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

RE: PHYs with 10.000 and 9.584640

Title: RE: PHYs with 10.000 and 9.584640


The insistence on a 10 Gb/s MAC is the desire for a MAC that is an order of magnitude faster than its predecessor.  That is a target the study group has decided to work towards.  It is better to get close and have a 9.58464 Gb/s MAC?  If you were designing only for the WAN, then the answer probably would be "yes."  But 802.3 is not a WAN protocol, it is a LAN protocol.  Because it is a LAN protocol, the goal of 10 Gb/s is an architectural and a precedence decision.

Speed control at the PHY could have a varying impact on the cost of the MAC.  In discussion with my office mate (our MAC designer), the cost of this is so insignificant.  The solution doesn't need to be complex, and this relative complexity is better incorporated in the chipset than in an architectural or a proprietary solution.

I think you might be confusing the data rate at which the MAC is designed to handle and the data transfer rate on the line as being one in the same.  In the past with LAN solutions, this has been true, but that doesn't mean it has to be true if we want to enter the WAN market (for which there seems to be an overwhelming desire).  There are means to lengthen the IPG between frames and for a WAN PHY to strip this IPG to obtain the data transfer rate that it requires.  This makes the architecture of tradition LANs easier to maintain and to sell, while also permitting 802.3 to enter the WAN space at an overall lower system cost.

Clear as mud?