Look up BS on the dictionary and youll find AMD...
Cant believe what theyve been doing the last year -.-

amen... they keep bringing out concepts that they themselves could not deliver in time...

I think they have out done themselves this time, the K10 performance estimates were at least calculated from the performance of a real, although slower, cpu. Now their claiming supremacy for something that isn't even spec'd out yet.

This seems superfluous: now that FB-DIMM is already out there, widespread and accepted by the market, why try to develop something intended to do the same thing but which is different and and has an incompatible implementation? Is it really a good idea to bank on the existence and widespread availability of, say, registered ECC DDR3 DIMMs?

On the other hand, very little detail is given in the article. I can see it being something more generic such as a Hypertransport tunnel with a memory controller attached to it, like AMD has always said could be done but which never was in practice. That's reasonable enough, but if so, it's more news for AMD's platform division than it is for us.