RE: Straw poll
- To: <stds-802-3-hssg-speed@xxxxxxxx>
- Subject: RE: Straw poll
- From: "Vipul Bhatt" <vipul.bhatt@xxxxxxxxxxx>
- Date: Fri, 2 Jul 1999 11:36:52 -0700
- Importance: Normal
- In-Reply-To: <37E04260A7BBD211AF0F0090272A827117B505@xxxxxxxxxxxxxxxxxxx>
- Sender: owner-stds-802-3-hssg-speed@xxxxxxxxxxxxxxxxxx
Title: Straw poll
Walter,
My grudging vote is for 9.58 Gb/s.
As
someone who will be involved in developing the PMD, I am agonizing over this
decision. Please allow me to share my agony.
The
PMD has to carry encoded data. Therefore, I cannot ignore encoding when voting
for a MAC/PLS interface speed. After looking into the crystal ball, my estimate
was that 9.58 Gb/s is likely to be implemented in conjunction with scrambled
encoding while 10 Gb/s is likely to be implemented in conjunction with 8B10B. I
know it does not have to be this way; but that's what my crystal ball told
me.
Therefore, by voting for 10 Gb/s, I thought I would be pushing my destiny
towards 12.5 Gbaud. Implementing a direct laser drive at 12.5 GBaud (including
obtaining those lasers from multiple sources in the market), and controlling
performance of receivers is not just 12.5/9.58 times harder; it is much
harder.
But
scrambled encoding's implications - frame delimiting, for example - for a
cost-effective PHY have not been fully discussed yet. Once you lose 8B10B, you
lose all the K characters, and a chance to leverage the robust 802.3z
implementation. By the time we hack out all the details of scrambled encoding
implementation, the PHY will begin to look a little like a POS device, I was
afraid. And then there is the jitter issue for PMD design - you really
appreciate 8B10B when you remember that it limits the run length to
5.
So the
only way to have my cake and eat it too is to lean towards a "non-serial or
non-binary" PMD solution (like WWDM or MAS). But sorry, I am not yet ready
to go that far. I think the serial solution has a lot of
potential.
In the
end, the immediate benefit of lower bandwidth (for PMD design) tilted my vote in
favor of 9.58 Gb/s.
Thank
you.
Vipul
Vipul Bhatt
Finisar Corporation
274 Ferguson
Drive
Mountain View CA 94043
Phone:(650)691-4000
x113
Fax:(650)691-4010
Email: vipul.bhatt@xxxxxxxxxxx
==============================
Ok group. In one more attempt to see if there is consensus
within the ad hoc as represented on this reflector, please vote for one and
only one of the following two motions. This straw poll will be conducted by
email, so either post your vote to the reflector or send it directly to
me.
Proposed motions:
______________________________________________________________
1. Adopt as an objective:
"Support speed of 9.58464 Gb/s at the
MAC/PLS service interface."
______________________________________________________________
2. Adopt as an objective:
"Support speed of 10.000 Gb/s at the
MAC/PLS service interface."
______________________________________________________________
Walter Thirion
Level One
Communications
- References:
- Straw poll
- From: "Thirion, Walt" <wthirion@level1.com>