Personally Intel should have gone with higher uncore clocks even if it meant 5-10w higher TDP it would have made it a better cpu over all but for those who have more cash than common sense or have time sensitive work this is the obvious best option.

Ha, I disagree.
The increase in cores and clock make this a nice upgrade.
Enough to last till next year.
Everyday usage is needed to really notice the difference.
So this review is understandable...but incorrect.

Uncore clock:

How do you calculate the Uncore Frequency of Bloomfield ?
Bloomfields Uncore clock is always 2xDDR3 clock, thus if you use DDR3-1600 you'll get 3.2 GHz Uncore clock.

However, this is already overclocking, because Intel does qualify Bloomfield only with DDR3 up to DDR3-1066, not even DDR3-1333 is allowed.
Proof:

http://ark.intel.com/Product.aspx?id=37153

Now it seems that the multiplier for Gulftown is lowered to 1.66, but maybe this is just an automatic setting. Check the uncore clocks with DDR3-1066, if you want to verify.

How was the new intel Fan coping with the heat/noise?

Those LinX GFlops numbers look awfully low - are you sure the system is "stable"? The 980x in the review gets about 61 Glops at 4.1GHz. That's wayyy off the max theoretical of 4.1 x 4 x 6 = 98 Glops.

In contrast, at 4.2GHz my 980x consistently achieves about 90-92 Gflops using the popular "Intel Burn Test" version of Linpack, and using intel's own batchfile driven Linpack. Reasonably close to the max theoretical of 4.2 x 4 x 6 = 101 Gflops.

I have noticed my system showing unpredictable and low Gflop figures (anywhere between 50-something to 80-something Gflops) when something is unstable (yet stable enough for Linpack to finish 10 or 20 runs of a large problem size) and the CPU voltages are insufficient. I wonder if this is some error correction mechanism in the QPI interface which "thottles" the lanes when it detects errors. With enough voltage the Gflop numbers stabilize.

Would love to hear your comments on this.

Ok, so there's still a problem. Everywhere I look they say that the L3 cache associativity is still 16. The problem with that is that 12MB / (64Bytes * 16) = 12288 which is not a power of 2. The associativity must be either 12 or 24. Does any one know which it is?

Hi everyone.
please could someone tell me what is the difference in latency given in number of clock cycle vs in nanoseconds? is there a relationship between them?

example here L3 latency is 44 clocks and in lavalys 4.8ns

Also, I knew that the L1 cache latency is 0ws (wait state) so why here it is 1.2ns?
thank you very much
Azzed2010