I was going to point to a performance preview at Inq via TR, but it seems that Theo cribbed all those figures from the ppt Vincent linked.
Well, bungled is more like it. How he confused the English graph on p.70 that we've all seen before and that clearly states estimated performance of a simulated 2.6GHz Barcelona against the Xeon 5355 with the still simulated and still English(and apparently even more "confidential" than the rest of this "public" document) Barcelona 2.3GHz vs. Opteron 2222 on the following pages, who knows.
More confusing is that I don't see a 2222 in the appendix that starts on p.84, but given that the Barcelona 2.3 vs. 2222 were the only two charts actually labelled with "NDA" it's possible the 2222 system and results weren't meant to be detailed.
Well, bungled is more like it. How he confused the English graph on p.70 that we've all seen before and that clearly states estimated performance of a simulated 2.6GHz Barcelona against the Xeon 5355 with the still simulated and still English(and apparently even more "confidential" than the rest of this "public" document) Barcelona 2.3GHz vs. Opteron 2222 on the following pages, who knows.
More confusing is that I don't see a 2222 in the appendix that starts on p.84, but given that the Barcelona 2.3 vs. 2222 were the only two charts actually labelled with "NDA" it's possible the 2222 system and results weren't meant to be detailed.