[802.3_B400G] Reconciling the P802.3dj SMF test methodologies

["Mark Nowell (mnowell)" <00000b59be7040a9-dmarc-request@xxxxxxxxxxxxxxxxx>]

Dear Task Force members,

With the release of D2.2, the optical clauses now include definitions for four separate optical transmitter test methodologies.  This is perhaps overkill.

We got here slowly over time after the Task Force understood that the TDECQ method used previously didn’t fully yield results that were predictive of link performance and interop.  Various proposals have been brought forward (and adopted into the draft) to attempt to reconcile this issue:

• Improvements to the TDECQ approach
• Alternative TDECQ approach considering codeword error ratios
• New link-measurement based function symbol error mask
• Jitter testing

These have all been well proposed and adopted and created in response to the perceived weaknesses of the original TDECQ approach. The justification to adopt them into the draft was to clearly document them to allow review and consistent testing/analysis of them.  

But we now have more than I suspect is necessary and sufficient.  We’ve also heard concerns from participants that in aggregate these tests will add too much burden to the ecosystem to support.  Unfortunately, the depth of data/analysis has not been enough for us to validate these tests and understand which one or which combination will result in the ultimate goals of guaranteeing interop, yielding consistent results, and minimizing unnecessary yield impact.  We’re also hearing concerns that teams are resource stretched being able to support all this testing along with their commercial priorities.

So, in a desire to bring this to closure as quickly as possible, I’ve reached out to many of the participants to request they do a short term push to gather test data or do analysis to help validate what is the right subset of tests that we need to hit our goals.  To support this, I will be submitting comments in the D2.2 review cycle against each test methodology to propose it is removed without any validating data or any further supporting evidence/analysis.  We are a data driven organization, and data always matters.  

I wanted to share this across the Task Force now so TF participants can work with their teams and partners to accelerate this work.  Hopefully, we will come out of November meeting with a much-reduced approach which will alleviate the ongoing pressure for teams.

Knowing that some may have concerns about sharing data due to commercial sensitivities, I ask you to reach out to me privately.  We can work to find an approach to anonymize any data.

Regards,

Mark 

P802.3dj Vice-Chair, Optical Track Chair.

---

To unsubscribe from the STDS-802-3-B400G list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-B400G&A=1