Intel's Core i7: Blazing Fast, But Crippled O/C

[BenjiMo]

You could get 2.5 9950BE's for about $425 from newegg and thrown in a board for $125 and you are at $550. The flagship i7 965 and a mobo are $1200 at a minimum. Obviously you can't just buy 2.5 cpu's but if you take Tom's comparison literally you realize how much of a price premium there is, and not just over the 9950 but the Q9300 and so on. Not really the best way to compare things is what I'm trying to say - it's like saying you need 2 honda s2000 i4's to equal the horsepower of a 911 turbo. True, but its not really helpful comparison when you factor in the $

 

[kamkal]

http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3448

apparently no problem in getting the 2.66ghz chip upto 3.6

 

[kamkal]

[citation][nom]jckrieger[/nom]You guys talking about the i7 being locked are idiots. Toms, you guys need to overclock the lesser models so people can get over the power consumption limit. I run a decent air cooler on my Q6600 and I can say I went from 2.4ghz to 3.0ghz with no voltage increase. With that said, it should be possible to overclock the i7 from say, 2.6ghz to over 3ghz without even thinking about hitting the current/power limit. Intel implemented the power limit so people wouldn't fry the poor cpu's from excessive voltage increases and poor cooling. [/citation]


have to agree here, you guys make it sound like the i7 is somehow not OC'able unless you get the extreme edition

i mean even core2 has mechanisms to throttle itself back when temperature gets too high

 

[Bodhammer]

Great Article - I also would like to MS FSX benchmarks, that is the one program that taxes my current system and I've been waiting for i7 to upgrade my system from a P4EE Galitin

 

[bounty]

You guys said that you will have to buy a new cooler, but it looks like you guys used the CNPS 9700 LED. What gives, did you guys hack it? If so please share.

 

[Area51]

That’s the exact same bandwidth that the classic FSB of the Core 2 offered at 400 MHz.

Not True. This is a BI-Directional path SO you must multiply your bandwidth by two.

 

[FrancoisPiednoel]

Let me add a little detail. In the BIOS, you have a setting call "CPU VR current override", and if you select this to ENABLE, your CPU will ignore the TDP limits, and the TDC limit. There is no "cripple Overclocking", we proactively putted this feature to make sure the Overclockers will have fun.
Try it yourself, and have fun!

Thanks!
Francois Piednoel
Intel Corp.

 

[cruiseoveride]

Boy! do we need competition!

$1000.00 CPU? What is this..the 1980s?

Lets hope AMD comes out with another HD4870, Core i7 killer, soon!!! This is ridicule.

 

[NightLight]

Excellent work by Intel. I don't really care about oc'ing, because if the processor runs fast enough, you don't _need_ to overclock it.
It will be the end of an era with the old fsb gone. I remember changing jumpers on my 486 to get the fsb & multiplier right
I've always been an intel enthousiast, but it's sad to see that amd isn't able to keep up in any way, and it will show on the high end market prices. That's why we need AMD (I kid)

 

[silversound]

Thanks for the review, but i think for gaming, you need to provide the setting, how do you get 150FPS in crysis? Lew setting? So overall gaming arent much faster, i guess i will just stick with my e8400 then.

 

[megamanx00]

Well, it looks like other reviews are able to somewhat overclock the lower end I7 models, but I don't like the idea of them being limited that way, especially when even the lower end models and the board to use them are so expensive right now. This really puts the pressure on AMD. Their next chip needs to be at least 35% faster to be competitive in the mid range. That's a big hurdle especially since the 45nm Phenom to be released in January will only have a 15% increase in core speed to 3.0GHz over the 9950.

 

[silversound]

Bert, your conclusion seems a little vague, you should include the performance increase over the same clock core2. Over 2.6 times faster than the 9950BE sounds like a pro-intel comment. You should be neutral like other review sites compare same clock to clock and in same categaries like:
"The Core i7's general purpose performance is solid, you're looking at a 5 - 10% increase in general application performance at the same clock speeds as Penryn. Where Nehalem really succeeds however is in anything involving video encoding or 3D rendering, the performance gains there are easily in the 20 - 40% range"

 

[cangelini]

Just updated the piece to reflect confirmation that Turbo mode is enabled on the lower-end i7 parts (using Core i7 920 and ASUS P6T Deluxe)

 

[Worf101]

Well AMD lover's warned you what would happen when Intel re-established its stranglehold. We warned you and warned you now you see it in Black and White:

Locked Mulitpliers
Throttles
New expensive chipsets
New expensive memory
Fast tech only a millionaire can love.

