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

Re: bit ordering on XSBI vs SFI-4



In a message dated 3/23/01 5:03:02 PM Pacific Standard Time,
Tom_Alexander@xxxxxxxxxxxxxx writes:



Justin,

Instead of changing the order of transmission for 10G Ethernet (and
affecting

three clauses), isn’t it preferable to place a note in the XSBI portion that

states that while the bit numbering assigned to the XSBI may differ from
that of implementations, the first bit on the fiber must always be the one
labeled in the standard as XSBI bit 0?

Cheers,

- Tom

P.S. I think a problem here is that there is a notion of “MSB” and “LSB”
with respect to the OIF SFI-4 interface (the OIF spec says that bit 15 is
the “MSB”). I believe this is silly, and it’s also what causes all these
assumptions about requiring inversion of connections on the board and
so on. I don’t know why a SERDES device should associate arithmetic
significance with its inputs and outputs. Perhaps we should get OIF to
change!




Hello Tom,

You and I discussed this a couple of times and I do agree that it is up to the
user to be careful AND that it is a logical mindset issue. However I am
sensitive to
the fact that people have informed me that it would be helpful to reduce the
risk of people messing things up by this reordering. I am getting the inputs
from practical users (modules makers) that are concerned their customers
would get it wrong. Say for example that there are two modules in front of a
person ... one is a older SFI-4 module and another is the "newer" XSBI
module. If one is not careful about the knowing the difference between the
transmission bit ordering, I can see that they will just assume that a
particular bit (15 or 0) will be transmitted first which would be wrong.

I must point out, like you did, that no matter what lables one puts on the
pins it is NOT a show stopper to use either a XSBI or SFI module in their
application.
It does leave room for a possible confusion. The cost in making this change
is more work on three different clauses at his point. I must admit, I saw this
issue early on in the standards effort and I think I probably can't win what
ever
order I put it in. :)  ... BUT I will offer a chance for this public forum to
voice their
positions.

Justin Chang
Quake Technologies, Inc.
2880 Zanker Road, Suite 104
San Jose, CA 95134
Tel: 408-922-6888 x108
Fax: 408-922-6827
email: justin@xxxxxxxxxxxxx
internet: www.quaketech.com