The age-old shut down vs hibernate debate

[Pleonasm]

The argument against shutting down a computer or smartphone would be that while the components are having temporary rest, it excerts more stress to warm up cold components, rather than keeping them at the same level. However, this happens whenever an inactivate /computer/phone starts getting used anyway.My question is thus as follows: is there less stress on a system going from luke to warm, as opposed to cold to luke warm?

No matter how you slice or dice it, the components are, however minimally, in progress during hibernation. Surely from a long term perspective, shutting them off would prolong their lifespan? They have a certain amount of hours on them, and those are pushed further into the future with 8 hours complete shut off everyday.?

 

[rcgldr]

In some cases (computers), the only difference between hibernate and shutdown is that hibernate saves the machine state (including ram image) in a file before shutting down and on restart, restores the machine state from that file, otherwise it's the same as a shutdown. A "standby" or "standby with quickstart" or "sleep" feature does a partial shutdown.

 

[Pleonasm]

In some cases (computers), the only difference between hibernate and shutdown is that hibernate saves the machine state (including ram image) in a file before shutting down and on restart, restores the machine state from that file, otherwise it's the same as a shutdown. A "standby" or "standby with quickstart" feature does a partial shutdown.

It's not the same as a shutdown. A computer in hibernation still consume battery in a way which a shut down computer does not.

 

[anorlunda]

Since you posted in the materials forum, I expect it is not the computer science you are asking about.This us not my field, but I expect that the most rapid temperature changes resulting in the highest temperature gradients is what effects lifetime. I further expect that to happen when warming up.

 

[rcgldr]

It's not the same as a shutdown. A computer in hibernation still consume battery in a way which a shut down computer does not.

My desktop saves state in a file, sets a hibernated flag in a boot information file, and does a full shutdown. On powerup, the flag in the boot file causes the boot process to restore the saved stated from the file, otherwise it's the same as a normal powerup. Wiki article.

https://en.wikipedia.org/wiki/Hibernation_(computing)

There is a battery in the computer for at least the clock, and optionally to allow powerup from an external source, such as wake on lan. Wiki article:

https://en.wikipedia.org/wiki/Wake-on-LAN

As mentioned in my prior post, a "standby" or "sleep" feature will do a partial shutdown.

 

[Pleonasm]

My desktop saves state in a file, sets a hibernated flag in a boot information file, and does a full shutdown. On powerup, the flag in the boot file causes the boot process to restore the saved stated from the file, otherwise it's the same as a normal powerup. Wiki article.

https://en.wikipedia.org/wiki/Hibernation_(computing)

There is a battery in the computer for at least the clock, and optionally to allow powerup from an external source, such as wake on lan. Wiki article:

https://en.wikipedia.org/wiki/Wake-on-LAN

As mentioned in my prior post, a "standby" or "sleep" feature will do a partial shutdown.

How about the discrepency between switching a phone off and on, as opposed to simply turning the screen off/ on? Will the stress from starting the phone from a cold state even out the time won by having it shut off at night?

 

[rcgldr]

How about the discrepancy between switching a phone off and on, as opposed to simply turning the screen off/ on? Will the stress from starting the phone from a cold state even out the time won by having it shut off at night?

Smartphones run relatively cool when in an idle mode, turning off the screen, slowing down the clock rate for the cpu, ... , reducing the battery discharge rate, which translates into less heat produced, and cooler circuits. My impression is that most of the components are relatively cool unless doing some high performance activity, screen on, cpu running at faster rate, perhaps wifi running at max rate.

Similarly, but not as extreme, desktops will run at lower than maximum cpu and gpu clock rates when there isn't much load, reducing power consumption, and fan speed, since less heat is being generated.

 

[Pleonasm]

Smartphones run relatively cool when in an idle mode, turning off the screen, slowing down the clock rate for the cpu, ... , reducing the battery discharge rate, which translates into less heat produced, and cooler circuits. My impression is that most of the components are relatively cool unless doing some high performance activity, screen on, cpu running at faster rate, perhaps wifi running at max rate.

