Sleep vs. Shutdown

On Thu, 01 Sep 2016 09:47:26 -0400, Spalls Hurgenson
wrote:

Personally, I think that - for average PCs at least - you are more
likely to replace the device because of its obsolescence long before
any damage (from restarts or keeping it running) accrues to the point
of noticability.

I agree with that. Both sides have their pros and cons so there's no clear
winner that everyone should adopt. I leave 3 machines on 24/7 because I use
them frequently, but other machines that are used much less often are
normally powered down.

I tend to shut down if I know I'm not going to use
the computer for some hours (overnight) and otherwise just lock it
(running but with a password) or let it sleep. I schedule a full
shutdown/restart of all my Windows computers once a week though
because, well, it's Windows ;-)

I used to try quite hard to wrap up every loose end so that I could reboot
my laptop every month, but that has slowly crept up to 6 weeks and then 8
weeks and then beyond 8 weeks. Earlier this week, I came back to the laptop
to find that it had rebooted on its own, which was a first for this machine,
so now I figure 9-10 weeks is about the limit (with only one data point so
far) to how far I can go between reboots, given the way that I use this
machine.

My desktop machine, which is used in a server role, gets rebooted when the
power goes out, which is once or twice a year. If I were to put that PC on a
UPS, I suppose it would just run 'forever', or until I need to make some
kind of hardware change/upgrade.

I remember having to boot 95/98 daily to keep them stable, but those days
seem to be long gone. XP was the first WinOS where I started going a week or
three between reboots, and even then most reboots were necessitated by
installing some kind of Win update.

--

Char Jackson

Char Jackson wrote:
On Thu, 01 Sep 2016 09:47:26 -0400, Spalls Hurgenson
wrote:

Personally, I think that - for average PCs at least - you are more
likely to replace the device because of its obsolescence long before
any damage (from restarts or keeping it running) accrues to the point
of noticability.

I agree with that. Both sides have their pros and cons so there's no clear
winner that everyone should adopt. I leave 3 machines on 24/7 because I use
them frequently, but other machines that are used much less often are
normally powered down.

I think there is a clear winner. With the
right component selection for your computer
build, shutting it down is the right answer.

80+ power supply - Active PFC functions as inrush limiter
- reduces stress on main cap
(older designs used a PTC resistor for inrush
limiting, with designs easily drawing 40 amps
for a cycle or two, at 120V. That's what makes
the lights blink in your computer room)

SSD - hibernation with a small set of applications open, only
uses a couple gigabytes of writes. Your SSD has a 150TBW
rating. Even if an average day has 20GB of writes (18GB
for web surfing, 2GB for hibernate), you have a
150000/20 = 7500 day rating.

Rotating hard drives are rated for 300,000 head load/unload cycles.
Plenty for shutdown and startup.

There is no question, in a powered state, it is ready
immediately for usage. So there is the convenience factor.
But in terms of hardware issues, I think with modern hardware
installed, there's no reason to leave the power on.

Paul

On Thu, 01 Sep 2016 13:46:19 -0400, Paul wrote:

Char Jackson wrote:
On Thu, 01 Sep 2016 09:47:26 -0400, Spalls Hurgenson
wrote:

Personally, I think that - for average PCs at least - you are more
likely to replace the device because of its obsolescence long before
any damage (from restarts or keeping it running) accrues to the point
of noticability.

I agree with that. Both sides have their pros and cons so there's no clear
winner that everyone should adopt. I leave 3 machines on 24/7 because I use
them frequently, but other machines that are used much less often are
normally powered down.

I think there is a clear winner. With the
right component selection for your computer
build, shutting it down is the right answer.

80+ power supply - Active PFC functions as inrush limiter
- reduces stress on main cap
(older designs used a PTC resistor for inrush
limiting, with designs easily drawing 40 amps
for a cycle or two, at 120V. That's what makes
the lights blink in your computer room)

SSD - hibernation with a small set of applications open, only
uses a couple gigabytes of writes. Your SSD has a 150TBW
rating. Even if an average day has 20GB of writes (18GB
for web surfing, 2GB for hibernate), you have a
150000/20 = 7500 day rating.

