When recommending PC parts to people, I tend to be most strict about my power supply recommendations, basically because there are a few brands (Corsair, SeaSonic, XFX) that have a proven track record of building solid, quality units. Oddly, these recommendations tend to draw more controversy than any other. More often than not, I'll have others fire back at me: "Look at this great deal on a [second-rate brand] PSU! It's much cheaper than what you recommend, and look! [Hardware reviewing website] gave it their Gold Standard Seal of Approval!" I've always felt that these PSU reviews were suspect, but could never quite articulate why. Thankfully, I recently discovered this article by Gabriel Torres over at HardwareSecrets that sums it up better than I ever could.

tl;dr for people who don't want to wade through the entire article: Most power supply reviews are wrong because...

1) The reviewer's use of a multimeter is not an accurate measurement of a PSU's stability. Only very, very cheap PSUs fail the multimeter test. Instead, reviewers should measure current by applying a correctly distributed load to the PSU.

2) Most reviewers do not use an oscilliscope to measure noise and ripple levels. High noise and ripple levels can damage your PC's components.

3) Most reviewers hook their test PSUs up to a PC instead of an active load tester. Your average PC cannot generate the kind of power needed to adequately test a 600W+ PSU under full load. Torres cites the example of this website giving a "Gold Award" to a 750W PSU that literally cooks itself if you apply just 500W load to it using a load tester.

So basically, most hardware reviewers know their PC hardware, but they aren't electricians. The best measures of a PSU's capabilities are its noise and ripple levels and its performance under heavy current load, and most review websites don't even test those two things using the correct methodology. Think about that the next time you consider putting a second-rate but "highly rated" PSU in your shiny new desktop.

lol...I agree with this. But I mean, just like with everything, you get what you pay for. I have an Antec CP-850W...I paid a pretty penny and the thing runs a dual Nvidia GPU graphics card with no effort.

Its even certified to run the GTX470 in 3 SLI supposedly...

I wouldn't listen to reviews either...but sites like this are nice http://www.slizone.com/object/slizone_build_psu.html

Coolmax is trash. We get low power ones for around 12-15 bucks. I wouldn't install that in any computer with more than an integrated chipset and a single hard drive.

The rating is most likely peak power, therefore it's absolutely certain that it will never sustain 750w for more than a short instant (milliseconds). Forget about continuous 500w.

As far as methodology goes, I agree that testers need more sophisticated tools, like load testers. It's just not worth making a review in other conditions, might as well stare at it for a hour and see if anything abnormal happens. :/

Huh, good to know, I'll be needing a new power supply in may when I get a new GPU.

I don't know what all the fuss is about power supplies. In my 15 years of being around other people's computers, building my own computers, and building computers for others, I never had trouble with the cheap power supplies I used, except for the rare failed ones (cheaply replaced), and only once did I see a massive (i.e. complete) failure of a PC which might have been a power supply thing (but not from a voltage fluctuation since the other computers form the area felt nothing, most likely a MB short circuit as I remember, not mine).
Since I never go for a "max theoretic power plus 200W" philosophy, that I see is common, I barely reached the 500W barrier recently. Who needs more for a dual core, 2 Gigs of RAM, a couple of HDDs, some random optical drive, and a good enough GPU combination anyways?

I always thought the PSU reviews are full of it. Cheap PSUs worked fine for me.

I need more than 500W easily...GTX295 graphics card power draw is below.

I'm thinking of upgrading to GTX470 SLI setup which will put my video card power draw at 550W (under load).

I know Tech Report do serious testing on that front, they still use this rig for stress testing:

http://techreport.com/articles.x/16073

Although that's not why I looked it up this review, it has this to say about CoolMax: "Before we had a chance to strap it to The Beast, the CoolMax unit died an unspectacular death in our real-world test system."

double post

Now you're just showing off :P. (good for you)

Plus, if I had those babies in my computer (or more likely an equivalent ATI setup) I really take care when buying the PSU, or maybe a good enough PSU and a protection something in front of it.