Similarly, but not as extreme, desktops will run at lower than maximum cpu and gpu clock rates when there isn't much load, reducing power consumption, and fan speed, since less heat is being generated.

but the circuits are still active. Any activation is surely faster deterioration than none at all.

 

[russ_watters]

The argument against shutting down a computer or smartphone would be that while the components are having temporary rest, it excerts more stress to warm up cold components, rather than keeping them at the same level. However, this happens whenever an inactivate /computer/phone starts getting used anyway.

My question is thus as follows: is there less stress on a system going from luke to warm, as opposed to cold to luke warm?

No matter how you slice or dice it, the components are, however minimally, in progress during hibernation. Surely from a long term perspective, shutting them off would prolong their lifespan? They have a certain amount of hours on them, and those are pushed further into the future with 8 hours complete shut off everyday.?

While I have seen people ask about this "debate" over the years, I've never actually seen the debate itself, and I'm not sure I believe it actually exists. Can you point to a reference where this "debate" is being argued, perhaps in scientific research?

And short even of the debate, I have never seen reference to computers having a lifespan based on degradation of their electrical components. Can you reference a reliable source indicating such an issue exists?

As far as I know, solid state devices such as computers and even more so cell phones do not "age" in an appreciable way, and so the "debate" is moot: it does not affect the lifespan noticeably, whether you shut down your computer/phone or let it go to sleep.

It's not the same as a shutdown. A computer in hibernation still consume battery in a way which a shut down computer does not.

No, that is not correct. You are confusing hibernation mode with sleep mode.

 

[rcgldr]

but the circuits are still active. Any activation is surely faster deterioration than none at all.

True, but try comparing time duration between standby and active mode on a modern smart phone, depending on power saving options, standby drain can be 10% or less than active drain on a battery.

 

[Pleonasm]

While I have seen people ask about this "debate" over the years, I've never actually seen the debate itself, and I'm not sure I believe it actually exists. Can you point to a reference where this "debate" is being argued, perhaps in scientific research?

And short even of the debate, I have never seen reference to computers having a lifespan based on degradation of their electrical components. Can you reference a reliable source indicating such an issue exists?

As far as I know, solid state devices such as computers and even more so cell phones do not "age" in an appreciable way, and so the "debate" is moot: it does not affect the lifespan noticeably, whether you shut down your computer/phone or let it go to sleep.

No, that is not correct. You are confusing hibernation mode with sleep mode.

Should I Turn My Computer Off at Night?

The issue isn’t as simple as you might think; there’s no single answer to this perennial question.

https://askleo.com/should-i-turn-my-computer-off-at-night/

Footnotes & references
1: Surprisingly, not quite true on some machines. Even when you turn the power completely off, there are often components that remain powered to some degree. Only pulling the power plug prevents this tiny trickle of electricity from being used if it is on your specific machine.

 

[elusiveshame]

Should I Turn My Computer Off at Night?

The issue isn’t as simple as you might think; there’s no single answer to this perennial question.

https://askleo.com/should-i-turn-my-computer-off-at-night/

Footnotes & references
1: Surprisingly, not quite true on some machines. Even when you turn the power completely off, there are often components that remain powered to some degree. Only pulling the power plug prevents this tiny trickle of electricity from being used if it is on your specific machine.

This isn't really an argument/debate - it's just some persons opinion posted on their blog. It's as relevant as someone's opinion in a forum post. The only concern you should potentially have is lowering your electric bill. You may have some components fail over time (hard drive, RAM, fans), but I've seen failures happen at the same rate on PC's that have been powered down as well as left on 24/7. I have a computer at home that hasn't been turned off in 8 years, except for when I lost power, and nothing has failed on it.

The footnote seems a bit silly, too. It shouldn't be a surprise that something plugged in is using a tiny trickle of electricity - just about every device does (coffee maker, VCR, TV, etc). That's how status LED's work to let you know if a device is powered on or not.

