Shall we hide the GUI for Hibernate in Natty?

[Chow Loong Jin]

On Tuesday 22,February,2011 11:03 PM, Phillip Susi wrote:
> On 2/22/2011 6:22 AM, Chow Loong Jin wrote:
>> Sounds just about as disastrous as laptops running out of battery during a
>> suspend, actually. Which is not as serious as running out of battery while
>> attempting to hibernate, which seems to be occurring much more often than the
>> case of running out of battery while suspended. On the other hand, like Marco
>> mentioned, TuxOnIce is likely to be able to allow more laptops to safely
>> hibernate on low battery as it finishes a lot faster.
> 
> I would think the likelyhood would be the other way around.  If the
> battery gas gauge is properly calibrated, you should be hibernating with
> well over 30 seconds of power left and so not running out of power.  On
> the other hand, people frequently close the lid and leave the laptop
> there for hours or days, not even thinking that the battery is still
> running down.

My previously mentioned point still stands though. Many people close the lid of
the laptop, stick it into a bag, and start walking. Considering the I/O
intensive nature of the hibernation process, that practice can damage the hard
disk pretty much.

> 
> If you are running out of power before the hibernate finishes, then one
> of two things is wrong:
> 
> 1)  Your gas gauge is not properly calibrated and is reporting that you
> have more remaining power than you actually do
> 
> 2)  Your "dead battery" threshold where you go to hibernate is set too low
> 
> Raising that threshold will fix the problem.
> 
>> Actually no, my laptop battery's full charge state can last for an hour, but
>> even after showing 0%, it can stay in suspend-to-RAM mode for more than half a
>> day. I haven't tried letting it run longer though.
> 
> Then your gas gauge is wrong and you either need to recalibrate it or
> get a new battery.  From the laptops that I have seen, it seems normal
> to drain about 1% of capacity per hour in suspend ( give or take ), so
> if it lasts half a day, then that 0% reading is quite a ways off.
> Staying at 100% after an hour of use is another sure sign that the gauge
> is broken.

The gauge isn't broken. Well, not completely, anyway. The only issue is that it
jumps from 5% directly to 0% without passing 4%-1%. But on the other hand, while
there's enough power to last for half a day without running out of battery,
there isn't always enough power to finish hibernation before it keels over and
dies (starting from 0%).

Therefore, I'm assuming that the amount of power needed to complete a full
hibernation run can sometimes exceed the amount of power needed to maintain a
suspended-to-RAM state for half a day. But like I said, this is mostly due to
the current hibernation implementation in Linux taking too long to stuff the RAM
into the swap. A quicker, more reliable implementation would do the trick.

Another thing is that it's pretty typical to have 4G or even 8G of RAM these
days, and the time taken to resume from a hibernated state is more or less
proportional to the amount of RAM in use, i.e. it can take anywhere between a
minute and five, or even more than that. My Ubuntu installation boots in 2
minutes, and that's slow by normal standards. I think there's something
seriously wrong when you resume from hibernation slower than a cold boot.

 ***

But anyway, back on topic: I don't believe we should get rid of the hibernation
option. Suspend and hibernate have very important differences, and both of them
have their own fair share of use-cases.

-- 
Kind regards,
Loong Jin

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 262 bytes
Desc: OpenPGP digital signature
URL: <https://lists.ubuntu.com/archives/ubuntu-devel/attachments/20110222/7766b95c/attachment.pgp>