I've not the faintest idea why THG would be referring to a FSB when the Hammer doesnt HAVE a FSB to speak of. The FSB is the connection between the microprocessor and the memory controller- the memory controller is on-die with the Hammer'.
The roadmaps given were clearly very old and out-dated.
AXP 2600-2800+ are Barton?
The 2700-2800+ are 266MHz FSB? Not to mention according to those roadmap's the 2700-2800 arent due for release until next year.
I'm sure AMD would love to know this, they seem to be under the impression the 2600-2800+ are Throughbred, and already launched.. or perhaps I should say paper launched, while the 2700-2800 are 333MHz FSB parts.
Another error:
They claim the AMD 8111 southbridge only supports USB 1.1, and is therefore outdated--
AMD seems to be under a different impression, as their official spec sheets states USB 2.0 capable.
AMD Spec Sheet
Frankly a few of those mistakes strike me as rather obvious errors that should have been caught, my opinion of Tom's employees remains relatively low. I sincerely wish he would get back to writing his own articles occasionally.
I found this comment interesting:
manufacturers have found out that an external memory interface with Dual-DDR could provide a solution for more performance than what AMD had internally predicted.
THG seems to be implying that VIA will not utilize the on-die memory controller, which in the case of the ClawHammer is supposed to be a PC1600-PC2700 capable single channel DDR interface.
VIA would seem to be implementing an external DualChannel DDR interface.... perhaps this is what the mobo references when listing 100-200MHz options?
I've little doubt the ClawHammer would appreciate the extra memory bandwidth, but the latency would assuredly be a considerably order of amgnitude higher as well.
The Opteron/SledgeHammer of course already features the DualChannel DDR interface, but this is supposdly a ClawHammer mobo.