Rotating hard drives are rated for 300,000 head load/unload cycles.
Plenty for shutdown and startup.

There is no question, in a powered state, it is ready
immediately for usage. So there is the convenience factor.
But in terms of hardware issues, I think with modern hardware
installed, there's no reason to leave the power on.

With all due respect, Paul, I think you just made a case for "no clear
winner". If you use a machine regularly, leave it on. If you use it
infrequently, turn it off. With that as guidance, there's still a gray area
between them that's a mile wide, so no clear winner at all.

--

Char Jackson

Paul wrote:

....
There is no question, in a powered state, it is ready
immediately for usage. So there is the convenience factor.
But in terms of hardware issues, I think with modern hardware
installed, there's no reason to leave the power on.

....plus there's the energy saving...

In message , Spalls
Hurgenson writes:
[]
You are correct. I was thinking of the "hybrid shutdown" that became
the default shutdown for Win8+. This is essentially a hibernation mode
but - assuming your hardware supports it - allows Windows to bypass
the hardware power-up tests. That's why Win8+ boots so much faster
than Win7.
[]
What are there "hardware power-up tests"? How long _do_ they take?
(Unless you've got a lightning-boot, I can't see them making _that_ much
difference to boot time - they certainly wouldn't to me, I don't _think_
- but then some modern systems _do_ boot very fast.)
--
J. P. Gilliver. UMRA: 1960/1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

Madness takes its toll. Please have exact change
[via Penny Mayes )]

In message , masonc
writes:
On Tue, 30 Aug 2016 16:02:54 -0400, SteveGG
wrote:

What's the relative merits of these states for an inactive PC ?

How inactive - I mean, how often do you turn it on (or rather come out
of the inactive state)?

Sleep appears to have some function of the CPU active
but Shutdown is more thorough and involved. Opinions ?

Especially, what is best for the life of the computer,
especially the moving parts?

By "moving parts", I assume you mean intentionally-moving, which I can
only think of as meaning disc drives and fans, rather than parts flexing
as they warm up and cool down (and the - mostly infinitesimal, but
repeated many times - movements of transformer cores, both in normal
operation at their switching frequency, and occasionally larger as the
load changes significantly).

Disc drives have two things that move - the platters rotate, and the
heads move in and out (including moving out altogether - or to their
park zone - when told to do so).

For any recent PC, I get the impression that both fans and disc drives
stop altogether in sleep mode, so there's no difference. It is possible
that some systems, especially in hot environments, might power up the
fans very occasionally when in sleep, though IMO if that's happening you
need to look at your arrangements anyway.

As for powering down versus hibernate, sleep, or leaving fully on in
general, I think Paul's right, for modern equipment you're far more
likely to replace it due to obsolescence rather than failure.

Paul, Spalls, and Char have covered other aspects: basically, powering
up (from cold or one of the rest states) puts _some_ stress on some
components (current surges and temperature fluctuations [which flex
things like connections, especially those inside components]), leaving
running all the time puts other stresses, probably mainly on fans and
drives but also hot components like the processor and graphics. A
complete shutdown causes the macvhine to "run better" (after startup of
course!), at the expense of the time you have to spend waiting for it to
boot. The various shutdown modes (including off completely) _do_ use
less energy, to two levels if your habitat is cooled anyway (when on
they both use more energy themselves, and produce heat thus making your
aircon have to work harder); conversely, if your home is being heated,
the heat they produce means your heating has to work _less_ hard;
however, in both cases, the effect is so minuscule as to probably be
insignificant, for those of us fortunate enough to live in the western
world anyway. (Do you go round unplugging TVs, hifis, 'phone chargers,
your home heating controls such as the timeclock, ...)
--
J. P. Gilliver. UMRA: 1960/1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

Madness takes its toll. Please have exact change
[via Penny Mayes )]

Insignificant !

Figure it out. How long to use only 1 KWH ?

