Sign in with
Sign up | Sign in
Your question

Could reaching high temps for a long time cause permanent damage?

Last response: in Graphics & Displays
Share
May 24, 2012 10:47:17 AM

Hi everyone

I have a question. Could laptop gpus and cpus get permanent damage when it reaches high temps for a period of time? I have a laptop with gt 425m, and it started to fail after just 1 year of use. the gpu hit 80 degrees celsius in every modern game i play for 1 year. The cpu and gpu has reached 87 degrees celsius due to cheap thermal paste.
Right now my gpu reaches a max of 68 c and cpu 72 c (Cleaned fan, Used high quality thermal paste, And polished heat sink).

And is there a solution to that? i saw a videos in youtube about rebailing and reflow stuff, will that solve all my gpu problems or make it even worse? thanks.
a b Î Nvidia
a b D Laptop
a c 176 à CPUs
May 24, 2012 10:58:48 AM

Willadin said:
Hi everyone

I have a question. Could laptop gpus and cpus get permanent damage when it reaches high temps for a period of time? I have a laptop with gt 425m, and it started to fail after just 1 year of use. the gpu hit 80 degrees celsius in every modern game i play for 1 year. The cpu and gpu has reached 87 degrees celsius due to cheap thermal paste.
Right now my gpu reaches a max of 68 c and cpu 72 c (Cleaned fan, Used high quality thermal paste, And polished heat sink).

And is there a solution to that? i saw a videos in youtube about rebailing and reflow stuff, will that solve all my gpu problems or make it even worse? thanks.

those videos were the old nvidia laptop gpu that had qa issues from nvidia.most all laptop mb are waved solider now if the laptop were to have a re flow issue it would have shown up when new. laptops due to there size unless the laptop is made for gaming and have lots of fans or big heat sink...a gamer can cook the gpu and cpu with high temps. if the laptop has a warranty i would send it in for repair. then i look into making a micro atx xbox size gaming pc..7770/560 4gigi of ram and i-320 or cheaper cpu. somthing that do some gaming without killing your laptop.
May 24, 2012 11:17:19 AM

aww man, should have got a nice desktop that time. Anyways my laptop does not have anymore warranty. The laptop randomly shuts down while web browsing or playing games when the gpu is in original clock speeds. My current solution is to underclock it, but it still crashes. Is there any solution to this other than taking it to the shop (they charge me nearly 50 usd just for them to look at the problem)
Related resources
a b Î Nvidia
a b D Laptop
a c 142 à CPUs
May 24, 2012 11:29:10 AM

Hi there Willadin,

All electronic components will experience a degraded lifespan or even immediate failure if they are run under extreme operating conditions for an extended period of time. However, extreme operating conditions are generally in excess of 90-100 degrees centigrade depending on the component. Many graphics cards can even withstand temperatures upwards of 110 centigrade (although it is by no means recommended to sustain this temperature). Component failure will manifest in many different ways and sometimes it can be hard to diagnose or even misdiagnosed.

As a rule of thumb any CPU operating temperature under 80 degrees is perfectly fine as is any GPU operating temperature under 90. Although I would personally recommend replacing thermal paste if temperatures are consistently higher than 10 degrees less than these values.

Your existing maximum temperatures of 68C and 72C are within the ideal operating temperatures.

If your laptop is no longer under warranty and you don't mind playing with a little bit of fire you can attempt to boost the video card's core voltage a little bit (not more than 2-3%) as this might compensate for some component degradation without drastically impacting temperatures.
May 24, 2012 11:40:25 AM

Thank you so much for your reply. Yes i have replaced the thermal paste and cleaned my fan. Although during my first thermal paste replacement i thought cheap thermal paste would work as good as high quality thermal paste. I did not know my cpu and gpu went up to 87 degrees celsius and played for 2 days. I got a high quality paste and max temp for gpu was 72c and cpu 75c. since my laptop is out of warranty i was thinking of lapping my heat sinks although high grit sand papers are hard to find. So i polished it with some metal polish (Works with copper) until it became so shiny i could see my reflection. Temps improved, gpu became 68c max and cpu 72c.

Oh and i was also thinking that voltages could be my problem. My gpu runs v0.83 while idling and v0.9 when active. Is that enough to run default clock speeds? oh and is my cpu voltages required? I dont know how to change voltages on a laptop though. there is no option in the bios. thanks.

a b Î Nvidia
a b D Laptop
a c 142 à CPUs
May 24, 2012 11:49:40 AM

Lapping components isn't necessary except to assist in extreme overclocking. The goal isn't to make it shiny, it's to make it flat. I'm sure it couldn't have hurt though.

Your GPU will automatically run at whatever voltage it is set to for each speed step as it was configured at the factory, same with your CPU. GPUs are generally not overclockable from within the system setup (although you can sometimes change the PCIe bus voltage) and certainly not on laptops. You will have to do some research on your particular laptop model and see if there's a software mechanism that works for it. Unfortunately there's no universally standard way for changing system parameters and most OEMs deliberately obscure them for marketing reasons or prevent them from being changed at all. If its a more recent model you will probably have a bit more luck.

