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

Re: [10GBT] Functional block diagram



Title: RE: [10GBT] Functional block diagram
'link_control' is a primitive from autoneg and 'link_status' is a primitive from the link monitor and is provided to both autoneg and the MDIO.
As Pat said, they are not part of the XGMII.
I say leave them in the diagram.
 
Brett
-----Original Message-----
From: stds-802-3-10gbt@IEEE.ORG [mailto:stds-802-3-10gbt@IEEE.ORG]On Behalf Of pat_thaler@AGILENT.COM
Sent: Friday, July 09, 2004 4:46 PM
To: STDS-802-3-10GBT@listserv.ieee.org
Subject: Re: [10GBT] Functional block diagram

Tim,
 
I don't particularly care whether the diagram is horizontal or vertical, but the signals in the top right corner don't go through the XGMII. They are the MDIO interface signals (clause 45) which is already a separately defined interface in 802.3 for managing the 10Gig physical layers. In 10/100/1Gig, these signals are technically part of the MII, but they have there own 2 pin (clock and data) interface. They don't share signal lines with the data path.
 
It wouldn't be correct to make them part of the XGMII.
 
Pat
-----Original Message-----
From: stds-802-3-10gbt@IEEE.ORG [mailto:stds-802-3-10gbt@IEEE.ORG]On Behalf Of Warland, Tim
Sent: Friday, July 09, 2004 9:42 AM
To: STDS-802-3-10GBT@listserv.ieee.org
Subject: Re: [10GBT] Functional block diagram

 Sanjay,

Nice work on the block diagram, it really helps to clarify the functionality for 10GBase-T. A couple comments: I would prefer to see a simplified version of the drawing which is transposed to vertical hierarchy as you have shown in figure 48-2. This maintains consistency within the 802 document. Secondly, the signals on the top right corner should probably go to the XGMII layer through some form of encoding. Here again, you manage consistency of information flow.

Keep up the good work!

Tim Warland, P.Eng,
Independant

(twarland@ieee.org)

-----Original Message-----
From: Sanjay Kasturia
To: STDS-802-3-10GBT@listserv.ieee.org
Sent: 7/9/04 2:16 AM
Subject: Re: [10GBT] Functional block diagram

Brett,
 
See attached. I made the changes you suggested to reflect that the left
side interface is now XGMII and not GMII. I also removed the Carrier
Sense function since 10GBASE-T will not operate in half duplex mode.
 
I pulled the Transmit Enable function into the PCS Transmit block since
the removal of the TX_EN signal left all inputs to this coming from the
PCS transmit block so it can be detailed out in the description of the
PCS Transmit function.
 
In the attached file I have also included the clause 48 and clause 49
figures so others can easily take a look at them. In one sense it
doesn't matter what you start from. In the end, the functional diagram
for 10GBASE-T has to stand on its own merits. I find that the the clause
40 figure provides more detail on the MDI interface and many of the
functions in the PMA and MDI will operate logically as they did in
1000BASE-T and the structure of the interface primitives.
 
We will probably have to borrow some of the Synch, Frame and Deskew
functions from .3ae from the clauses you suggested to describe operation
of the XGMII to PCS operation.
 
Regards,
 
Sanjay Kasturia
Editor-in-chief, 802.3an
 
sanjay@teranetics.com <mailto:sanjay@teranetics.com>
cell (650) 704-7686
office (408) 653-2235
 

  _____ 

From: stds-802-3-10gbt@IEEE.ORG [mailto:stds-802-3-10gbt@IEEE.ORG] On
Behalf Of Brett McClellan
Sent: Tuesday, July 06, 2004 9:34 AM
To: STDS-802-3-10GBT@listserv.ieee.org
Subject: Re: [10GBT] Functional block diagram


Sanjay,
 
I'm wondering whether the 1000BASE-T diagram is the right starting
point. Perhaps We should start with the clause 48 or 49 diagram and
build on that.
Otherwise, your diagram need the following revisions:
 
- XGMII interface should be: TX_CLK, TXD<31:0>, TXC<3:0> and RX_CLK,
RXD<31:0>,RXC<3:0>
- remove: COL, TX_EN, TX_ER, RX_DV, RX_ER, CRS (actually the entire
carrier sense block should be removed)
 
Brett
 
 

 <<functional2.rtf>>