Sigh...

Da Worfster

 

[camulus]

Anandtech seems to have been able to overclock the 920 to 3.8 GHz. Personally I wouldn't think about overclocking it much higher, anyway. Stock speeds are fairly fast, and with a 43% overclock I'm guessing it would quite comfortably cruise past the 940.

 

[Lowdown]

I thought this was a good article, probably will be awhile before I upgrade. It will be interesting not having FSB speeds and stuff like that on a Intel build but I guess the newer AMD builds are like that too, I just haven't had a chance to build one of them. Question though, I have the Zalman cooler on my computer right now and I was wondering why you have it pointing vertically, most pictures that I have seen, along with how I have mine mounted is horizontaly. Any reason for that, I haven't seen any showing the correct way or which provides better results.

 

[WINTERLORD]

this is gonna sound noobish. but this is a great article. i just don't rember what all the benchmarks are for. a legend may be helpfull, i know sandra and PC advantage, or somethin. one is theoritical and the other is practicle real, life. not sure wich. but once again alswome article

 

[cerulean]

I agree with many of the others that this was a well written article. Good job.

On the negative side, this sure doesn't look good for AMD (and the end user on that note). I'm hoping for our wallets sake that AMD is able to turn things around. If they can supply a decent chip that still provides a means to continue tranditional overclocking, they might still stand a chance. I agree that part of the joy of building your own machine is to be able to crank up the performance a bit and save a few bucks in the process, which for many people actually puts their desired performance levels into a price that they can afford. If Intel is allowed to dominate the market like this, low to mid-level overclocking will be a thing of the past. While people may pay more for their systems I think it's really just going to encourage them to upgrade less often. Then again, without any competition, the market won't move as fast--so you won't have to upgrade as often.

Despite being a new chipset, $750 for the low end (not including other system components) is quite a bit--more than many might spend on a whole-system upgrade! Intel's new anti-overclocking policy really opens your eyes to the fact that they are really trying to lighten the load of your wallet. When they have the means to do so--they will employ it to its fullest extent. I know this is true of pretty much all corps, but it stings when it's so evident as this. This is why I have never had any brand loyalty.

Now we sit and watch for AMD to make its next move.

 

[hannibal]

What is really good about i7 is that it seems to be ralatively fast even with games. That was not something that preliminary rumours told about it.
Really interesting piece of technology indeed!

For the competition sake I really hope that AMD can make a quantum leap... or there is really lonely in the very top of CPU's.

 

[hannibal]

[citation][nom]FrancoisPiednoel[/nom]well, Fugger at xtremesystem got the new Core i7 running at 5.247Ghz, no problem for overclocking. http://www.xtremesystems.org/Forum [...] p?t=206571 [/citation]

Well they used the extreme edition CPU and a guite extreme cooling system too
Not something that can be done with normal air cooling. But yeh, it shows that extreme edition overclocks about as well as those "good ole" "old" core2's.

 

[Sapper123]

On page 34 the title says Fritz 12 while the table has it listed as Fritz 11.

 

[skalagon]

I can see a lot of people getting pissed off with this overclocking barrier. Talk about a way to reinforce their big bad mega company screwing the little guy image. If AMD release something good I could see a lot of people switching even if only for this reason.

 

[radguy]

I think you guys are jumping the gun a little bit on this overclocking lock thing (and definitly with the water cooling suggestion). 130 Watts probably isn't as big of a cap as your making it out to be plus water might be a good way to keep things cool and efficient. This lock is probably just a slightly worse than the current limit of not having an unlocked multiplier. Why don't you guys try overclocking it and tell us how bad it really actually is. I could be wrong as you guys actually have a sample chip but I want to see the data.

 

[gonchuki]

so, I'm not getting why are you using a vanilla 9950BE AND an old SB600 based motherboard on AMD side, to compare against a whole batch of Intel CPUs at much higher clock speeds.
Couldn't you just use an SB750 based board and OC the AMD part to 3.2ghz so you could at least make a clock for clock comparison?

That supposed 62.2% faster sure looks impressive, but the i7 965 also has a 23% advantage in raw clock speed on those tests. Level the field next time before you make those bold statements.

 

[jaragon13]

So this is a complete Intel fanboy orgy or what?