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

Re: XAUI and 64b/66b

Dear Rich,

> Rich Taborek writes:
> If XAUI is present, it MUST transport /A/K/R/.  If XAUI/XGXS is
> present at the Local device, 64B/66B would transport /A/K/R/ as would
> the medium in the direction of Remote device.  My point is that there
> is NO requirement to rever se translate from the XGXS adjacent to the
> PCS to insure that the PCS or medium transports only /I/.  There is no
> optional or mandatory XGMII defined between the XGXS and PCS. 

The easy solution here is to make XGMII the "universal gasket".  All
PMDs would be architecturally defined with input and output ports
logically equivalent to XGMII. 

XAUI may use all manner of fancy signalling over the copper trace, or
over WWDM or parallel fiber.  This signalling is necessary and
appropriate for alignment and deskew across the parallel physical
layers.  However, XAUI's output should always be converted to the
minimal set of XGMII idle codes.  (with perhaps with a concession for
busy_idle if such a flow control mechanism is chosen to be supported
across all PMDs). 

64b/66b likewise is only required to support XGMII-signalling.  It never
has any need to support or transport XAUI signalling semantics. 

All PMDs would be defined with pure XGMII interfaces at their
architected boundaries.  They are free to use custom signalling within
their own physical layer, but are never required to support or transport
special codes belonging to some other style of physical layer. 

This is like a strategy for avoiding the need to stock every possible
kind of coaxial cable adapter.  If I buy a complete set of adapters
between every connector style and another single type, say type F, then
if I need an adapter from BNC to PL-259, I can always do it in two steps
by cascading a "BNC to F" with an "F to PL-259". 

In this way, XGMII logical signalling is the lingua franca.  No PMD will
need to receive, or be allowed to deliver any codes that are not
transportable by XGMII. 

Of course in any actual implementation that concatenates PMDs such as a
XAUI to 64b/66b fiber module, there would not actually be any physical
XGMII.  It would collapse out to some efficient internal logical
connection.  The important point is that such a 64b/66b layer would never
be required, or permitted, to transmit any XAUI-specific signalling such
as /A/.

kind regards,
Rick Walker