 

[Pleonasm]

I have a computer at home that hasn't been turned off in 8 years, except for when I lost power, and nothing has failed on it.

How does that refute the other side though? It's an all else equal argument.

 

[CWatters]

I've had several PC last 8+ years and others fail within 2 years. I've yet to detect any particular usage pattern responsible.

The only thing I would say is that electrolytic capacitors can have a short life if they are run hot. So in a hot climate or similar I would probably buy a better quality and over spec PSU and fit more fans to the case than I would otherwise. Make sure things like the graphics card is well cooled.

 

[mfb]

There are different possible failure modes, and it is not clear which one will make your device fail first.

• Things with a battery will see a degraded battery capacity over time, mainly coming from the number of charge cycles. Lowering power consumption will make the battery last longer.
• Thermal cycles can break things, although I'm not sure if that is ever relevant in consumer electronics. Keeping the device running and active will limit the number of thermal cycles.
• The risk to overheat is only present if the device is switched on, but if the device tends to overheat it will probably do that soon anyway.
• Single event effects from radiation can only occur when things are switched on. Keeping the device off reduces their probability. Overall you shouldn't have a problem from them in usual applications.
• Accumulation of radiation damage depends on the powering state, but different components will behave differently. This is not an issue in consumer electronics in usual places, however.

All these things become more important for specialized electronics, with intense ionizing radiation, with high voltage components and so on. Particle detectors prefer "on" or "off" depending on their individual conditions, sometimes even a mixture (e.g. HV off, LV on).

 

[elusiveshame]

How does that refute the other side though? It's an all else equal argument.

That was kind of my point :P

 

[berkeman]

Since you posted in the materials forum, I expect it is not the computer science you are asking about.

After a Mentor discussion, we'll move this thread to the General Physics forum for now. It does involve more than just materials or even EE.

 

[anorlunda]

This source cites the following three main causes of PC failure. I think it is notable that none of the three are failures in the semiconductors, and that thermal cycling is significant in all three.

1. Component Package Failures ... Mechanical failure of the package can be caused by a number of factors including thermal stress, chemical cleaners, and ultraviolet light.
2. Solder Joint and Contact Failures ... Thermal cycles are also a prime cause of solder joint failure, especially if the thermal expansion rates of the materials (component pin, solder, PCB trace coating, and PCB trace) are different.
   
3. PCB Failures ... The solder flux used in attaching all of the components to a PCB may remain on the surface of a PCB which will eat away and corrode any metal it comes in contact with. Solder flux is not the only corrosive material that often finds its way on to PCBs as some components may leak fluids that can become corrosive over time and several cleaning agents can have the same effect or leave a conductive residue which causes shorts on the board. Thermal cycling is another cause of PCB failures which can lead to delamination of the PCB and play a role in letting metal fibers grow in between the layers of a PCB.

 

[Pleonasm]

This source cites the following three main causes of PC failure. I think it is notable that none of the three are failures in the semiconductors, and that thermal cycling is significant in all three.

1. Component Package Failures ... Mechanical failure of the package can be caused by a number of factors including thermal stress, chemical cleaners, and ultraviolet light.
2. Solder Joint and Contact Failures ... Thermal cycles are also a prime cause of solder joint failure, especially if the thermal expansion rates of the materials (component pin, solder, PCB trace coating, and PCB trace) are different.
   
3. PCB Failures ... The solder flux used in attaching all of the components to a PCB may remain on the surface of a PCB which will eat away and corrode any metal it comes in contact with. Solder flux is not the only corrosive material that often finds its way on to PCBs as some components may leak fluids that can become corrosive over time and several cleaning agents can have the same effect or leave a conductive residue which causes shorts on the board. Thermal cycling is another cause of PCB failures which can lead to delamination of the PCB and play a role in letting metal fibers grow in between the layers of a PCB.

