| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
|
Here is a link to my slides, where I show how we can optimize for 2 cable lengths – “short” cables (3m), and “long” cables (5m). http://www.ieee802.org/3/25GSG/public/adhoc/architecture/salunke_082814_25GE_adhoc.pdf I will be voting for Option 3 (separate objectives for each length). --vineet From: Mark Nowell (mnowell)
Dear 25Gb/s Study Group members, In today’s ad-hoc call we had a great discussion on language options for picking an objective for the twin-ax cable application. I captured the options in the
discussion live on the call and they are posted here: We have a number of options available to us and I took the action to set up a straw poll to gather input. I’ve established a SurveyMonkey to use for the straw
poll. The Poll will close on Tuesday morning before the adhoc call. The link for the survey is here: Survey question is asked twice. Once as Chicago Rules and once as a regular straw poll (i.e. Pick one). Background information ==================== Without the benefit of the discussion, the straw poll may need some explanation. The tension between the options revolves around the balance of reach and budget
allocation that people want to be able to explore in task force. When thinking through this, please remember, that the
key decision we need to make is to find language for the objective that will give everyone confidence that when they get into Task Force they will be able to do the work they need to do and drive towards the outcomes they want. Below I will give the options as they are set up in the Survey Monkey, I’ll also add my
personal commentary on what choosing this option mean. This commentary is intended to capture the intent of the discussion and the interpretation I shared on the call but it is obviously not the definitive view. Straw poll options for Twin-Ax Objective wording: 1. Define a single lane 25 Gb/s PHY for operation over links consistent with copper twin axial cables, with lengths up to at least 3m Commentary on #1: This is felt to be the simplest objective. Allows the TF to work on a 3m specification and it is felt that it also enable the TF to also include a mode of operation to
support 5m. Concern raised that this would not guarantee that a 5m specification would happen and budget allocation could cause interoperability issues for 4x25G breakout applications from a 100G CR4 port. Concern that 3m alone would impact Broad Market
potential. 2. Define a single lane 25 Gb/s PHY for operation over links consistent with copper twin axial cables, with lengths up to at least 5m Commentary on #2: Variant on #1, but explicitly calling out the 5M reach. It was felt the TF would be able to still define a mode of operation to support 3m. Similar concerns to above around
whether that would occur in TF. 3. Create two objectives (from options above) Define a single lane 25 Gb/s PHY for operation over links consistent with copper twin axial cables, with lengths up to at least 3m And Define a single lane 25 Gb/s PHY for operation over links consistent with copper twin axial cables, with lengths up to at least 5m Commentary on #3: This approach would explicitly instruct the TF that it needed to work on specifications for a 3m reach and for a 5m reach. As it is framed this way as two separate objectives,
this does not preclude the TF from developing two different PHYs, one for each reach. However, I do not think that there is any industry interest in that happening so believe the TF will work to keep the PHYs converged. But two objectives could allow that
divergence to happen. 4. Define a single lane 25 Gb/s PHY for operation over links consistent with copper twin axial cables, with lengths up to at least 3m and up to at least 5m Commentary on #4: To address the concern about more than one PHY in the option above, re-word the objective to define a single PHY that supports both reaches. This option would explicitly prevent PHY divergence, but runs the risk of being
a very odd objective having two reaches in it. Working Group feedback on that would be unknown. The comment on the call was it could be sent back to tell the SG to make a decision. 5. Define a single lane 25 Gb/s PHY for operation over links consistent with copper twin axial cables, that re-uses the host silicon transmitter and receiver characteristics specified in IEEE Std 802.3bj-2014 Clause 93 Commentary on #5: a different approach to addressing the same problem. Rather than define reach, define that the TF will re-use the industry specified transmitter and receiver specs as outlined in Clause 93. This provides both FEC and
no-FEC variants, and defines an end-to-end channel with no partitioning. The downside of this approach is that no reach is indicated which isn’t usual and not very informative to the non-expert. TF could theoretically meet the objective with a 1m cable,
but I would not expect that, in reality, the TF would move forward without solving industry needs (of 3m and 5 m). Final comment on all of the above: I believe that everyone participating in these discussions and previous ad-hocs are fully aware of the scope of the work that will be addressed in the TaskForce. I do not think that any of these
options will really preclude that work happening but everyone has their own interpretations. Challenge for the SG is picking one with 75% support. Regards…Mark |