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

Re: XAUI, XGMII extender?


Your comments address a specific 10 GbE application, that of a single port NIC.
The NIC will have to contain one copy of all required elements of the stack from
the MAC to the MDI. You'll have to find a way to connect the transceiver module
(which itself may not fit on a PCI-X) card to the MAC. I envision prototype and
early transceiver modules as being of the "daughter card" variety. For a NIC,
this may mean that a large part of PCI-X card itself becomes the transceiver
module. I don't want to burst your bubble, but to get 10 Gbps of data through
the PMD, you'll either be running 3.125 Gbaud traces for WWDM or 10.3125 Gbaud
traces for a Serial PMD. These are the leading PMD proposals at this time. 

You can choose to go with a MAC which connects to PCS chip via an XGMII
interface or other proprietary interface, or you can optionally use XAUI for
this interface. XAUI/XGXS is not being forced on anyone. However, a lot of folks
seem to like it.

As time marches on, I'd be very happy to help you out with an integrated MAC/PHY
which connects directly to smaller form factor Serial or WWDM transceivers, the
interconnect being XAUI. All of this should fit easily on a PCI-X card, consume
the least power, and have good EMI characteristics. For production 10 GbE NIC's,
I'd put the XAUI interface up against any other proposal.

Best Regards,

Roger Ronald wrote:
> ----- Original Message -----
> From: Rich Taborek <rtaborek@xxxxxxxxxxx>
> > Walt,
> >
> > I have never viewed XAUI/XGXS as REQUIRING the XGMII. XGMII has always been
> > proposed as an optional interface as was XAUI/XGXS. Therefore, XAUI/XGXS is
> > proposed as an additional interface. It was then located in the stack between
> > the XGMII and PCS.
> >
> > I don't believe that there were ever any presentations made to the HSSG nor any
> > reflector discussion requiring an XGMII for XAUI. The only folks I ever hear
> > suggesting such a requirement are those that dislike XAUI/XGXS for one reason or
> > another, mostly because it's not SONET.
> <snip>
> I can assure you that my dislike for XAUI has nothing to do with its lack of
> SONET purity. It has everything to do with XAUI's added complexity and speed.
> For companies that are interested in producing adapters and don't want to
> deal with 3 GHz signals, XAUI isn't much better than dealing with whatever
> 10 Gbit signaling is used on the other side of the chip.
> We would like to buy 10 Gbit Ethernet parts to get to a reasonable
> clock rate so that we can prototype and build adapters. If we have to do a 3 Ghz
> interface, our fab choices are severely limited and our protoyping choices are
> non-existent. At the XGMII rate, we can build with just about any fab and
> we can prototype in FPGAs (we run FPGAs almost that fast now).
> I can't see much virtue in a high speed interchip connection for my application.
> I'll be much closer than 3", simply because I don't have extra room on
> a PCI-X adapter board. I suspose I could use a an extra part to convert
> from XAUI (if someone makes one), but that seems awfully unattractive.
> RR
Richard Taborek Sr.                 Phone: 408-845-6102       
Chief Technology Officer             Cell: 408-832-3957
nSerial Corporation                   Fax: 408-845-6114
2500-5 Augustine Dr.        mailto:rtaborek@xxxxxxxxxxx
Santa Clara, CA 95054