How about moderate, external impacts? Could this not contribute to weakened connections and overheating? When I gave a smartphone display a "black eye" it suddenly started suffering from constant overheating, from originally withstandning work load of any kind. It got at worst luke warm.. Suddenly it overheated from any prolonged usage and disabled features.

 

[Pleonasm]

I would very much like to know how impacts could affect heat dissipation? The construct is what it is no matter if a pixel is bad on the upper left side, say. This phones heat regulation went south once there was clear signs of display damage.

 

[elusiveshame]

How about moderate, external impacts? Could this not contribute to weakened connections and overheating? When I gave a smartphone display a "black eye" it suddenly started suffering from constant overheating, from originally withstandning work load of any kind. It got at worst luke warm.. Suddenly it overheated from any prolonged usage and disabled features.

You could definitely jostle something out of place that could cause ill-desired effects. If you loosen any heat shielding/heat sink/cause thermal paste to chip away, then overheating could definitely be a thing.

I also agree with what @mfb said about the capacitors. There's a certain age range from the 90's where caps are known to have gone bad within a few years (much shorter than they should have), so if you cheap out on your parts, the components that make up those parts could definitely degrade much faster, causing a whole array of issues.

 

[Klystron]

Measurements provided by data center hosts to resident clients may be relevant to collections of computers and related equipment.

• electrical power consumption measurements.
• temperature variations after accounting for external temperature changes, HVAC usage, client visits, etc.
• network statistics.
  
• error and alarm statistics.

The large sample sizes and collections of equipment under similar if not identical conditions provide a statistically significant measure.

The OP includes mobile units. Cell phones can be dropped or otherwise physically damaged while electronics in data centers are fixed to support structures. So, gather statistics on many cell phones and also look at data center installations as units with computers as components..

Examples: during the October 1989 Loma Prieta earthquake Bay Area data centers dropped shook and twisted. Ames research center located along the south bay kilometers north of the epicenter experienced "S" and "P" seismic waves lasting "many minutes". Some systems were then exposed to thermal shock as super-heated gases vented from the nearby blow-down wind tunnel vessels. While evacuating and searching the buildings, computing equipment looked devastated, covered in thick dust. Once personnel were allowed to return and with power restored most equipment came back up after cleaning and reinserting loose boards.

For more data I suggest comparing Ames replacement equipment cost figures for ~5 years before and after the 'quake with sister site at Langely. Many, if not most, computers were upgraded/replaced 1989-1993. Comparison to Langely and other data centers not affected by earthquakes could provide failure rates for 'components' subject to this event.

 

[russ_watters]

Should I Turn My Computer Off at Night?

The issue isn’t as simple as you might think; there’s no single answer to this perennial question.

https://askleo.com/should-i-turn-my-computer-off-at-night/

That is neither a reference to the debate - only a guy who someone told he heard about it from a friend - nor a reference documenting the problem actually exists. At this point, I'm inclined to believe this is just a myth, like the old myths we used to hear about fluorescent light bulbs.

How does that refute the other side though? It's an all else equal argument.

There's nothing here to refute: you haven't given a source either documenting or arguing that the issue exists. Only that people have *heard* a "debate" exists.

I do like the blogger's answer though: since there is no evidence this problem is real, you should ignore it and operate your PC based on how it best works for you with respect to ease of use and electrical usage.

 

[Pleonasm]

At this point, I'm inclined to believe this is just a myth, .

So what is your contention? What is the prevailing wisdom on this topic?

 

[Klystron]

Oh, I was trying to supply hard data comparing lots of cell phones and computers. [BTW I found a different method than data center aggregates.]

Professional opinion: keep electronic equipment vital to communications such as phones powered up with conservation parameters set to maximize duty cycle. Keep personal computers not vital to emergencies powered on or off according to your preferences. I leave my computers powered up but let the OS set hibernation and related states such as screen saver mode based on my parameters. Your choice.

[Agree that "this debate" insofar at it exists is folklore. Use actual data from groups of equipment.]

