mismatches.
- Bob: This tests the simulator, not the model.
- Mike: It tests the combination of the two.

- Bob showed the [Test Load] template circuit diagram.
- Lance: This circuit diagram has flaws.
  - For example Rp1_near and Rp1_far must use the same V_term1.
- Bob: Simulators might handle this circuit differently.
  - You want simple circuits to validate models.
- Mike: This gives model makers a way to avoid testing with multiple EDA tools.
  - Their customers do it instead.

- Bob showed the 2010 BIRD proposal from Anders and Mike.
- Mike: We never made a list of the requirements for this.
  - Sometimes customers ask the vendor to provide silicon simulation data for a channel.
  - It would be good to hear what Maxim expects.
- Bob: Is this for classic IBIS buffers only or also for IBIS-AMI?
- Mike: It should not matter.
- Lance: This might help to enhance the dialog between vendors and users.
- Bob: No tools implement this though.
  - Tim Coyle had it in his Sharkfin product, that was the small company.
- Lance: We need models first.

Meeting ended: 12:13

Next meeting February 24