UPDATE: Core i7: Blazing Fast, O/C Changes
Features
By
Bert Toepelt
published
Analysis--Triple-Channel Memory Does Not Impact Performance
Analysis: Triple-Channel vs. Dual-Channel and Single-Channel
We ran the entire benchmark suite once with each of the memory configurations. Our conclusion? It doesn’t matter much to the Core i7, with its 1 MB L2 cache and 8 MB L3 cache, whether it is running its memory in a triple-, dual-, or single-channel configuration. In everyday tasks, there is absolutely no tangible performance difference: the performance delta between a tri-channel and a single-channel configuration is only 2% on average. If you already own a dual-channel DDR3 memory kit and are planning to switch to the Core i7, you don’t need to spend the extra money on a third module to run your system in tri-channel mode.
Swipe to scroll horizontally
| Benchmark | Dual-Channel | Single-Channel |
|---|---|---|
| Crysis | -1.0% | -2.4% |
| Unreal Tournament 3 | -0.2% | -0.9% |
| World in Conflict | -4.6% | -9.2% |
| Supreme Commander | -0.3% | -1.9% |
| AVG Anti-Virus 8 | 0.0% | -1.4% |
| Winrar 3.80 | 0.0% | -7.2% |
| Winzip 11 | 0.9% | 0.0% |
| Acrobat 9 Professional | 1.0% | 1.0% |
| Photoshop CS 3 | 0.0% | 0.0% |
| iTunes | -1.6% | -1.6% |
| Lame MP3 | 0.0% | 0.0% |
| Studio 12 | -1.8% | -5.5% |
| DivX | -2.9% | -3.5% |
| XviD | -0.7% | -2.6% |
| MainConcept H.264 | 1.2% | 0.0% |
| Premiere Pro CS3 HDTV | -2.1% | -5.3% |
| Cinema 4D Release 10 | 0.0% | 0.0% |
| 3D Studio Max 9 | 0.0% | 0.0% |
| Fritz 11 | -0.1% | 0.0% |
| Nero 8 Recode | -0.4% | -0.8% |
| Overall: | -0.6% | -2.1% |
Analysis: Overclocking to 3.60 GHz and 3.84 GHz
Overclocking the Core i7 is very well worth the effort: boosting our CPU from 3.20 GHz to 3.60 GHz yielded a performance increase of 8.1%; running it at 3.80 GHz sped it up by 12.4%. Depending on the memory configurations, higher clock speeds are also possible, but we assumed a "worst case" scenario based on DDR3-1333.
Swipe to scroll horizontally
| Benchmark | 3.84 GHz | 3.60 GHz |
|---|---|---|
| Crysis | 4.4% | 3.1% |
| Unreal Tournament 3 | 2.8% | 4.0% |
| World in Conflict | 4.6% | 3.7% |
| Supreme Commander | 3.3% | 1.5% |
| AVG Anti-Virus 8 | 15.5% | 9.9% |
| Winrar 3.80 | 8.7% | 4.3% |
| Winzip 11 | 15.1% | 9.4% |
| Acrobat 9 Professional | 18.4% | 11.2% |
| Photoshop CS 3 | 17.2% | 10.8% |
| iTunes | 13.1% | 9.8% |
| Lame MP3 | 16.7% | 11.4% |
| Studio 12 | 1.8% | 1.8% |
| DivX | 14.0% | 8.1% |
| XviD | 15.1% | 9.2% |
| Mainconcept H.264 | 14.1% | 10.6% |
| Premiere Pro CS3 HDTV | 14.9% | 9.6% |
| Cinema 4D Release 10 | 16.3% | 10.5% |
| 3D Studio Max 9 | 19.2% | 11.5% |
| Fritz 11 | 19.1% | 11.5% |
| Nero 8 Recode | 14.1% | 9.2% |
| Overall: | 12.4% | 8.1% |
Stay On the Cutting Edge: Get the Tom's Hardware Newsletter
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
Current page: Analysis--Triple-Channel Memory Does Not Impact Performance
Prev Page Analysis--Hyper-Threading Yields 6% Speed-Up Next Page Test Bed Systems And DriversTOPICS
156 Comments
Comment from the forums
-
pullmyfoot AMD’s HyperTransport protocol can transmit up to 25.6 GB/s at 3.20 GHz.Reply
You mean Intel don't you? Other than that little mistake, good article
-
wh3resmycar hmm, question. once this nehalems come out. will we ever see a dieshrunk c2q again after the penryns? im expecting the price of this procs along with the mobo and ram to be too far off from my budget. orReply -
skywalker9952 One of the first side effects of Intel's domination of the CPU market is beginning to show. Since they don't have to compete with AMD in any market segment the i7 occupies, they have limited (significant) overclocking to only extreme models.Reply
RIP AMD.
