| 1 |
On 7/18/07, Ryan Sims <rwsims@×××××.com> wrote: |
| 2 |
> On 7/18/07, Hendrik Boom <hendrik@××××××××××.com> wrote: |
| 3 |
> > On Sun, 03 Jun 2007 12:49:21 -0500, Dan Farrell wrote: |
| 4 |
> > |
| 5 |
> > > it takes just as much power to |
| 6 |
> > > spin up the drive as to keep it spinning for a few extra minutes. |
| 7 |
> > |
| 8 |
> > So ... spin it down after a few more minutes? |
| 9 |
> > |
| 10 |
> > -- hendrik |
| 11 |
> |
| 12 |
> No, only spin it down when the savings from the down cycle outweigh |
| 13 |
> the power cost of spinup+spindown (I don't know whether spindown uses |
| 14 |
> extra power, to "brake" the drive or anything). |
| 15 |
> |
| 16 |
> Say you have a drive that uses 1W/m (huge, but I'm being merciful to |
| 17 |
> my math skills) while in usage, and requires 5W to spinup. If you're |
| 18 |
> going to shut it down for 1m, you're looking at saving 1W and using |
| 19 |
> 5, net use of 4, when leaving it spinning would only use 1. However, |
| 20 |
> if it's going to be inactive for 30 min, you're using 5 and saving 30, |
| 21 |
> net savings of 25. |
| 22 |
> |
| 23 |
> -- |
| 24 |
> Ryan W Sims |
| 25 |
> -- |
| 26 |
> gentoo-user@g.o mailing list |
| 27 |
|
| 28 |
|
| 29 |
Ryan, |
| 30 |
You're certainly right that hard drives take more power to start |
| 31 |
up but I think the arbitrary values you used don't quite represent |
| 32 |
what really goes on. First though, let me help you with your units. |
| 33 |
Watts, a unit measuring power, is defined as energy per time period. A |
| 34 |
device that requires 5 watts and runs for 1 minute will use the same |
| 35 |
amount of energy as a 10 watt device running for 30 seconds. I think |
| 36 |
what you really meant to use was Joules, which measures energy. 1 |
| 37 |
joule per second is one watt. Now, as for the wattage values you |
| 38 |
supplied. |
| 39 |
A quick question posed to google lead me to |
| 40 |
http://www.digit-life.com/articles2/storage/hddpower.html where you |
| 41 |
can see a listing of power consumption for various hard drives (mostly |
| 42 |
models that would be used in servers, but they will do) when idle, |
| 43 |
under use, and most importantly at start up. Looking at the values, it |
| 44 |
seems that saying a drive uses 12W while active, 8W while idle, and |
| 45 |
30W at startup seem reasonable. I don't see anything indicating how |
| 46 |
long a drive takes to spin up, but I would assume it's something |
| 47 |
rather short. Let's say 5 seconds (which is probably longer than it |
| 48 |
actually takes). |
| 49 |
So here is your hard drive, happily powered up but idle, using 8 |
| 50 |
watts of power. Since it is idle, you might be wondering if it should |
| 51 |
be turned off to save power. Since it seems a drive uses 30 watts for |
| 52 |
5 seconds when powering up, this is 30x5 or 150 joules. At 8 watts, it |
| 53 |
will take 150/8 or 18.75 seconds to use 150 joules. Therefore, if this |
| 54 |
hard drive is going to be idle for more than 18.75 seconds it makes |
| 55 |
sense to shut it off. Of course real drives will almost certainly be |
| 56 |
different, but the point is it only would seem to take a few seconds |
| 57 |
of idle time before powering down makes sense. Also one could argue |
| 58 |
that this doesn't take into account the effects of wear and tear when |
| 59 |
stopping/starting drives, but I personally believe those effects are |
| 60 |
negligible. |
| 61 |
Finally, an interesting thing about hard drives is that when they |
| 62 |
are spinning down (at least when power has been unexpectedly cut off), |
| 63 |
the motor that spins the platters is used as a generator, taking the |
| 64 |
energy of the spinning drive to move the read/write heads to the |
| 65 |
parked position, so there is no power cost associated with powering |
| 66 |
down a drive. |
| 67 |
|
| 68 |
|
| 69 |
Julian |
| 70 |
-- |
| 71 |
gentoo-user@g.o mailing list |
Archives