WE ALL KNOW that Intel just launched its mainstream Nehalem-based PC CPUs, the "Lynnfield" Core i7 and Core i5 series sitting in the new LGA1156 socket.
Less DRAM channels, two instead of three, for a simpler memory system, a bit less speed and less total power draw - due to the P55 1-chip chipset savings - than their bigger current cousins in the LGA1366 socket, but otherwise it's all about same.
The new chips also improve on the original Turbo Boost brought along with the first Core i7 CPUs by adding more turbo stages, up to 10 steps in total above the stock frequency. Each Turbo Boost step adds 133MHz. For instance, a Core i7-870 with its default 2.93 GHz clock can race up all the way to 3.6 GHz if only one or two cores are active, and 3.2 GHz everyday with all four cores running - if you've got a good motherboard and decent cooling, of course.

Some journos got the meaning a bit skewed, and I couldn't miss registering a very funny point brought up by a fellow fast-talking Asian hack a few hours ago during the Intel press talk. That poor fellow asked something like, in a heavy accent far worse than even my own, "So, this Turbo Boost means you want to boost sales faster in a poor market?"
Then another journo interrupted to ask, "Is this Nehalem?" At that point the friendly regional PR guy facing me almost looked like he wanted to throw himself off the stage in despair. It looks like a year's worth of Intel PR hard work promoting the Nehalem stuff still resulted in unsuccessful hack brain programming aplenty.
How about a basic hardware course for the fresh journos once a year, Intel? And no, I don't mean IDF there. µ
I'm wondering about this boosting, how long is that active, 20 seconds?
Just long enough to skew most benchmarks before it would overheat?
The turbo boost function is tested at Anandtech.com and Anand shows some very large gains from turbo boost. Much better than the socket 1366 brethren can achieve.
In fact it seems that overclocking has now become largely obsolete. The CPU does it all on its own.
The boost is apparently dependent on the type of code being run as well as the number of cores. Under all circumstances you get as much power as Intel can provide you. So why do you want more?
I just tested a Core i7 920 and it goes from 46 degrees at idle to 73 degrees at full speed in a matter of seconds. It's the same with the power consumption. Yet the turbo mode stays on for the duration of the test.
So no, it's not a feature intended for cheating.
The program I use for stress testing can be found at www.withopf.com .
The only problem with turbo mode is with the early CO stepping, where C3 mode was problematic. Thats why most boards default to C1e mode only. Without C3 the cores per definition never sleeps and the faster turbo mode is never invoked.
Looks like the i5-750 is the way to go. It's a bit strange how (for now at least) the Lynnfield i7s are more expensive than the 920, with mainboards and RAM at a similar cost. Oh and it also may be worthwhile buying 6GB triple channel kits and selling one of the 2gb sticks if you want to be an early adopter - works out cheaper than the initial batch of "i5 memory" being sold. Wonder how many people are going to try to pair one of these with the incumbent, higher voltage DDR3 kits... I think the marketing doesn't make it as clear as it should be as yet.
@Tomas
If it stays on whenever needed without overheating then they can claim a lower TDP for the same CPU.
In that case it would still be deceiving people.
Would it be any different if AMD said TDP 65w for 955BE but with a base frequency of 2.5Ghz and boost to 3.2Ghz?
And after that I want to see your face when you buy Intel Celeron CPU for 5000$.
That would make my millennium. :)
Intel Lover, we told you several times, that without AMD, you would still be playing on your 3GHz Intel P4, with the case cover off and blowing air with a huge cooling fan from the outside.
No competition always means slower development...
Go forth, and multiply!