Platform upgrade: Core i7-920 vs i7-3770 at 4.2GHz featuring ECS Golden Series Motherboard and Kingston
Overclocking
Core i7-3770″K” CPUs are multiplier unlocked and can easily be overclocked to well in excess of 4GHz. On the other hand, the Core i3 CPUs are locked and cannot be overclocked in the traditional manner. Only the base clock may be overclocked which overclocks the entire system. Overclocks of 3-5+% with stability are reported and there can be a small performance boost. The Core i7-3770K also has an unlocked base clock although we did not attempt to adjust it.
We found that we would have liked more options in the ECS BIOS to turn off Turbo and perhaps have better fine-tuning options. However, we were able to match the 4.8GHz overclock of the Gigabyte mATX Z77 motherboard. We were unable to get 4.9GHz no matter how much reasonable voltage we applied. We could boot into Windows but the system was unstable. Evidently, the issues are not thermal-related as temperatures remained around or below 80C under load.
Issues with 8GB (2x4GB) of System RAM and Overclocking
When we upgraded our Ivy Bridge platform from 4GB to 8GB system RAM (2x2GB to 4x2GB) for the last round of benching games and for all of the other tests, we noticed that it took significantly higher voltage to stabilize the CPU overclock and the system. One RAM upgrade solution from 4GB to 8GB would be to contnue to use only 2 DIMMs but upgrade by using higher capacity RAM (2x4GB).
The Ivy Bridge Overclock versus Bloomfield’s
The highest overclock we could manage with our Core i7-920 was 4.2GHz and it was accomplished stably with our Thermaltake Frio OC CPU cooler. To achieve the same 4.2GHz with Core i7-3770K we only needed .03V more added to stabilize our system. To reach 4.8GHz required .13V and we needed our Noctua NH-DH14. If the reader is interested, they can check out the Noctua NH-DH14 evaluation to see how we reached our maximum overclock with Core i7-3770K or the Frio OC evaluation to check out the i7-920 overclock.
Here is Core i7-3770K at 4.8GHz using Noctua’s NH-DH14 to remain below 80C under full load.
Although Ivy’s temperatures were good – below 80C under maximum load – we simply could not reach any higher no matter how much reasonable voltage we added. In all of the tests between Bloomfield and Ivy Bridge, our CPUs are clocked to 4.2GHz. Ivy is clocked higher when we are trying to demonstrate the effects of CPU scaling on performance and these further overclocks are identified on all of our charts.
Kingston RAM upgrade: DDR3 PC1800 to PC1866 and 4GB to 8GB
Kingston sent us DDR3 PC1866 to replace the 3 PC1800 that they sent us three years ago. Evidently memory has got much faster and we want to demonstrate what performance increase there is in gaming. For the Bloomfield platform, we first tested with 6GB PC1800 at 1200MHz and then compared gaming benches with PC1866 at 1333MHz. For Ivy Bridge, we compared 4 and 8GB PC1866 at 1333MHz and 1867MHz. Of course, Bloomfield use Tri-Channel configuration and Ivy Bridge uses Dual-Channel.
We will increase our Core i7-920’s memory speeds from 1200MHz to 2000MHz and compare it to increasing Core i7-3770K’s RAM speeds from 1333MHz to 1867MHz.
4GB versus 8GB system RAM in Gaming
Since Kingston was very generous in answering our request to test their HyperX PC1866, we are now able to finally settle if there is any performance difference between 4GB and 8GB in gaming. 
As you can see there is no performance differences playing with 4GB or 8GB of system RAM and we can be sure that there is no difference between 6GB and 12GB in Tri-Channel. Of course, there is every other reason to upgrade your PC from 4GB to 8GB, but gaming is not one of them.
Effect of RAM overclocking on gaming performance
Now we are going to compare again using our entire game suite using Core i7-920 at 4.2GHz and Core i7-3770K at 4.8GHz with a GTX 690 – the only differences being the speeds of the 4GB of DDR3.
As you can see, there is a definite performance increase just from clocking the system memory higher which appears to be dependent on the individual game engine. Please review the results again in this evaluation’s Gaming Performance as part of a full chart.
