By what measure would you consider that calculator to be accurate? To be sure, it has lots of options, so you can very precisely assemble exactly the computer you're interested in. However, the data on which it bases its calculations is completely and totally wrong. In my experience - compared to actual measurements from my Kill-a-Watt - that site will estimate consumption to be double or occasionally even triple the actual value. For an actual, real-life example, it estimates my computer at 276 watts, which, with my 80+ PSU, would be 345 watts AC at the wall (where the kill-a-watt measures). In truth, I can barely break 157 AC watts. The SnooP PSU Calculator I mentioned in my previous post, however, estimates my computer at 131 watts. Factor in my PSU efficiency and you're looking at 164 watts. I'm willing to give them a 4% margin of error simply because they didn't have quite the right components and I was forced to use the most similar parts in the list. The Extreme PSU Calc, on the other hand, had exactly the right parts, so guessing 219% of the correct value is inexcusable. If you have a power meter which closely matches the results you've been getting from the Extreme PSU Calc, I suggest you replace the meter, as it is irreparably broken.
Do you measure your power consumption at peak load (Crunching Prime95 for example). Generally I use that calc. to measure peak wattage to calculate the size of UPS for server installations and the wattage is measured with a metered or switched PDU from APC, I find it hard to believe they are wrong. Perhaps there is a difference between the accuracy of their server vs. desktop parts?
The Extreme calculator is recommending me a 250W PSU, my metered figures show I'm only using 125W (and from the wall) running Prime small FFTs, the highest wattage I can generate. A few more metered checks here.
This is what we use - bargain! We've also got a Watt's Up Meter that makes nice graphs that I've just started to use
I measure my power consumption while running Orthos Small FFTs and either RTHDRIBL or ATiTool, a combination which gives me 100% load on both my video card and my CPU. My drives are more or less at idle, but that's only a few watts and is actually accounted for by the SnooP PSU Calc I've been suggesting.
I just put my list into http://web.aanet.com.au/SnooP/psucalc.php - had to choose a few near matches, but the result was practically bang on the money - wish I'd used that when speccing a PSU!