Here is a look at the Test bed Specifications used for this review. For the overclocked portion we settled on a CPU frequency of 3.37GHz and memory speed of PC2-1125 MHz. Once we broke above 1125MHz we had some weird results some applications would not benchmark at all or we got some strange results. Being that these were both 4GB kits we were testing in Vista x64. Again, this caused a lot of anomalies and forced us to remove some of the tests entirely as the results were so skewed testing in Vista x64 while compared to Windows XP x86 it was irrelevant to show them. Why does 3D Mark 2001 SE bench about 20,000 points lower in Vista?? Anyways, this review was difficult to pull together enough applications that would work properly between the OS and the applications it was a long and trying process.The kit used for comparison is the Mushkin 2 x 2GB, 5-4-4-12 PC2-800 set.
Initially I set out to test a lot more applications; however it became clear early on that these results were being echoed over and over again. So I decided to hack some of the charts and just provide the few benchmarks we all live by.
Looking at the results with the Theoretical testing we can see that keeping the CPU at 2.4GHz and just moving from PC2-6400 to PC2-1066 impacts performance ever so slightly. SuperPI went from 4min 8sec (1066) to 4min 15sec (800) and when overclocked to 3.37Ghz, PC2-1125MHz dropped to 3min 1sec. Bandwidth followed suit showing a nice theoretical gain moving from PC2-1066 from PC2-800 and again once overclocked. Everest Latency showed more of the same.
So while the theoretical testing came as no surprise. Faster ram = faster results. How would this affect the real world testing?
Quake 4 seemed to favor the lower timings of the Mushkin PC2-800 kit with the CPU kept at 2.4GHz using DDR2-1066 didn’t really matter at all. Overclocking the CPU and memory provided a nice gain however. Call Of Juarez (DX10) apparently doesn’t even care about the memory used and the score changed so slightly the results were moot. The last real world application we used was Unreal Tournament 3 and again the timings look to have affected the score more than the frequency. All tests were run at 1024x768 to highlight the performance differences in the memory speeds.
|
Alright, so basically what these results tell us is that somewhere along the way the frequency removed the bottleneck of memory throughput. This would be disappointing information for all you speed freaks that loved your low latency, high clocking DDR1. Basically you can use a kit of PC2-800 and not lose a whole lot of performance over a kit of 1066. I even went back and set the clock speeds higher to see if the CPU was now the bottleneck. What I saw time and time again was that it just didn’t matter. The speed of the memory really didn’t affect the results. Once you break into DDR2 800MHz you’ve already removed the memory bottleneck.
Conclusion
Results aside, the Corsair Dominator kit performed flawlessly. Not a single crash or glitch was due to the ram. The Corsair Dominator Cooling Fan kept the memory cool to the touch even during extended runs. The biggest disappointment with this review was trying to get Vista x64 to co-operate. I had no problems with this kit in Windows XP x86 however I could only address 3.25GB of the ram due to the addressing limitations of a 32-bit OS. I would rather run and test in XP over Vista any day after the nightmare Vista gave me during this review. The UT3 tests were at least 20FPS higher across the board in XP, 3D Mark 2001 SE was 20,000 points higher in XP than Vista. Some applications like Super PI would crash in Vista. I had to force it to run as Administrator and change some other properties just to get the PC to complete. No such issues occurred in XP with the same memory kits at the same settings. I ended up reformatting about 3 times to complete this review as the OS issues were just killing me.
So in closing, the Corsair Dominator PC2-1066 was a great kit to run and test. Stay away from Vista x64 and everything was a pleasure. We cannot fault the memory for the experience of the poor Operating System however. In Windows XP I ran this kit flawlessly for two weeks, not suffering a single crash or glitch. And while the real world results didn’t show any astounding performance differences I will say that while using the Dominator PC2-1066 gaming performance felt smoother to me. The Achilles heel here is going to be the price. It’s hard to justify a $300 purchase when the $120 kit of Mushkin ran neck and neck with it. However, for those overclockers out there that push every component to the edge of stability this kit should satisfy that urge. One last note is that the Corsair Dominator Cooling Fan worked wonderfully and kept the memory cool to the touch throughout all of the testing performed. This product can be purchased separately.
Newegg stocks both this memory kit and the Dominator fan unit.
Pros
Performance.
Quality and lifetime warranty.
Includes Dominator Fan Cooler.
Cons
Expensive.
Back Home