 

[russ_watters]

So what is your contention? What is the prevailing wisdom on this topic?

The blogger and I just told you: The prevailing wisdom is that the the issue you are asking about does not exist, so you should ignore the "debate".

 

[anorlunda]

There is a separate but closely related issue for laptop computers that should be mentioned.

When your laptop is shut down, the battery charging circuit is still active. Spontaneous fire in lithium ion batteries is a risk and heightened while charging. The risk is not just computer failure, but also a structure fire. For that reason, the recommendation for laptops is to remove the battery if you leave it plugged in all the time. The laptop will still function normally when plugged in without the battery (but it will stop on a momentary power failure).

Some office laptops are used as desktops, always plugged in and never intended to use as a portable device. Especially in those cases, the battery should be removed.

I confess that I don't follow that advice myself, but I do acknowledge the risk-convenience trade-off.

 

[russ_watters]

There is a separate but closely related issue for laptop computers that should be mentioned...

Another is that because computers tend to draw a little even when shutdown, taking the battery out of a laptop you don't intend to use for a while will make it live longer.

 

[Vanadium 50]

It's not the same as a shutdown. A computer in hibernation still consume battery in a way which a shut down computer does not.

That is simply not the case. A computer in hibernation reads a file containing the state before hibernation and resets that state. That's the only difference,

 

[anorlunda]

<<Moderator: This thread was originally posted in Materials and Chem E. I think it may be more appropriate here.>>

 

[jedishrfu]

There’s another failure mode in fluctuating AC power that can affect the lifetime of semiconductors and of course the talked about but yet to occur EMP in any significant way.

With respect to debate, I remember as a child the issue of Color TVs and the instant on feature where folks were worried that it would age the tube faster. I’m sure this carried forward into PCs as well. I do remember discussions among peers over the benefits of sleep vs hibernate vs shutdown modes. We would shutdown our PCs over the weekend and leave them running overnight. Now we leave them running all the time so that needed admin work and disk backups can run after hours.

 

[anorlunda]

I searched for a study comparing PCs which stay on always with those that power off every night. I couldn't find one. If there was such a study, it could provide the definitive answer.

There are now publicity campaigns trying to convince people to stop wasting electricity (and being more green) by powering down all devices that draw standby power overnight. That is more than PCs, entertainment center boxes are a major culprit. There are also EU and US regulations mandating reduced standby power consumption.

The situation today is much better than in 2000, yet we still see public estimates such as below. I am skeptical of that 10% number, but I have no hard data. Nevertheless, it is foreseeable that some lawmaker will one day forbid leaving personal computers on 24x7, and perhaps forbidding standby mode for all consumer devices.

Devices on standby consume electricity which must be paid for. The total energy consumed may be of the order of 10% of the electrical energy used by a typical household, as discussed below.

 

[Pleonasm]

I searched for a study comparing PCs which stay on always with those that power off every night. I couldn't find one. If there was such a study, it could provide the definitive answer.

There are now publicity campaigns trying to convince people to stop wasting electricity (and being more green) by powering down all devices that draw standby power overnight. That is more than PCs, entertainment center boxes are a major culprit. There are also EU and US regulations mandating reduced standby power consumption.

The situation today is much better than in 2000, yet we still see public estimates such as below. I am skeptical of that 10% number, but I have no hard data. Nevertheless, it is foreseeable that some lawmaker will one day forbid leaving personal computers on 24x7, and perhaps forbidding standby mode for all consumer devices.

What I find crazy is that some modern flat screen TVs no longer have an off button. Only standby mode is possible unless you pull out the plug.

 

[Pleonasm]

https://www.cnet.com/news/apple-genius-turn-off-your-phone-for-better-battery-life/

Why would battery longevity imrpove from nightly shutdowns? I lose 2% from just starting it up,... about the same that procentage lost over night if the phone is on airplane mode.