J. P. Gilliver (John) wrote:
In message , Spalls
Hurgenson writes:
[]
You are correct. I was thinking of the "hybrid shutdown" that became
the default shutdown for Win8+. This is essentially a hibernation mode
but - assuming your hardware supports it - allows Windows to bypass
the hardware power-up tests. That's why Win8+ boots so much faster
than Win7.
[]
What are there "hardware power-up tests"?

When PNP OS is set to No, the BIOS plans the
memory map, and sets the address decoding according
to all the equipment installed in the computer. The
chipset has to be programmed so every piece of hardware
lives at a unique address. And the memory controller
has to be set according to the size of the RAM modules
and so on.

The BIOS can do a quick RAM Test.

The BIOS can zero the RAM before the OS gets it.

The BIOS waits up to 35 seconds for disks to spin
up and obtains the text ID string from them. You
can see this info in the popup boot menu. On
computers where the "monitor SMART" setting is
enabled, the BIOS reads the SMART on each drive,
and issues a warning if any thresholds have been
surpassed (too many reallocated sectors).

The BIOS does not do a fully structured test
like that $50 box of software you bought at the
computer store.

Some computers have a switch setting that triggers
the manufacturer test. That tests the response of the
keyboard, and does all sorts of tests. Such a test could
take 20 minutes. (Sun Computers was the best at doing
that sort of test. The tests actually did something and
weren't pure window dressing.) You don't want the BIOS
implementing such a feature for consumers ("it would only
confuse them"). Everything the BIOS does is "fast",
relatively speaking. A 20 minute test, that should
be saved for that $50 box of software while the OS
is running.

This is one of the reasons that a decent RAM Test is
gradually disappearing from BIOS designs. One way or
another, the BIOS is just too fast, to actually
be doing something like that any more. If any traditional
behavior doesn't scale well in new computers, it's just
tossed on the side of the road. I'm really surprised
that "PNP OS = No" is the default, as I'm sure the OS
people would love to do all the BIOS work themselves :-)

The Insyde BIOS in my Acer laptop starts in only a couple
seconds, and it's not even clear whether it waits for anything.
Insyde is a relative newcomer to writing BIOS for computers,
so they're not saddled with any "traditions". But they do
still have to handle the default "PNP OS = No" thing, and
program the chipset properly before handoff. That would
include passing hardware-specific ACPI tables to the OS.
(Like the SLIC table that tells the OS it's an Acer,
and it's OK to boot an Acer-stamped royalty OS.)

*******

In the above description, the things done is a function
of ACPI state. The above might be appropriate for restoration
from S5. But the activities done in S3 or S4 could be different.
And I'm not going to do all the combinations, because I'll
only get it wrong. It's not even clear who is responsible
in the other ACPI states, for setting things up. (Recorded
in hiberfil.sys, rediscovered by BIOS ???)

Paul

J. P. Gilliver (John) wrote:

Paul, Spalls, and Char have covered other aspects: basically, powering
up (from cold or one of the rest states) puts _some_ stress on some
components

This is becoming less and less with time.

I have no concerns about turning my Test Computer
on and off multiple times a day.

S5 Soft Off for example, is not stressful.

And flipping the switch on the back is safe, as long
as you wait at least 30 seconds from switch off to
switch on again.

Do not rapidly toggle the AC switch on the back. A guy
in one of the newsgroups, blew an ATX supply by
"toggling the switch 50 times". Well, somebody had
to test that...

*******

I've had ATX power supplies fail here, but upon opening
them, you can see the "leaking cap" problem. And that's
just bad chemistry at the factory (pH is wrong for
one thing). The full story is covered here of how
it happened. The chemical will eat through the aluminum
outer shell, even with zero stress applied (supply
sitting in storage room, unpowered). On my Antec, it
took less than two years in storage for that to happen.
I had the orange goo and everything, just like the
picture here. The four +5V caps were bad. No amount
of "babying it" would have saved it. It would have
been dead in two years anyway.

https://en.wikipedia.org/wiki/Capacitor_plague

Paul