Now that the installation has been covered, it's time to crank this baby up. Before doing so, OCZ recommends turning the potentiometer counter-clockwise until you hit the stop. This ensures that the Booster starts up at the lowest setting of 2.6V as opposed to, say, 3.9V. When you power the system on, the numeric display will indicate the current voltage level. In my experience testing the Booster, I found that it was a good idea to verify the voltage reported on the Booster using the motherboard BIOS or a monitoring utility like Motherboard Monitor. A smaller potentiometer (think jeweler's screwdriver small) under the numeric display allows you to adjust the calibration if you find it to be off.
Personally, I would recommend relying on the LED display only as a general guide. This has less to do with the calibration of the device and more to do with its precision. When dialing in 2.8V for example, you could be dialing in anywhere from 2.79V to 2.89V without knowing it, because the display on the Booster lacks a hundredths digit.
Once the system is powered on, OCZ recommends you enter the system BIOS before adjusting the voltage, because the act of adjustment can cause an operating system to lock up. I did manage to blue-screen Windows a couple of times by turning the knob too quickly, but generally I found that if I moved in slow, small steps, the system tolerated it just fine.
In terms of operating the device, that's basically all there is to it. Disappointed? What did you expect for a device whose control mechanism consists of a single knob?
Testing notes and our testing methods
OCZ's description of the DDR Booster is interesting:
OCZ DDR Booster Diagnostic Device with Patent pending PowerClean Technology™ allows users to supply cleaner power to their memory modules, resulting in more stable memory. Additionally, users are able to view memory module voltage with the digital LEDs on the DDR Booster, allowing simple and inexpensive troubleshooting of the memory modules.That's the official version. I suspect the unofficial version is more akin to 'It lets you overvolt the snot outta your RAM!" but I don't have any confirmation on that. Nonetheless, the "cleaner power" claim is an interesting one, and it bears further examination. As a result, I decided to set up four testing scenarios: Stock motherboard voltage (2.6V), maximum motherboard voltage (2.8V), the equivalent of maximum motherboard voltage supplied by the DDR Booster (2.8V) and a significantly higher voltage supplied by the DDR Booster (3.3V). The idea with the second and third test conditions is to see if letting the DDR Booster supply the power will result in higher stable RAM speeds than the motherboard alone. If so, it would imply that the power coming off the DDR Booster is of higher quality than the motherboard power.
OCZ sent along some memory modules which they said responded well to higher voltages, but I wanted to test a variety of RAM, so I assembled a total of five sets of dual-channel DIMMs from a number of manufacturers. Timings were set manually to 2.5-3-3-8 to set a level playing field. In all cases, these were looser timings than the SPD of the memory, which should hopefully allow for higher speeds.
The objective here was to crank the memory speeds as high as possible. Motherboard Monitor was used to set the voltage as close to the target as possible; I found that relying on the DDR Booster display alone could result in significant variance in the actual voltage, and a few hundredths of a volt could significantly affect the maximum speed of the memory.
Our Sphinx speech recognition benchmark was used as a quick stability test for the memory. (Over time, we've found that Sphinx is very good at exposing iffy memory configs.) Essentially, I would start at a memory clock of 205MHz (5MHz over stock) and run Sphinx to determine if the test configuration was stable. If so, the memory clock got bumped 5MHz higher and the process was repeated until Sphinx crashed. Then, the speeds between the last stable setting and the unstable setting were tested to determine the highest stable speed to the nearest MHz. This procedure was repeated for each combination of DIMM type and voltage.
Since I was only testing for maximum memory speed, I didn't want any other component to get in the way. Consequently, I lowered the multiplier of the CPU from 12X to 8X, locked the AGP and PCI busses at their default speeds, and lowered the HyperTransport multiplier to 2X.
Finally, I decided that I would just go hog-wild on whichever RAM performed the best in the above tests. That RAM would undergo one final max stable speed test, this time at 3.7V. Speaking of which, OCZ maintains that while the DDR Booster does get very hot, active cooling is only necessary if the voltage is 3.4V or above. After running some initial tests, the heat of the Booster and the memory just made me too nervous, and I resolved to come up with an advanced, high-tech cooling solution. Behold!
Our test system was configured like so:
|Processor||Athlon 64 FX-53|
|Memory size||1024MB (2 DIMMs)|
|Memory type||Corsair XMS3200LL PC3200 DDR SDRAM|
Crucial Ballistix PC3200 DDR SDRAM
Kingston HyperX PC3200 DDR SDRAM
Mushkin LII V2 PC3200 DDR SDRAM
OCZ Gold Edition VX PC3200 DDR SDRAM
|RAS to CAS delay||3|
|Graphics||ATI Radeon 9800 Pro 256MB|
|Hard drive||Seagate Barracuda V 120GB SATA 150|
|OS||Microsoft Windows XP Professional|
|OS updates||Service Pack 2, DirectX 9.0b|
We used the following versions of our test applications:
The tests and methods we employ are generally publicly available and reproducible. If you have questions about our methods, hit our forums to talk with us about them.
|Toshiba's OCZ RD400 512GB SSD reviewed||7|
|Gigabyte shows off its thin Aero laptops and Aorus RGB Fusion Keyboard||7|
|Deals of the week: 25% off Das Keyboard 4 and more||4|
|Everyone and their gran announces non-reference GTX 1080s||32|
|AMD FirePro S7100X is ready to virtualize blade-server graphics||5|
|Thermaltake Pacific water coolers gain hard tube option||9|
|Rumor: Google shames partners into updating Android||36|
|First GeForce GTX 1080 driver out with new VRWorks features in tow||28|