 

[Klystron]

[snip!] Why would battery longevity imrpove from nightly shutdowns? I lose 2% from just starting it up,... about the same that procentage lost over night if the phone is on airplane mode.

Interesting and quantifiable point. Begs the question: should chargeable devices be left plugged in?

My Moto-G phone instructions warn against over charging but are vague on details. In light of @anorlunda warning concerning overheated batteries, caution dictates plugging in the phone only when required.

 

[Pleonasm]

Interesting and quantifiable point. Begs the question: should chargeable devices be left plugged in?
.

It stops charging after 100% so it doesn't matter, if 100% charge is your goal.

 

[anorlunda]

https://www.cnet.com/news/apple-genius-turn-off-your-phone-for-better-battery-life/

Why would battery longevity imrpove from nightly shutdowns? I lose 2% from just starting it up,... about the same that procentage lost over night if the phone is on airplane mode.

Lithium-ion suffers from stress when exposed to heat, so does keeping a cell at a high charge voltage. A battery dwelling above 30°C (86°F) is considered elevated temperature and for most Li-ion a voltage above 4.10V/cell is deemed as high voltage. Exposing the battery to high temperature and dwelling in a full state-of-charge for an extended time can be more stressful than cycling. Table 3 demonstrates capacity loss as a function of temperature and SoC.

 

[essenmein]

I searched for a study comparing PCs which stay on always with those that power off every night. I couldn't find one. If there was such a study, it could provide the definitive answer.

There are now publicity campaigns trying to convince people to stop wasting electricity (and being more green) by powering down all devices that draw standby power overnight. That is more than PCs, entertainment center boxes are a major culprit. There are also EU and US regulations mandating reduced standby power consumption.

The situation today is much better than in 2000, yet we still see public estimates such as below. I am skeptical of that 10% number, but I have no hard data. Nevertheless, it is foreseeable that some lawmaker will one day forbid leaving personal computers on 24x7, and perhaps forbidding standby mode for all consumer devices.

I don't have the time to find it, but I think Google or someone had some interesting data.

My understanding of the question is more do you let it run 24/7 or shut down at night, google or facebook, one of them, used domestic computers for their server farms so had interesting data on MTBF for things like spinning drives, and I think the standard theory was less power cycles = more run time, but based on the statistical data they measured (from 1000's of PC's and their component failure rates) they didn't notice any effect (between power off at night and run 24/7).

Re original question re shut down vs hibernate, I doubt there is a lot of difference, both are basically "off", but if looking at keep running 24/7 vs some low power state (hybernate/shutdown/sleep) the thought was the life saved due to the "off" period would be less than the wear a power cycle/thermal cycle added, therefore its better to not turn it off.

Even though the data says there was not statistical diff between off at each night and run 24/7, I just feel better if my spin drives are running at a constant temp constant speed.

Note that from a greeny perspective, we heat the house at least half the year so at that point its parasitic heating that reduces the load on the furnace :P

 

[eq1]

As far as I know, solid state devices such as computers and even more so cell phones do not "age" in an appreciable way, and so the "debate" is moot: it does not affect the lifespan noticeably, whether you shut down your computer/phone or let it go to sleep.

This is not correct. See below.
https://semiengineering.com/transistor-aging-intensifies-10nm/

When hibernating a SOC has more of its die area biased [*] which means the time to dioxide breakdown or NBTI failure could be reduced, in theory, but the engineers should've assumed the part was biased 100% of the time in the lifetime model so whether or not the part is in shutdown it should still meet the stated lifetime goal making the point moot.

* Part of the die area (for example, the real-time clock circuit) is always biased. Unless the battery is physically disconnected of course.

 

[Rive]

Just two small points.

• maybe worth to look up 'BGA reball' and the mentioned failure reasons on the pages coming up as result. Temperature change does have an effect.
  
• worth checking the temperature profiles of the most frequently affected components (CPU and GPU). Usually both has more changes during load<>idle than idle<>OFF. Loads like games are especially bad.