Low temps for longevity?

[cisto1999]

Hi,

 

i am wondering if keeping GPU (or CPU) temps under a certain point does really something for longevity of the product? I mean... watercooling looks cool and it may help doing higher stable OC... but does it also something for longevity? GPU and CPUs will start throttle so that they won`t kill themselves or harm other components. So my question or point of discussion is: Is it ok for you if your GPU runs at 70-85 degrees when playing games or doing some rendering or do you have the fear that you might decrease the life-span of your product by doing so?

I found myself in the position that i am spending a lot of money for the best cooling possible in concern of keeping the product in the best shape possible for a long time even if it might upgrade anyway after 2 or 3 years ... but then i could have bought a better product in the first place if i would have spent less money for cooling. Is the concern for temps just part of marketing (beside the little performance-boost you get with better ccoling)?

[Velcade]

I wouldn't worry too much about your temps as long as they're below 80^oC on the CPU and 90^oC on the GPU

 

 

[TimonL]

@Velcade under what circumstances should you put your GPU under such load that it goes beyond 90° Celcius?

[Velcade]

4 minutes ago, TimonL said:

@Velcade under what circumstances should you put your GPU under such load that it goes beyond 90° Celcius?

You're GPU shouldn't be getting that warm.  If it is something else is going on... poor air flow, etc

[Unter Dog]

I'd generally worry about your electrolytic caps, as at higher temperature you can generally say 10 C lower doubles lifespan.  I'd assume you'd get an acceptable lifespan out of your silicon and caps if you are running 85C max and 75-80C normally with a normal consumer load cycle. Note a year is 8760 hours.  Note caps are generally rated in hours at their max ripple current at a specific temp.  https://www.xppower.com/Blog/Electrolytic-Capacitor-Lifetime-in-Power-Supplies  (there are a ton of articles online like this i just grabbed one quick).  Electrolytic caps many times are the weak link in a system.  

 

I design high reliability electronics for harsh environments and for most components higher temperature lowers life (things like NAND flash have caveats) but at what point does it not matter anymore which can only be answered by you. 

 

Electomigration is a common failure mode https://www.sciencedirect.com/topics/materials-science/electromigration  for silicon for time at temp.  Read up if you are interested.  

 

One also thing to note is the rate at which things change temp also affects reliability as coefficient of thermal expansion differences can cause long term failures like solder joint cracks and internal part delamination as stress/strain occur when parts expand/contract at different rates.  

 

 

[LogicalDrm]

Most parts already have 8+ year lifetime when used in normal operations. Most will change parts long before they die. Which is also why getting used stuff is good business. Running components near or over the danger limits, so over or at what manufacturer target with their own cooling, then there is degration. It will mean that eventually you need to up the voltages to keep same clock speeds, or that they will run at higher temps. But it would take extreme temps to actually kill stuff.

 

So running things cool gives more placebo than realistic benefits. Not saying that you shouldn't. But no need to bust money on cooling if you don't feel the need to.

[Taja]

Just some personal experience:

When I was a noob, I got a pc with a I7 2600. That was probably before the 3rd gen came, so pay attention to its lifespam.

 

I used it with the stock cooler and paste for maybe 4 years? It was a really long time, but anyway, one day the pc started shutting down. I was a bit more tech literate at the time, so I figured it out: it was shutting down because of THERMALS. Yep, it was getting 105°C hot, so I changed the cooler and the problem was solved... I kept it for 2 more years after that (I think) so the chip is 6+ years old, and the IHS is not making proper contact with the silicon (thermal dilation) but with a hyper t4 cooler it manages temps lower than 85°C. 

 

So yeah, the chip will become obsolete before he dies because of heat.

[Sentile]

Ok. Streight Answer. Yes , it will increase the longgevity  but Even Cheap Products Are Rated For Many Years At Normal Temps So You Shouldn't Worry Too Mutch. UNLESS Overcloking , when Overcloking Try To Keep The Lowest Temps You Can (Reasonably)

[Mira Yurizaki]

The rule of thumb I've heard is that for ever 10C higher the part runs for, its life expectancy cuts in half.

 

Except here's the problem: we don't know what the life expectancy is supposed to be in the first place. It could be 3 years at 70C, it could be 3 years at 40C.

[bowrilla]

Most chips, capacitors and so on that are used in circuits in which they will become hot are usually rated for 105°C. If you keep your parts within stock specs, the parts will usually last out beyond the point of obsolescence. There's no guarantee of course, some parts may fail sooner but then some will last long beyond that point.

 

If you go beyond specs (overclocking) degradation is a different story. You can have a stable overclock within commonly agreed safe margins regarding voltage and even temps, but at some point it might suddenly just become unstable due to degradation. You'll have to up the voltage at some point, which adds heat and the degradation process will accelerate. 

 

GPUs are very temperature sensitive/adaptive. The internal logic will adjust boost clocks according to temps until the point at which it meets the power limit. If you really want to lower your temps on the GPU you'll need quite a lot of cooling power to compensate for this boost logic, because the GPU will boost higher resulting in more heat, which can just compensate for slightly better cooling efforts.

 

I wouldn't bother much with GPU temps between 70°C and 80°C if it's a high end card. Even 90°C isn't that rare on the GPU if it's very powerful and boosting high enough. 

 

CPUs are way less sensitive generally speaking. The boost logic is simpler especially on current Intel chips. Ryzen is a bit more complex in regards to overclocking itself. You won't benefit from lowering temps by 10°C though once the max boost/self overclock is reached. There won't be a difference between your CPU running under full load at 40°C or 50°C. 

 

Why do have people "oversized" cooling solutions? Because this way everything can run more quiet.

 

Rule of thumb: CPUs should stay below 85°C at peak loads, GPUs (as long as they boost the way they are intended to) should stay around the same max temp. 

 

You'll have a hard time killing hardware these days by overheating it. All major components usually have protection circuits that will either throttle or shut down.