> I'm sorry to be a fact Nazi in this, but I think that if we're talking
> about saving the world from predicted environmental disaster based on
> scientific research, then we should use more accurate statistics.
I'm speaking at a higher level than specifics. The fact is we're not
going to get to the exact figures, and even if we could find perfectly
precise figures, I can't see them modifying my point, unless they turn
out to actually contradict me. They would only put a more concrete
figure on the magnitude of the benefits of developing more efficient
computing practices.
Unless it turns out that computers are in fact feeding power into the
electrical grid, there's really no way to argue against coming up with
strategies to deliver computing with less power.
> I found the "Energy efficiency" section of this article interesting:
>
>
http://en.wikipedia.org/wiki/Computer_power_supply
>
> And this:
>
>
http://en.wikipedia.org/wiki/Energy_efficiency
>
> Computers are much more efficient than I expected. 93% efficiency is
> pretty impressive. So why are we making such a big deal about it? It
> seems that as long as we use our computers for useful things while they
> are on (like writing this email), its not really all that wasteful.
a) 93%-efficient power supplies are available, but they are generally
70-75% at peak according to your article, though this figure only
applies at 50-75% load.
Anything less than or more than that drops this figure, more significant
on the low-load side than the high-load side.
So an idling computer is in fact the least efficient from the
mathematically perspective as well as the fact that it's basically a
power-sucking paperweight.
b) However, let's pretend computers are 100% power efficient.
In this magical world where the power supply is basically just
superconducting power from the grid into the box, it's still more
efficient to write your e-mail on a box that draws less than 10W than
one that must draw 60W if it's staring at the wall (though honestly, I'd
be surprised if an idle computer only drew 60W...I'll have to borrow a
kill-a-watt from work to measure this stuff). You're spending less
energy to get the exact same result.
What is your return on investment for that extra wattage? Wobbly
windows? ;-)
> So in this way, servers are generally not wasting much energy. Its
> obvious that people understand this though because the desktop is where
> all the power saving technologies are like turning off the monitor, hard
> drive sleep, etc.
Servers are the most efficient part of our computer use habits,
depending on how you structure it. This is why my ideal includes big
servers back-ends with the small workstations front-ends.
> Ok, I've rambled on long enough. Scott Blaydes could tell you about
> his dream of having a data center that is solar powered though. :-)
It's definitely doable, and economically advantageous in the long term.
He'll need to figure out how to make enough money with it to pay for the
solar panels in the short-middle term.
He'll also need to carefully calculate the "bang for his buck" of each
power-using component in his data center, as I've been talking about.
He'll want to make sure that every Watt he's spending (and thus every
square inch of expensive solar paneling he needs to have up) is bringing
him more money than it's costing him.
His dream highlights my very points.
I started thinking about this stuff as I considered the possibility of
building a completely power-independent home (and solar power seems the
way to go). I'm a total computer geek, and with the number of computers
I have now, they're a major chunk of my energy needs.
*yawn*
I'm going to bed, ya'll, have a great night!
Simón
P.S. Incidentally, I don't intend to say our computers today need to no
longer be used, throwing them away would have equally if not more
disturbing ecological consequences. I'm simply saying tomorrows
computers can be compacts, motorcycles, bicyles, with the same power as
today's SUVs.
_______________________________________________
BLUG mailing list
BLUG@linuxfan.com
http://mailman.cs.indiana.edu/mailman/listinfo/blug
No comments:
Post a Comment