db said:
Try the following on that Asus page. In the PCI section, change the
"Modem, Network LAN..." entry from zero cards to one card. The total
system power increases by 50 watts. When is the last time you saw something
like an Ethernet card eat 50W ? It's just absurd. An Ethernet chip might
consume a couple watts. Test some of the other entries, and you'll find similar
anomalies. Their rounding function makes it hard to tell what is
going on.
Another item of contention, would be their handling of hard drives.
If I select "10 hard drives" in the Asus page, that adds 350 watts,
or 35 watts per drive. Real hard drives have two power numbers to
consider. One is the "spinup power", measured during the first
ten seconds. The other is the "running power", the power they consume
the rest of the time. Running power for a modern 3.5" drive, is 12W or less.
Spinup power is 12V * 2.5A + 5W or about 35W. So the Asus calculator
is using the spinup power, for the entire time the computer is running.
The Asus page should be formulated, to point out the difference between
those two scenarios. (That is why, when I do a power estimate for
someone, I sometimes do two calculations - one for spinup power, one
for operating power, just to see whether spinup power is high enough
to require a different power supply selection. The Asus approach is not
separating those considerations in a manageable way. I only do a
spinup calc, if a computer has four or more hard drives.)
There aren't too many power supply estimators I trust. There was one in
Japan, with a spread sheet style output, which did a good job, but the
web site was shut down. Some of the other sites available, estimate
double the proper amount of power. You'd be better off asking the
clerk at Staples what to buy
This is an archived copy of the Japanese site. The component database won't be
up to date, so newer hardware won't be represented in here. I like this approach,
because their data is visible, and you can examine it easily for sanity. This
is a model for someone seeking to do their own estimator. Most estimator sites
try to hide their math, so you can't use common sense to check the results.
http://web.archive.org/web/20040411032947/http://www.takaman.jp/psu_calc.html?english
HTH,
Paul
