Very Good - 73 Watts for Four Cores
To determine the maximum power consumption of our review sample, we stress all four cores using Prime95. Once again, we were more than surprised by the result. Compared to its direct quad-core predecessor with the Kentsfield core (Conroe), the new Penryn-based CPU with its High-K Metal Gate technology draws 39.25% less power. Mind you, we measured the power consumption of both processors in the same motherboard, namely the Gigabyte GA-X38-DQ6.
Although the new Penryn now sports an additional 4MB of L2 Cache, its power consumption of only just under 73 Watts undercuts that of its predecessor by roughly 47 Watts.
Even though the number of transistors has increased from 586 million to 820 million, power consumption has dropped by 39%.
Things currently look very bleak for AMD. Its flagship processor, which only features two cores, draws 132.76 Watts. Intel’s current top model, a quad-core processor, only needs 72.98 Watts. Even the AMD platform’s lower northbridge power consumption can’t help close the 50 Watt gap.
Looking at these results, an official 4 GHz variant is entirely feasible for next year. A new stepping could allow for a lower core voltage and would thus be easily possible. We’re not sure why Intel specifies a TDP of 130 W for the QX9650. In our measurements, the processor and voltage regulation modules together drew much less power and would have easily fit into a thermal envelope of TDP 95 W.
Just a small typo on the first line on thelast page... i'm sure the word that is meant to be there is "reticent". Other than that an excellent, informative and thorough review, thank you
wow... poor amd
the green bars are getting smaller for AMD
i still going to get an AMD64 5000+ Black edition as my server needs an new cpu, mobo and ram
i do not want to replace my motherboard just yet as it performs Fully stable for me and i not looked into stable/good OC boards for Intel yet
but darn that CPU has some power behind it lol
Yup.... AMD seems pwned for the time being
Bad for them and surely bad for us 
I hope their quadcore (on a single chip) solution is a success...
Are THG listening to the readers? This is a lot better than articles of recent times!! Now what about the follow up article... 'Penryn Hitting 5Ghz on ambient watercooling!!'.
Thanks guys!!
Bob
PS Like the detail on the new SSE4.1 instructions + history of SSE instructions - nice touch!!
I agree with Bob- It's nice to get a decent, in depth technical article, the likes of which got me reading THG in the first place!
I like the picture at the bottom of page 9. Nice toastie air blowing right into the arse end of that video card, just as well its not a scorchy pentium 4
)
Yes indeed it will be nice to see how well it performs with a watercooling system, who knows even liquid nitrogen will be nice.. If they reach with Intel Pentium 4 more than 5GHz I wonder how much GHz might give a 45nm processor heheh
P.S. With liquid nitrogen I'd like to see how well it performs TOO!! [of course a more simple solutions like a watercooling overcloking still is a good thing, especially if it's in a video to see exactly what is going on]
Sounds like a 805 ... only more expensive, faster and twice the cores
I thought I left some feedback a few days ago.. oh well I was just saying why not a review of an overclocked intel processor reaching lets say 5Ghz? or even 6GHz with some liquid nitrogen perhaps. I currently have an Athlon 64 X2 5000+(running on more than 3.1GHz on water from 2.6GHz and planning for some peltier stuff soon to reach over 3.5-4 GHz)
I think 4GHz is slightly disapointing and production samples will rapidly push past 4GHz once they hit their second spin.
If we're already seeing 3.6GHz+ air cooled on Q6600's @ 65nm then another 20% hike in speed should be pushing it to 4.2GHz easily.
The scary thing for all should be what the hell happens when Intel take away AMD's only defense in the way of adding a memory controller to the die. Low latency access with a shared L2 and ridiculously scalable frequency leaves AMD looking a rather poor second choice.
With Intel sandbaging already and enjoying healthy margin that is only going to be worse for AMD later on.
Whilst the article makes for very interesting reading why did TH choose to test the QX9650 CPU? This is one of the most expensive processors on the market today and well beyond the reach of all but the most well off users.
I understand that you want to achieve the best results possible and therefore chose the best performing processor for the job. However, this particular processor is the only one available with an un-locked multiplier. The majority of users don't have this luxury and only have an option to increase the FSB if they want to overclock their system. This sort of makes the test you've performed redundant for the majority.
A much better processor to test would have been the Q9550. While it's eventual clock speed wont match that of the QX9650, percentage wise it should be near as damn identical.
Otherwise, great article :-)
Whilst the article makes for very interesting reading why did TH choose to test the QX9650 CPU? This is one of the most expensive processors on the market today and well beyond the reach of all but the most well off users.... Blah Blah
Wakey, wakey idiot... This thread is months old and thats all that was out at the time...
Bob