[SOLVED]Nvidia GPU overheats

iruneachteam · 2018-01-22 19:14:07

Hello there.
There has been an overheating problem with my Nvidia GeForce GT 630 GPU lately. The temperature is around 70 °C when idle in cinnamon DE, around 81-90 °C when using Firefox and over 100 °C when trying to play very simple games (Such as Civilization V in the lowest possible resolution.) Pointing a bunch of fans to my computer did not help at all. I am using the Nvidia drivers. all up to date. It is not unusable, but the computer crashes whenever I try to play a game or do something related to rendering. This problem has started recently (In fact, it started in Jan 20). I was able to run any games without any problems before. I'd appreciate some help and thank you in advance.
Edit:It is also possible that the sensors are wrong, nvidia-settings tells me that the temperature is over 70 °C just after booting up (Shouldn't be possible to heat up that much in that little time.)

Last edited by iruneachteam (2018-01-23 19:03:20)

ewaller · 2018-01-22 19:21:09

What is the ambient room temperature?
What temperatures where typical before whatever changed changed?

iruneachteam · 2018-01-22 19:24:11

ewaller wrote:

What is the ambient room temperature?
What temperatures where typical before whatever changed changed?

The room temperature is around 21°C. Honestly, I do not know what the temperatures were like before, but my computer has just started crashing.

ewaller · 2018-01-22 19:34:19

It is crashing rather than just getting scary hot. The room temperature seems nominal. It is January, I was just making sure you are not in the southern hemisphere facing 35°C+ temperatures. Nvidia says their GPUs should be good to 105°C
What changed on January 20? Will you post your pacman log?

iruneachteam · 2018-01-22 19:37:38

ewaller wrote:

It is crashing rather than just getting scary hot. The room temperature seems nominal. It is January, I was just making sure you are not in the southern hemisphere facing 35°C+ temperatures. Nvidia says their GPUs should be good to 105°C
What changed on January 20? Will you post your pacman log?

Nothing, I think. I just installed a couple of games on Steam. Do I need to post the whole log, or just the past few days?
Edit:Also, it gets up to 105 °C just before crashing.

Last edited by iruneachteam (2018-01-22 19:38:23)

seth · 2018-01-22 21:09:25

Notebook or PCI card?
Actively or passively cooled?
Just to be sure: the GPU temperature, not the CPU or environment? Do those rise as well?

If you can, check the cooler - many PCI cards have a stupid layout w/ the heat spreader on the bottom. The fan juggles the cooler and it ultimately detaches from the heat spreader. The GPU will go hot pretty fast then. (The cooling paste goes south in this case, so you'll have to replace that)

If the environment temperature rises as well, that's probably not the issue, but RAM and CPU can not stand the temperatures the GPU could (while in any case the GPU should not heat up w/o major load)

iruneachteam · 2018-01-23 09:28:47

seth wrote:

Notebook or PCI card?
Actively or passively cooled?
Just to be sure: the GPU temperature, not the CPU or environment? Do those rise as well?
If you can, check the cooler - many PCI cards have a stupid layout w/ the heat spreader on the bottom. The fan juggles the cooler and it ultimately detaches from the heat spreader. The GPU will go hot pretty fast then. (The cooling paste goes south in this case, so you'll have to replace that)
If the environment temperature rises as well, that's probably not the issue, but RAM and CPU can not stand the temperatures the GPU could (while in any case the GPU should not heat up w/o major load)

PCI card. Not sure about the cooling, sorry. The CPU and the environment temperatures stay relatively normal compared to the GPU. But I think I know why it overheats : I had opened the chassis many times before, but never before did I check whether or not the GPU fan is running (It points downwards, never paid much attention) and it turns out the fan is not running. Dead. I now (Probably) know why the GPU overheats, but I do not know how to make the fan run again. The overheating might have another reason, but for now I should try to fix the fan first.

seth · 2018-01-23 09:38:37

"Never" like in "never"? Not event right after the system powers up (eg. in BIOS/Grub)? In that case it's a HW issue.
You could check the cooler on whether it picks heat. In case that's not obvious:
USE A SUFFICIENT THERMOMETER AND NOT YOUR HAND! ;-)

iruneachteam · 2018-01-23 09:43:20

seth wrote:

"Never" like in "never"? Not event right after the system powers up (eg. in BIOS/Grub)? In that case it's a HW issue.
You could check the cooler on whether it picks heat. In case that's not obvious:
USE A SUFFICIENT THERMOMETER AND NOT YOUR HAND! ;-)

Thanks. Yeah, never as in never, the GPU is pointed downwards and this is a pretty old system, didn't pay attention to the fan at all. Although I am positive the fan used to run (Up to a few days ago) since it has just started overheating. What do I need to do to check whether or not it is a hardware issue or not? I'd be very happy if it was a software issue.

