RE: Proposal for accomodating 10.0000 and 9.58464 line rates
Brad, Dan:
I agree with Dan. The HOLD solution is the simplest and most general. If the PHY doesn't rate control the MAC then the MAC will need a pacing mechanism.
Paul
At 05:03 PM 8/17/99 -0500, Booth, Brad wrote:
>>>>
RE: Proposal for accomodating 10.0000 and 9.58464 line rates
Hi Dan,
My biggest concerns with the HOLD signal implementation are the following:
* the PHY ends up rate controlling the MAC
*
* the granularity of the HOLD signal (would it be on a byte-by-byte basis or a 32-bit basis)
*
Other than that, I see the basic concept and end result being effectively the same. As for point #1, a valid concern that makes the HOLD signal a bit better (by the way, MAC deferral is half duplex only). J As for point #2, the MAC wouldn't have to do anything for a 10 Gb/s PHY because minimum IPG would be required for that implementation and variable IPG would not exist.
I think that whether we select variable IPG or a HOLD signal implementation, they both want a 10.0 Gb/s MAC/PLS data rate that will permit a 9.58464 Gb/s payload rate.
Thanks,
Brad
Brad Booth
bbooth@xxxxxxxxxx
Level One Communications, Austin Design Center
(512) 407-2135 office
(512) 589-4438 cellular
-----Original Message-----
From: Dan Dove [SMTP:dan_dove@xxxxxx]
Sent: Tuesday, August 17, 1999 4:38 PM
To: stds-802-3-hssg@xxxxxxxx
Subject: Re: Proposal for accomodating 10.0000 and 9.58464 line rates
Hi Brad,
I just returned from vacation and finally got around to reading on the HSSG
reflector. I am concerned about the packet-length-based-IFG concept for two
initial reasons.
1) It would seem to add more complexity to the MAC deferral process
by making it based on a calculation rather than a simple count. While
this might not be too difficult, the coding in the standard might be.
2) For PHYs that run at a full 10.0000, the MAC will either run at a
lower efficiency than necessary, or there will have to be some level
of negotiation with the PHY up front... again more complexity.
I believe that adding a "Hold" signal at the XGMII is simple, allows for
a broad range of PHY implementations, and should eliminate this continued
discussion about what speed to run the MAC/PLS interface. I expected to
come back from vacation only to find total agreement on this point and
everyone looking for something interesting to talk about... so much for
my great expectations. :)
Regards,
Dan Dove
Paul A. Bottorff, Director Switching Architecture
Enterprise Solutions Technology Center
Nortel Networks, Inc.
4401 Great America Parkway
Santa Clara, CA 95052-8185
Tel: 408 495 3365 Fax: 408 495 1299 ESN: 265 3365
email: pbottorf@xxxxxxxxxxxxxxxxxx