OCZ 4GB PC3-10666 Platinum Low Voltage Dual Channel OCZ3P1333LV4GK
Apr 20th, 2010 | By StevenDetailed below are the system specs that will be used during testing.
- CPU : AMD X2 555 Unlocked to x4 and overclocked to 3.997 GHz
- MB : Gigabyte 790XTA-UD4
- GPU : 2x ATI Radeon HD5770
- PSU : Corsair HX620
- HDD : 2x Western Digital Caviar Black 1TB in Raid 0
- Case : NZXT Hades
- Cooling : Custom water
All benchmarks and tests were executed five times and an average was taken to derive the most accurate results. Most importantly, our tests will be run with the two highest stable memory clocks as well as the stock memory timings.
- 8-8-8-30 @ 800 MHz / DDR3 1600
- 7-7-7-30 @ 733 MHz / DDR3 1466
- 7-7-7-30 @ 666 MHz / DDR3 1333
Now that we have that out of the way, let’s take a look at the first set of benchmark results. Everest consistently delivers good information about memory performance.
From the first set of benchmarks, we see that more MHz doesn’t translate into higher memory performance. Timings are just as important for total bandwidth and latency. Thus, the 7-7-7-30 @ 733 setting clenches the performance crown in this first set of tests.
Next up, FutureMark takes the stage for testing with 3DMark 06 and PCMark 05.
Here we see results that are interesting. It seems that both 3DMark 06 and PCMark 05 both determine that the higher clocked system memory delivers higher system performance. This inconsistency is the reason why I run multiple different benchmarks to determine an all-around winner.
Last up is WPrime. For those of you who are unfamiliar with WPrime it uses a recursive call of Newton’s method for estimating functions. Basically, it tests CPU floating point performance using square roots. I use both the 32 million as well as the 1024 million tests for benchmark purposes. Keep in mind a lower number here is better.
And once again, the 7-7-7-30 @ 733 setting performs better than the higher clocked 8-8-8-30 @ 800 setting.
