Question the hotter the better

[theking2]

I love to keep my CPUs on 90 degrees. just below throttle. Why is this a problem?

 

[punkncat]

Why does this feel like being trolled?

Is there some real discussion you are looking for out of this, or just messing about?

 

[hotaru.hino]

Why is this a problem?

Electronics wear out faster when they run hotter.

 

[sizzling]

I love to keep my CPUs on 90 degrees. just below throttle. Why is this a problem?

What is it you love about doing this?

 

[kerberos_20]

room heater

 

[bignastyid]

room heater

Actually pulling more heat from the CPU(running it cooler) would make it a better room heater than letting it run hot.

 

[USAFRet]

Actually pulling more heat from the CPU(running it cooler) would make it a better room heater than letting it run hot.

CPU is producing the same amount of heat,l to be pushed into the room.
Just a difference on where the heat transfer happens.


The question for the OP is...."Under what workload is this 90C?"
If at idle, that is a problem.

 

[TerryLaze]

Electronics wear out faster when they run hotter.

That's not really true, as long as you stay within the rated range of operation it will wear out the same.
If you go above the save temp or constantly push it then it will wear out, the soonest, within the rated hours. 90 is safe even for ryzen CPUs that have a lower tolerance.
It's not "when they run hotter" but rather "when they run too hot"

I love to keep my CPUs on 90 degrees. just below throttle. Why is this a problem?

It's not a problem on its own, but it does make the CPU much less efficient you can save a good bunch of power by running it at a lower temps.

 

[kerberos_20]

there is also a thing,when something gets hot, it will stretch, and once it gets cold, it will withdraw, atleast on GPUs this is normal thing where you have to bake it in owen to make those cracked joints make happy again
so ya cpu should be ok, but mainboard may take some hits later on

 

[USAFRet]

where you have to bake it in owen

Please...can we make this old turd die?

Pretty please...?

 

[kerberos_20]

you can save a good bunch of power by running it at a lower temps

cpu will draw same power no matter temperature as long its not throttled
if youre talking about derating curve...well we are past the point where CPUs had stability issues after 65C
but technically if hes running it at high temperature (low fan speed), then he is saving power as fans do draw power, running them faster will eat more power

 

[hotaru.hino]

That's not really true, as long as you stay within the rated range of operation it will wear out the same.

Do you have anything saying that this is the case? Because I can find plenty of material saying that higher temperature/heat accelerates how quickly electronics become unreliable.

 

[TerryLaze]

cpu will draw same power no matter temperature as long its not throttled
if youre talking about derating curve...well we are past the point where CPUs had stability issues after 65C
but technically if hes running it at high temperature (low fan speed), then he is saving power as fans do draw power, running them faster will eat more power

No, the maximum limit of how much it can draw will not change, but the cooler the CPU the less watt it will use for the same amount of work.

View: https://youtu.be/FnwhYEUHw7k?t=487

Do you have anything saying that this is the case? Because I can find plenty of material saying that higher temperature/heat accelerates how quickly electronics become unreliable.

Every single piece of electronics comes with a spec sheet that states how many hours on average it is guaranteed to work at or below a specific temp.
If you can find any material that says that a lower temp within the safe range does a difference then yeah please show that.

 

[kerberos_20]

Do you have anything saying that this is the case? Because I can find plenty of material saying that higher temperature/heat accelerates how quickly electronics become unreliable.

that depends on how that electronic part was designed, with more heat, theres more resistance, which changes current flow
most electronic devices says its operating temperatures, take iphone as an example operating temperature 0C minimum, as a result people getting stuck screen during winter xD

 

[USAFRet]

And still, we do not know why or how the OP is maintaining the CPU at 90C.

Heavy load all the time, reducing fan speed...what?

 

[Tac 25]

not a problem if you live in Siberia.

Spoiler

feels like OP is just fooling around.

 

[hotaru.hino]

Every single piece of electronics comes with a spec sheet that states how many hours on average it is guaranteed to work at or below a specific temp.

I'm looking at a random electronics data sheet (https://www.st.com/resource/en/datasheet/stm32f205re.pdf) and I don't see anything resembling what you're talking about.