If you can find a tool that works for your laptop you can try boosting the GPU core and CPU core by 0.02 volts each at the max and see what happens. This much will not damage either component
May 24, 2012 12:03:03 PM

Thanks for your reply. Yes i have been doing some research on how to change voltages for my laptop although i still cant change the voltage. Well i guess im fine but all i want is to play my games at base clock speeds like when i first got this laptop. only problem is i dont know the exact cause on why my laptop is randomly shutting down while web browsing or gaming. Weird thing is why the issue gets solved when i underclock my gpu?
a b Î Nvidia
a b D Laptop
a c 142 à CPUs
May 24, 2012 12:37:46 PM

When circuits age they undergo a process called electromigration. What happens is the mobile charge carriers (electrons and holes) slowly push and pull on the metal and metalloid atoms that form their conductive path. Over long periods of time or when subject to extreme conditions this can actually cause the atoms to become dislodged from the crystal lattice and thus degrades the conductive path. The gradual degradation of the conductive path impairs the path's ability to conduct electricity. However, since a constant amount of charge buildup within a minimum period of time is require for stable CMOS logic operation it stands to reason that as time goes on, gradually more voltage will be required to sustain operation of the circuit. Fortunately, chip manufacturers ship chips with standard supply voltages that are high enough to keep them going for several years past their warranty period. Even if the standard supply voltage becomes insufficient, additional voltage can always be supplied by the user in an attempt to return to stability.

The dangers of bumping up the voltage are two fold. First, it increases the push/pull that occurs on the conductors, thus increasing electromigration. Second, we still have to obey basic circuit principles which state that voltage, current, power and heat are all related. Increasing temperature also increases electromigration because it weakens the crystal lattice further. Combined, these two factors form the basis of why overclocking shortens the lifespan of integrated circuits. Electromigration has been shown to scale quadratically with voltage and cubically with temperature beyond a point, so it's important to be be careful.

The alternative to increasing the core voltage in order to reach a stable output within the required time is to actually increase the time allowed for the output to become stable. This is what underclocking does. The number of cycles per second is lowered, and thus the time that each cycle has for each stage in the entire logic network to become stable increases.
May 24, 2012 12:47:55 PM

Pinhedd said:
When circuits age they undergo a process called electromigration. What happens is the mobile charge carriers (electrons and holes) slowly push and pull on the metal and metalloid atoms that form their conductive path. Over long periods of time or when subject to extreme conditions this can actually cause the atoms to become dislodged from the crystal lattice and thus degrades the conductive path. The gradual degradation of the conductive path impairs the path's ability to conduct electricity. However, since a constant amount of charge buildup within a minimum period of time is require for stable CMOS logic operation it stands to reason that as time goes on, gradually more voltage will be required to sustain operation of the circuit. Fortunately, chip manufacturers ship chips with standard supply voltages that are high enough to keep them going for several years past their warranty period. Even if the standard supply voltage becomes insufficient, additional voltage can always be supplied by the user in an attempt to return to stability.

The dangers of bumping up the voltage are two fold. First, it increases the push/pull that occurs on the conductors, thus increasing electromigration. Second, we still have to obey basic circuit principles which state that voltage, current, power and heat are all related. Increasing temperature also increases electromigration because it weakens the crystal lattice further. Combined, these two factors form the basis of why overclocking shortens the lifespan of integrated circuits. Electromigration has been shown to scale quadratically with voltage and cubically with temperature beyond a point, so it's important to be be careful.

The alternative to increasing the core voltage in order to reach a stable output within the required time is to actually increase the time allowed for the output to become stable. This is what underclocking does. The number of cycles per second is lowered, and thus the time that each cycle has for each stage in the entire logic network to become stable increases.


I learnt more in a single post than I did in 5 weeks of uni...sadly :p  More interesting as well...
May 24, 2012 2:06:22 PM

Pinhedd said:
When circuits age they undergo a process called electromigration. What happens is the mobile charge carriers (electrons and holes) slowly push and pull on the metal and metalloid atoms that form their conductive path. Over long periods of time or when subject to extreme conditions this can actually cause the atoms to become dislodged from the crystal lattice and thus degrades the conductive path. The gradual degradation of the conductive path impairs the path's ability to conduct electricity. However, since a constant amount of charge buildup within a minimum period of time is require for stable CMOS logic operation it stands to reason that as time goes on, gradually more voltage will be required to sustain operation of the circuit. Fortunately, chip manufacturers ship chips with standard supply voltages that are high enough to keep them going for several years past their warranty period. Even if the standard supply voltage becomes insufficient, additional voltage can always be supplied by the user in an attempt to return to stability.

The dangers of bumping up the voltage are two fold. First, it increases the push/pull that occurs on the conductors, thus increasing electromigration. Second, we still have to obey basic circuit principles which state that voltage, current, power and heat are all related. Increasing temperature also increases electromigration because it weakens the crystal lattice further. Combined, these two factors form the basis of why overclocking shortens the lifespan of integrated circuits. Electromigration has been shown to scale quadratically with voltage and cubically with temperature beyond a point, so it's important to be be careful.

The alternative to increasing the core voltage in order to reach a stable output within the required time is to actually increase the time allowed for the output to become stable. This is what underclocking does. The number of cycles per second is lowered, and thus the time that each cycle has for each stage in the entire logic network to become stable increases.


Wow, never knew that. So it is possible that my gpu is undergoing electromigration? mmmm so i only have 2 options. Underclock the gpu or increase voltages. Ill try looking around for an easy guide to change gpu voltages for laptop computers. Thank you so much for sharing such information, i wish to learn more about this.
a b Î Nvidia
a b D Laptop
a c 142 à CPUs
May 24, 2012 8:47:12 PM

Yeah, the fact that it works well at low frequencies but not at high frequencies despite not being excessively hot anymore suggest that might be the case
May 25, 2012 2:05:00 AM

mmm since now i am more aware of my temperatures, I guess adding a few volts would be fine. Im still looking for a guide. Although my gpu is somewhat stuck together with the motherboard (something like you can remove it). I have seen a few people saying if i want to change voltage, i would have to modify the hardware.
!