OCZ ProXstream 1000W review - Testing setup and prodedures
Author: Vedran Dakic
Date: 14 Feb 2007

If you're testing a PSU with this kind of power, you really try EXTRA hard. So, coming back to our "How to shoot yourself in the foot" article I was fully aware that I have to "add" another parameter to the equation. By going with a large amount of hard drives and optical devices I'd run out of Y-power cables long before I get the load I want so - I decided to be plain mean. I mean, it's Valentine's day, do you know of any better day to be mean? :-)

That parameter is called - third 8800 card. I didn't have a third GTX in the lab at that time so I just used the GTS card. So, initial testing setup looked like this:

- EVGA 680i motherboard 250
- Intel Core 2 Extreme QX6700
- OCZ Platinum XTC PC2-8000 (2x1GB)
- EVGA 8800GTX (x2) in SLI
- WD RaptorX (x2) in RAID 1

In this "initial" phase, I really didn't think that ProXstream was under some "load". The cooler was rather lazy in action, which kind of reminded me of last year's CeBIT experiments with a couple of Powerstreams (the difference in power consumption between 7900GTX in SLI and 1900XTX in Crossfire was VERY noticeable). I have to say that I'm usually very cautious about the overall cooling performance on setups that are so hungry for power, so I took the liberty of adding a couple of more coolers on the front and the back of the ProXstream. Just to be extra sure. After 20 minutes 3dMark 2006 and three Prime's, I just accepted the fact that the score is 1:0 for the ProXstream. This was expected.

I went thinking for awhile and came up with a really mean plan. I actually decided to bring in the second machine and "feed" the additional 8800GTS from the ProXstream, and let the PSU of the second machine (GameXstream 850W) to power the rest of the machine. The second machine was based on ABIT's AW8 MAX and Intel's Pentium D 820, 2GB of OCZ's DDR2 memory and a whole bunch of other stuff not so relevant for this review. So, what I did was that I turned the second machine the other way around (so, back looking at the front), so the cables don't come too short. So, I had the ProXstream powering two EVGA's 8800GTX on the first testbed and one Point of View's 8800GTS from the second machine. This was actually working in Windows, both machines booted without any problems. While I was still recovering from this fact, I decided to turn on the same combination of benchmarks - 3dMark 2006 + three Prime's on the Kentsfield, and 3dMark 2006/Prime combination on the second machine. After removing my jaw from the floor when I relaized that this was actually working, it was time to declare the score. OK, it would be unfair to the ProXstream to count this as one point, so while still going "Mumbly" (you know Mumbly, the Hanna-Barbera cartoon character?) and doing only monosyllabic grunts for half an hour causing a couple of people to question my sanity, I took a pen and wrote "3:0" for ProXstream.

But it wasn't until I replaced 8800GTS with 8800GTX and added a couple of PCI-E power adapters to the mix that I was completely satisfied with my work. Both machines booted Windows, I started the whole benchmark suite again and after twenty seconds everything went bye-bye and beep-beep. Luckily, I didn't destroy all of the hardware - only ProXstream's fuse. This also shows the fact that this is really a good PSU - it may be killed, but it won't necessarily kill everything with it. So, the final score? Well, now, this has to be worth more then a point, right? So after a short period of "Tom and Jerry - Tee For Two" (the episode where Tom plays golf and uses Jerry for a tee while cheating on the game), I wrote 3,5:1,5. I believe it's only fair to give credits where credit's due. To ProXstream, of course.