From: Nuutti Kotivuori <naked_at_iki.fi>
Date: Wed, 04 Dec 2002 20:01:05 +0200

I did a few tests with the hard drive power-off toggle and this is what
I came up with. My hard drive is: FUJITSU MHN2200AT.

* Current

These tests were made with bad equipment under bad circumstances - I
can replicate the results easily, so they should have an internal
consistency, but do not expect them to correlate to anybody else's
test results :-) The unit is A, but it's not really relevant since the
meter shows something way different.

 Playing, backlight off, HD off: 0.07
 Playing, backlight on, HD off: 0.11
 Playing, backlight off, HD sleep: 0.08
 Playing, backlight on, HD sleep: 0.13
 
 (all the rest are backlight off
 Booting up : 0.46
 HD spun up and reading :~0.50
 Spinning up peak power : 0.64
 HD just before shutting down : 0.21

So, there is a difference between HD sleep and HD off. And that was
basically what I wanted to know.

* Timing

Then I made some timing tests between hard drive power-off and
sleep. Some of these are taken by hand, so they are not too reliable.

** Sleep

In sleep mode, when ever the unit needs to read more, the hard drive
led will light up, the drive will spin up, the led will flash a bit,
and then go dark and the drive spins down again.

 Spin-up time given by firmware : constant 20ms

There's gotta be a bug of some sort there, I could never see anything
else in the timing by the firmware.

 Total time taken by one buffer fill : 3.1 s

** Power-off

In power-off mode, the behaviour is a lot different. First the hard
drive led will light up, then the drive will spin up - but it will
take around 5 seconds for the drive to spin up. Then the led goes dark
again for about a second. And after that it lights up again for a
couple seconds or less.

 Spin-up time given by firmware (1) : 5100 ms
 Spin-up time given by firmware (2) : 1900 ms

The first spin up time will show in the space between when the led
goes dark - and the second time will show after the second read is
finished. Then on to manual timings.

 Time taken by first spin-up : 5.3 s
 Delay between spin-ups : 2.2 s
 Time taken by second spin-up : 1.4 s
 Total time taken by one buffer fill : 8.9 s

That last total time actually is taken from a separate timing, it just
happens to hit exactly the same values.

* Problems

With the power-off enabled, all my 256-kbit MP3's were breaking
up. Sometimes it was just a pause in playing, sometimes the sound
stopped altogether and would not come back until after another
buffer-fill.

I had to increase the low-water mark considerably, to take into
account the long spin-up times. That made performance relatively good,
but there was one obvious difference. It might have been for the still
inadequate low-water mark, but whenever I moved around with my player,
the loading started to break up. When I was walking, I hardly ever
heard any sound even. So it could be that the power-off spin-up is
more easily disturbed by movement.

* Conclusions

I'm not sure what to make of all this, results are weird. But atleast
right now, power-off is not for me. Because of the long spin-up time,
the low-water mark problems and the relatively small power savings -
it doesn't seem worth it.

Hopefully this helps some.

-- Naked
Received on 2002-12-04