Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [HSSG] 40G MAC Rate Discussion


This is a good clarification. Along the same lines, it would be helpful
if the 40GE proponents made a presentation detailing out what their
views are for PHY requirements.

To date, we have had to piece together the 40GE PHY requirements from
various conversations, emails, and presentation snippets. 

Here is what I think is being asked for.


Despite a WDM long haul bullet in Schelto's March presentation, the
assumption is that there is no 40GE SMF PHY requirement.


>100m 4x10G OM3 MMF parallel optics ribbon fiber


This has been mentioned as needed, but the details of what is needed are
a bit of a mystery. Is the copper requirement meant for reach of tens of
meters? We have seen emails suggesting 4x10GBASE-T, which would be
consistent with such reach, but this would be puzzling in view of the
MMF requirement above.

Presumably 40GE MAC will have 10G I/O (in line with Pat's comments,)
which makes sense for ICs for 2008/2009 time frame. Is that mean that
there is a second 4 x XAUI PCS layer that is added between the 10G I/O
PCS and the PHY? Is an inverse 4 x XAUI core planned for the Quad
10GBASE-T PHY that is promised to be available soon, so that the I/O is
at 10G? So now we add another 4W to the 15W or 20W needed for the PHY?

Or is the proposal for copper PHY meant to only cover 5m or shorter
distances? Is it then use a subset of lanes being proposed for the 100GE
copper PHY, for example 4x10G wires, versus 10x10G wires (one of the
copper alternatives,) to piggy back on the 100G effort?

Or is the copper PHY requirement something different?

In addition to assisting in scoping the effort required to write the
standards and do the development, it can enable identifying PMD
alternatives which can be made available even in the absence of a 40G
standard being adopted.


-----Original Message-----
From: Pat Thaler [mailto:pthaler@BROADCOM.COM] 
Sent: Wednesday, April 11, 2007 10:58 AM
Subject: Re: [HSSG] 40G MAC Rate Discussion


Everyone proposing 40 Gig (or at least all those that I have had contact
with) is proposing 4 lanes using the same PCS as most 10x10G proposals:
10GBASE-R - therefore 64b/66b encoded. I expect any exposed serial
electrical interface would also be 10GBASE-R based - again similar to
many of the 100 G proposals but 4 wide instead of 10 wide. 

My preference would also be to use the same multiplexing mechanism for
distributing the traffic to the lanes for both 100 Gig and 40 Gig. There
are at least a couple of proposals out there for multiplexing and I see
no reason why they couldn't be used for both speeds.

Given that the market is data center, I think a parallel optical
interface could be appropriate which would allow using an existing 10
Gig optical PHY as is though I don't think we have gotten to discussions
yet of that vs. wavelength multiplexed.


-----Original Message-----
From: Marcus Duelk [mailto:duelk@ALCATEL-LUCENT.COM] 
Sent: Friday, April 06, 2007 7:18 AM
Subject: Re: [HSSG] 40G MAC Rate Discussion

Hi everyone,

I have a question regarding a possible PCS implementation of 40GbE. To
my knowledge the block code is chosen according to the PMD
configuration, i.e. a 4-lane PMD will most likely use a different PCS
than a 10l-lane PMD or a serial PMD.

For 40GbE we have been arguing that the prime candidate would be 4x10G.
Looking at 10GBase-LX4 I could imagine that the PCS of choice is 8B/10B.
This would, however, result in 12.5 Gb/s per lane. Is that a reasonable
assumption or are there any reasons that 4x10G could use the same PCS as
10x10G (e.g. 64b/66b) ?


Marcus Duelk
Bell Labs / Alcatel-Lucent
Crawford Hill HOH R-237
791 Holmdel-Keyport Road
Holmdel, NJ 07733, USA
fon +1 (732) 888-7086
fax +1 (732) 888-7074