May Abu Dhabi restore you to life soon so we don't have to suffer through more Intel ripoffs. -
sonar610 "The fastest Core i7, the 965 Extreme, is more than 2.6 times as fast as AMD’s current flagship CPU, the Phenom X4 9550 BE."Reply
This seems like an editing mistake maybe it should be 9950BE. -
cryogenic Core i7 is a great CPU, the article is not. I can't believe after all this time you still stack overclocked CPUs with unoverclocked ones. It's great to find out the overclocking potential of Nehalem but, at least include some overclocked Penryns in there too, to see how overclocked Nehalem stacks agains OTHER overclocked CPUs, because it's fairly evident that and overclocked new gen CPU will stack well with older non overclocked ones.Reply
-
joseph85 CryogenicCore i7 is a great CPU, the article is not. I can't believe after all this time you still stack overclocked CPUs with unoverclocked ones. It's great to find out the overclocking potential of Nehalem but, at least include some overclocked Penryns in there too, to see how overclocked Nehalem stacks agains OTHER overclocked CPUs, because it's fairly evident that and overclocked new gen CPU will stack well with older non overclocked ones.If it's evident then who cares?Reply -
fender22 skywalker9952One of the first side effects of Intel's domination of the CPU market is beginning to show. Since they don't have to compete with AMD in any market segment the i7 occupies, they have limited (significant) overclocking http://en.wikipedia.org/wiki/Overclocking to only extreme models.RIP AMD. May Abu Dhabi restore you to life soon so we don't have to suffer through more Intel ripoffs.Reply
My thoughts exactly... I wonder if there will be some sort of resistance to this sort of thing... It's like buying a car, you can do whatever you want to it (within the limits of the law) to make it as fast as you want. Sure, you may void your factory warranty, but it's your deal. You don't see car companies making it impossible for you to do what you want to their cars so you have to buy their expensive high end just to get your kicks... (not a perfect comparison, but it works)
I dunno, it's just pretty weak. And they are just taking advantage of the situation... -
cangelini sonar610"The fastest Core i7, the 965 Extreme, is more than 2.6 times as fast as AMD’s current flagship CPU, the Phenom X4 9550 BE." This seems like an editing mistake maybe it should be 9950BE.Reply
Fixed, thanks! -
onearmedscissorb Aside from the all too prevalent and potentially misleading typos, which someone needed to get a handle on as of months ago, I must say that the overall quality of this article is MUCH better than pretty much anything I can remember of the last few months. It's actually informative and thought out, rather than being a mess of assumptions that many people reading already know better than.Reply
Keep it up, and maybe I'll pay attention to this site like I used to.
But just for the record, I don't believe that it's particularly appropriate to use the overall average performance percentages as a basis for comparison between the "speed" of Core 2 vs. Core i7. Obviously, most people are going to be interested in the difference with games, where it's likely to be pretty minimal. But here and there, you have something like the 55% WinRAR difference pretty much skewing what otherwise would have been an accurate depiction of average expectations across the board.