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

RE: Gearbox reality check

Eric, Brad,

I am not able to discuss about OIF/SUPI interface.
What I have understand is that there is a technical limit for the
electrical GMii MAC interface to produce a 1 channel in 10 Gbps.
This because of decisions related to compactness silicon versus reliability
of operations.

Do you agree that it is decided that GMII MAC will be 4 channels of 2.5
Ben beeftink
product manager Fiberoptics

Kannegieter Electronica bv
the Netherlands

-----Oorspronkelijk bericht-----
Van: Erik Trounce [mailto:erikt@xxxxxxxxxxxxxxxxxx]
Verzonden: Wednesday, June 21, 2000 3:57 PM
Aan: Booth, Bradley
Onderwerp: RE: Gearbox reality check


Ben, does this question sound familiar?

My view of this so far is that without the WIS (i.e. UniPHY LAN
mode), data coming out on the OIF/SUPI interface is simply the PCS
output, albeit at a slightly higher rate.  For example, assuming
a 16-bit OIF network interface and an XGMII MAC to PCS interface,
the clock rate on the OIF is 644.53125 MHz, and the data is 66-bit PCS
codewords, split up 16-bits at a time.

Now, if you add the XAUI interface between the MAC and PCS
which includes the 8b/10b encoding, I think you need to go even
faster on the OIF interface (805.6641 MHz!), to maintain 10 Gb/s
from the MAC.  This can't be right.  Are 8b/10b encoding and 64b/66b
encoding mutually exclusive?  Or is the 8b/10b encoding only used
across the XAUI interface with the two XGXS interface layers
encoding/decoding the RS data (MAC sends ethernet, RS encodes as
XGMII, XGXS does 8b/10b encoding, sent across XAUI, XGXS 8b/10b
decodes, PCS receives XGMII data, performs 64b/66b encoding, etc).

Can anyone clarify?

Erik Trounce

In message "Gearbox reality check", Booth, Bradley writes:
>Gold star!
>Now, can you explain how it works without the WIS? :-)