In the pic of the "Octal Core AMD" AIDA results, the L2 cache speed is slower than the L3.
Is that also the case with retail samples (in the results that have been posted) or is it an AIDA bug? It seems a bit nuts for L2 to be slower than L3 I would think.
I'm more or less comparing the pic between thuban. But okay.
erocker's post here is may hold the answer. We could say, easily, that the L2 cache of pre-release chips was held back, speed-wise, to give false numbers, and not give any indicator of true performance early.
Erocker's post was more saying, look at the raw speed...then i posted up showing that "speed" (or MHz)isn't always the answer.
So, let's give it a few days, and see if that L2 is still that slow. If it is, then the Lab501 benchmarks are probably 100% accurate. If not, the performance increase will 100% be contained there, IMHO.
After all, I noticed these differences for a reason....because pretty much all I do every day when writing reviews is run benchmarks. I knew where to look.
Matose is defending his numbers. Whether they are indictive of retail performance, I do not know...but it does seem to be that he thinks they are real, for whatever reason. I am not a CPU reviewer, and don't want to be, so it'll be up to others to find out...I'll be looking at performance in relation to board performance...in the future, my reviews will seperate AMD and Intel from each other, so platform compares aren't something I am "professionally" interested in.