seth · 2018-01-23 10:44:56

If it doesn't run up for a self-test, it's broken.
You could try running nouveau instead of nvidia, but the fan should start running on power and then be controlled down.
If this isn't the case, it's a HW issue (could just be a loose connector)

iruneachteam · 2018-01-23 10:47:40

seth wrote:

If it doesn't run up for a self-test, it's broken.
You could try running nouveau instead of nvidia, but the fan should start running on power and then be controlled down.
If this isn't the case, it's a HW issue (could just be a loose connector)

Nevermind, turns out the fan cables disconnected! Thanks for the help.

seth · 2018-01-23 15:15:10

Cool. I assume the heating problem is gone?
In case please mark the thread as solved by editing your first posts subject. Thanks.

iruneachteam · 2018-01-23 15:35:39

seth wrote:

Cool. I assume the heating problem is gone?
In case please mark the thread as solved by editing your first posts subject. Thanks.

I soldered the wires of the fan back, then the GPU got so hot again that the solder melted. Still not gone.

seth · 2018-01-23 16:23:15

Did the fan run until this happened?
Solder should be good for 150°C - that's too hot for a GPU. (I assume it's the fan-end of the cable on top of the cooler? The board would probably start to melt?)

iruneachteam · 2018-01-23 16:31:09

seth wrote:

Did the fan run until this happened?
Solder should be good for 150°C - that's too hot for a GPU. (I assume it's the fan-end of the cable on top of the cooler? The board would probably start to melt?)

Honestly, I have no clue. According to the Nvidia X Settings app, the temperature was at 100°C all the time. (Decided to play a game to test the temperatures. Later when I was using Firefox, the PC crashed. The fan's solder was melted at that point (It is very much possible that I couldn't solder it correctly.)

ewaller · 2018-01-23 18:58:11

seth wrote:

Solder should be good for 150°C - that's too hot for a GPU. )

Eutectic solder melts at 183 °C.

iruneachteam · 2018-01-23 19:03:01

ewaller wrote:

seth wrote:
Solder should be good for 150°C - that's too hot for a GPU. )
Eutectic solder melts at 183 °C.

Then perhaps I couldn't connect the board and the cables well. Either way, it is not normal for a GPU to reach 70 °C when idle. I probably will contact Nvidia support. Thanks for the help!

seth · 2018-01-23 20:19:17

If the GPU never powers down (performance mode all the time, overclocked) and is designed to be actively cooled and the fan is broken the temperatures are easily reached.
If you can, check nvidia-settings for the GPU mode/clocks before it goes south.

----

ewaller wrote:

Eutectic solder melts at 183 °C.

There is not "the" eutectic solder and eg. Sn42Bi58 already melts at 138°C
You can also have Au80Sn20 which is good for 280°C so I just took a generic value that you can expect.

Ftr, 183°C would be Sn60Pb40 which is a big no-no in at least Europe (because of the lead)

Arch Linux

#1 2018-01-22 19:14:07

[SOLVED]Nvidia GPU overheats

#2 2018-01-22 19:21:09

Re: [SOLVED]Nvidia GPU overheats

#3 2018-01-22 19:24:11

Re: [SOLVED]Nvidia GPU overheats

#4 2018-01-22 19:34:19

Re: [SOLVED]Nvidia GPU overheats

#5 2018-01-22 19:37:38

Re: [SOLVED]Nvidia GPU overheats

#6 2018-01-22 21:09:25

Re: [SOLVED]Nvidia GPU overheats

#7 2018-01-23 09:28:47

Re: [SOLVED]Nvidia GPU overheats

#8 2018-01-23 09:38:37

Re: [SOLVED]Nvidia GPU overheats

#9 2018-01-23 09:43:20

Re: [SOLVED]Nvidia GPU overheats

#10 2018-01-23 10:44:56

Re: [SOLVED]Nvidia GPU overheats

#11 2018-01-23 10:47:40

Re: [SOLVED]Nvidia GPU overheats

#12 2018-01-23 15:15:10

Re: [SOLVED]Nvidia GPU overheats

#13 2018-01-23 15:35:39

Re: [SOLVED]Nvidia GPU overheats

#14 2018-01-23 16:23:15

Re: [SOLVED]Nvidia GPU overheats

#15 2018-01-23 16:31:09

Re: [SOLVED]Nvidia GPU overheats

#16 2018-01-23 18:58:11

Re: [SOLVED]Nvidia GPU overheats

#17 2018-01-23 19:03:01

Re: [SOLVED]Nvidia GPU overheats

#18 2018-01-23 20:19:17

Re: [SOLVED]Nvidia GPU overheats

Board footer