EDIT: Digging around some more, I did find some companies reporting MTBF for products which did report a temperature at which they conducted their testing to come up with that figure. So looking around a bit more:
https://www.vicorpower.com/documents/quality/Rel_MTBF.pdf

MTBF – mean time between failure (Mean Time Between ... - Delta

shopdelta.eu

MTBF Predictions :: Mean Time Between Failure Calculations

MTBF predictions based on MIL-HDBK-217 or commercial (Bellcore) standards, calculate your Mean Time Between Failure (MTBF) today.

www.halthass.co.nz

https://www.anandtech.com/show/2468/5

Industrial computer w. 28 year MTBF & -40 to +70C motherboard

All of these seem to suggest higher operating temperature leads to a shorter MTBF time.

 

[kerberos_20]

I'm looking at a random electronics data sheet (https://www.st.com/resource/en/datasheet/stm32f205re.pdf) and I don't see anything resembling what you're talking about.

page 71?

Table 14. General operating conditions (continued)

Symbol	Parameter	Conditions	Min	Max	Unit
T J	Junction temperature range	6 suffix version	–40	105	°C
T J	Junction temperature range	7 suffix version	–40	125	°C

 

[hotaru.hino]

page 71?
T J Junction temperature range
6 suffix version –40 105 °C
7 suffix version –40 125 °C

That's not saying how long the part will last at some given temperature or how temperature affects the longevity/reliability of the device.

EDIT: I'm aware there are some curves in the datasheet that plot accuracy of some part of the part against operating temperature, but also those point out that higher temperature reduces the accuracy of whatever part it's concerned with.

 

[hotaru.hino]

Another thing to add to this: https://qr.ae/pv4Whm

Let Vt=0.5 V, now when the temperature increases from 300 K to 350K, the leakage current increases by a factor of 15!

While it goes on to mention why this is bad (higher current = more heat = higher leakage current), current is also a factor in an equation used for predicting failure time due to electromigration: https://en.wikipedia.org/wiki/Black's_equation

That also explains why the CPU consumes less power when it's colder: there's less leakage current.

 

[kerberos_20]

That's not saying how long the part will last at some given temperature or how temperature affects the longevity/reliability of the device.

EDIT: I'm aware there are some curves in the datasheet that plot accuracy of some part of the part against operating temperature, but also those point out that higher temperature reduces the accuracy of whatever part it's concerned with.

ye, well as this being for something like mobile device...usualy battery kicks bucket first

well you remeber when bitcoin started? GPUs were dying so fast as they werent designed for that kind of perma work....well look at aging tests for pc electronics, few hours a day to get into 5years lifetime, office pc is a baseline btw xD
unless you get military grade equipment...which was used in gaming pcs aswell, later renamed to "gaming" which idk how valid is now due to gaming being RGB now

 

[geofelt]

I love to keep my CPUs on 90 degrees. just below throttle. Why is this a problem?

Why do you think this is a good thing?

 

[JWNoctis]

Why do you think this is a good thing?

So there'd be some interesting conversation like above, maybe. There's a classic xkcd for this.

More realistically, maybe the OP wouldn't have to (if not actually couldn't) pay for their own hardware, and there'd be a nice upgrade if, on the off-chance, their processor died an untimely thermal death this way.

 

[TerryLaze]

https://www.anandtech.com/show/2468/5
All of these seem to suggest higher operating temperature leads to a shorter MTBF time.

Which means, for a given target product lifetime, it is possible to derive with a certain degree of confidence - after accounting for all the worst-case end-of-life minimum reliability margins - a maximum rated operating temperature that produces no more than the acceptable number of product failures over a period of time.

You stay within operating temps you get the stated MTBF.
You stay below the safe Vcore you get the stated MTBF.
Anything more than that is 100% down to luck, you can have electronics sit on a shelf in a controlled environment and some will fail after so many years, and you can have office PCs that are filled solid with dust running for decades.

 

[hotaru.hino]

You stay within operating temps you get the stated MTBF.
You stay below the safe Vcore you get the stated MTBF.
Anything more than that is 100% down to luck, you can have electronics sit on a shelf in a controlled environment and some will fail after so many years, and you can have office PCs that are filled solid with dust running for decades.

And one of those sites I provided listed several products where they gave two MTBFs depending on the operating temperature you ran it at, with the lower MTBF rating given at a higher operating temperature. I'm still not seeing how my argument that higher temperature degrades electronics faster is invalidated.

Also MTBF isn't even a gaurantee. You're not gauranteed at least the time listed in the MTBF. It's a value based on statistical analysis. The most "guarantee" you have is based on however many samples it took to get to that value, when you take the mean time, it should get you the MTBF rating. And even then, as you say, luck can just give you the middle finger.