OCZ PC2-8000 4Gb Platinum Edition Dual Channel Memory
Author: Frank Stroupe
Editor: Shawn Knight
Date: 02-25-2008
Provided by: OCZ Technologies
Pages:
Testing/Overclocking

Test Rig:

Intel Core 2 Quad Q6600, G0 stepping
Asus P5K Deluxe Wifi-AP
Zotac geForce 8600GTS
Rosewill Xtreme 850 gig PSU
OCZ Vendetta CPU Cooler
Zalman VF-1000 GPU Cooler
Microsoft Windows Vista Premium 64-bit

The test rig has proven to be an excellent overclocker, having seen CPU clocks over 3.2gHz, FSB over 500mHz, and pushing 2 gig memory kits to well over 1300 MHz. There is nothing here to hold the Platinum Edition back.

Keep in mind that you will need to use a 64-bit operating system to utilize the entire 4 gigs of memory. A 32-bit system will utilize a maximum of about 3.2 gigs (depending on your video card memory, etc). Also, if you are doing a new build, it is usually better to install Vista with only one memory module in the system during installation, when using a 4 gig kit.



The first thing I did after installing the memory was run MemTest86 on each module individually. It found no errors on either stick.

I found that they defaulted at DDR2-800, 5-5-5-15, 1.8v, which was the max setting on the SPD table. I went into the BIOS and merely changed the memory settings to DDR2-1000, 5-5-5-15, and 2.1v, and the system wouldn't boot. After a few tries, I raised the FSB from the stock 266 MHz to 300 MHz, reset to DDR2-1000 and it booted fine. I'm assuming that the stock FSB was too low to accommodate a 1000 MHz memory clock, but I'm just guessing. It isn't a big deal, as I never leave the CPU at stock clock anyway; I usually run the FSB from 330 to 350.

Normally, I test a kit of high-performance memory for flexibility, to see how it will boot at lower voltages and timings. As the memory performed just fine at DDR2-800 at CAS 5 and the DDR2 standard of 1.8v, I am satisfied at the range of frequencies and voltages that it will run at, which will accommodate most PCs built within the last couple of years. At DDR2-800, I tightened the timings to 4-4-4-20, after raising the Vdimm to 1.95. I am definitely satisfied with the flexibility of this memory.


For overclocking, I spent a significant amount of time experimenting with FSB, timings and memory clock frequencies. DDR2-1072 6-6-6-20 was the maximum stable overclock I could get with the PC2-8000 Platinum Edition.

I recently reviewed the OCZ ReaperX PC2-6400 4 gig kit. The PC2-8000 Platinum Edition actually gave about the same all-around performance as the ReaperX, a heatpipe-cooled kit costing $30-$50 more than the Platinum Edition. Of course, the ReaperX was at a pretty stout overclock to reach the advertised settings of the Platinum Edition, which might not be obtained with a rig not as conducive to overclocking as my test rig.

The Platinum Edition seemed to have some headroom, as it never felt more than just warm to the touch, even at max overclock. I'm a little surprised that it didn't overclock more, but then again, this line isn't designed with overclocking in mind.

I ran a battery of benchmarks just to see how the memory performance changed between advertised settings and overclock. Most of the benchmarks are memory intensive. For gaming, I ran the onboard benchmark on the RTS game Supreme Commander, and the Crysis Benchmark Tool created by "MadBoris" Vergiza. Supreme Commander is heavily multithreaded, and in Vista, really puts a system to the test. Crysis is also heavily multithreaded, and extremely memory intensive. Either game will crash a system if a memory overclock isn't rock solid stable. I ran both game benchmarks at medium settings and 1024 x 768 resolution, due to the limitations of my video card.

In every case but one, the overclocked scores were better than the scores with the memory at advertised settings. For some reason, every set of memory I've tested with the Crysis benchmark dropped FPS slightly as the memory reaches it's maximum overclock.

Below are the results from my benchmark and game testing. Things are pretty self-explanatory, so I will spare you the boredom of reciting what you see in the charts. Do keep in mind the actual numbers, as the charts can be a bit misleading (results looking more drastic than they actually are).


















  Sponsors Links

  Sponsors Links