4-8b/10b encoders or 1?
Hi Rich,
I was not sure how 8B/10B encoding is to be done for the XAUI interface.
Please refer to your presentation from the July 00 plenary:
http://www.ieee802.org/3/ae/public/jul00/taborek_2_0700.pdf
Going from the XGMII interface to the XAUI interface, which (if any) of the
following
is correct?
Option 1: The 32 data bits are converted to a byte stream, then fed to the
8B/10B
encoder. The output of the 8B/10B encoder is then spread across the 4 XAUI
lanes, starting with lane 0.
XGMII[31:24], XGMII[23:16], XGMII[15:8], XGMII[7:0] // 32 bit XAUI i/f
\ /
---------------------------------
|
XGMII[31:24] // data muxed into a byte
stream
XGMIII[23:16]
XGMII[15:8]
XGMII[7:0]
|
8B / 10 B encoder
|
__________________
/ \
EOUT[9:0], EOUT[9:0], EOUT[9:0], EOUT[9:0] // 10 bit output of encoder
// striped across 4 lanes
| | | |
XAUI Lane 3 XAUI Lane 2 XAUI Lane 1 XAUI Lane 0
Option 2: The 4 bytes of the XGMII interface are sent to 4 separate 8B/10B
encoders (conceptually), each operating on one of the 4 bytes.
The output of each 8B/10B encoder is then sent to one of the 4 XAUI lanes.
XGMII[32:24] XGMII[23:16] XGMII[15:8] XGMII[7:0] // the
XGMII interface
| | |
|
8B/10B 8B/10B 8B/10B 8B/10B // 4
separate encoders
| | |
|
XAUI Lane 3 XAUI Lane 2 XAUI Lane 1 XAUI Lane 0
I believe Option 2 results in better 0-1 distribution per lane. My
appologies if this has already been clarified before.
Regards,
Satish Sathe
Ample Communications, Inc.
4034 Clipper Court,
Fremont, CA 94538
ssathe@xxxxxxxxxxxxx