SSDs and Performance
We have performed many SSD versus HDD evaluations in the past and SSD is without question the single upgrade that will make your system faster and noticeably more responsive. Although there is no difference in gaming performance between games installed on a SSD and the same games installed on a HDD, loading games, saves, and levels will be much faster.
The only issues for most gamers is the cost of storage of a SSD versus a HDD. Only recently have the costs of storage dropped as prices have fallen close to $1 per Gigabyte for smaller capacity drives. Most gamers would want at least 240GB of storage.
We are using two 240 GB Kingston HyperX SSDs that have essentially identical performance – the only difference is that the HyperX “3K” SSD uses less expensive NAND and the writes may be slightly slower than the HyperX (5K) drive. You will only see these differences in SSD synthetic benches. We have installed Windows 7 using Bloomfield on the HyperX 3K SSD and an image of the same drive for Ivy Bridge on the regular HyperX SSD.
Since we have 240GBs of storage available to us and our game library is much larger, we have installed 17 games of the newest games on SSD and we duplicate the installation on a 500GB Seagate HDD with the addition of the rest of our games. All testing for this evaluation was performed on SSDs except for the gaming benchmarks or as noted. The reader should note that the synthetic benchmark scores are significantly higher when testing using a SSD over a HDD.
In a follow-up evaluation, we will compare not only these two premium SSDs, but also a Vnow 200+ Kingston consumer SSD against our HDDs. In the meantime, there are SSD performance benchmarks in the Summary and More Tests sections that will demonstrate the advantages of using PCIe 3.0’s higher bandwidth over Bloomfield’s older platform using USB 2.0
Let’s head over to our test configuration.
No offense, but there are two main summary charts that are identical to each other. Is the second one supposed to show different games?
Don’t forget about the “[insert chart]” part! 😛
None taken. Thanks for pointing it out!
Fixed.
Never mind my previous comment, as it’s now fixed, thanks!
Of course, Civilization V does better with Ivy Bridge than with Bloomfield, all clocked the same. At 4.8GHz (IB), CivV is like 30% faster than Nehalem (Bloomfield) at 4.2GHz.
BUT what amazes me is that Nehalem still seems to be doing just as well as IB overall, clock-for-clock. In Crysis 2, Nehalem is like 15% faster than IB. At 4.2GHz, Nehalem still beats IB at 4.8 GHz in DX11 mode, by around 15%!! I was thinking of blaming it on the system memory bandwidth (with Nehalem using triple channel), but even at only 1200MHz, Nehalem still shines pretty much just as much as it does at 2000 MHz.
The case reverses in the favor of IB but only by a few percentage points for Crysis 1, LP2, HAWX2.
It goes to show that unless one is a serious Civilization V gamer, an overclocked Nehalem would be just fine for even a GTX 690. I would guess that Starcraft 2 and Skyrim (other CPU-limited games not tested here) are the only other games that show a noticeable improvement on the Ivy Bridge.
Thanks, apoppin for all of this benchmarking that pretty much nobody else on the ‘net have done with all of these games, on the still mighty Nehalem!!
Amazing work!
It would be one of these, but with an Athlon II X4 OC with maximum range and I7 3770K OC max.
Athlon X4 ~ 3.5Ghz vs 5Ghz i7 3770K for example
It would be interesting to see one of these with a single GPU both as a dual system which is where I think you will notice more.
Sorry for my English.
I have a 3770k and a I7 920. Its was a nice review because many people with older i7s are itching for an upgrade. I do think this review is biased in its language. The non scientific language of “Way Faster” vs saying its 8 percent faster is a dead give away. “What faster is subjective where a percentage is objective. Way faster to them seems to be any test that the Ivy beats the bloomfield in.
USB 3.0 ports are nice but can be installed in an older X58 board (they fail to mention)
For gaming its not worth the upgrade at all. For most windows tasks its not worth it.
For tasks that take hours then it might be worth it. Ie encoding, but here a 6 core would be better than a quad. So an upgrade to a 6 core might be cheaper on x58 than doing entire MB/RAM swap.
Not trying to rain on the parade but this review seems to want to justify the upgrade by using subjective wording rather than just stating the percentages.