Overheating GTX1080

[--<[ Blei ]>--]

Hi

When i bought my 1080 i did some stress tests and noted that the GPU was temperature limited at 62 degree celsius by the manufacture company (KFA2).

KFA2 GTX1080 EXOC

So, all good and fine...now i was playing cuisine royale and the monitor, out of nowhere, turned out (no incomming signal/data).
Well just a bug...no...next match the same.
Maybe overheat by too many dust? Surprisingly not^^ (from what i can see/tell).
Both fans are working too.

Started GPU-Z and FurMark (stress test)...themperature goes higher than the 62 dagree...near 70 i turned off the test.

Then i switched to "advanced" at GPU-Z just to see this mess:
wtf is this? (picture)[url]92 degree maximum? this can't be right !So it SEEMS like some kind of software "mistake".Updated my really old driver and it still goes the same.Any idea how to handle this?[ibb.co]

[Kaihekoa]

GPUZ is probably reading the temperature wrong. There is no reason to run Furmark anyway; it's an unrealistic load on the GPU. Use MSI Afterburner or HWInfo 64 to read the GPU temps while gaming. As long as they are less than 80C, you're fine.

[John Doe]

Furmark will stress your GPU heavier than anything else will, use Superposition stress. That is the "maximum" temp limit before the card does something. As long as your load temps hover around that same old 60 degrees or even 70, you should be fine.

With a dual fan cooler card like that though, you might be able to see better temps on the 1080. Try at night time when ambient is cooler and work on your airflow or something.

[--<[ Blei ]>--]

Originally posted by Kaihekoa:

GPUZ is probably reading the temperature wrong. There is no reason to run Furmark anyway; it's an unrealistic load on the GPU. Use MSI Afterburner or HWInfo 64 to read the GPU temps while gaming. As long as they are less than 80C, you're fine.

both, GPU-Z and FurMark showed the same value, so i doubt it.
Installed MSI Afterburner. Same value.
Made another test an it pretty much "stuck" at 82 degree celsius, which isn't really nice :/

So...i watched that the "max. heat amount" is set to 83degree celsius in MSI-A.
So i just decrease the value and i should be good to go?

[Kaihekoa]

I thought you were saying there was a difference in temperature reports between GPUZ and Furmark. Like John Doe said, Furmark is a very heavy stress test. Unrealistically so IMO, use Superposition or 3DMark if you want to stress your GPU.

What exactly do you mean by GPU temperature is limited to 62C by the manufacturer? The monitor turning off could have been because your drivers were nearly a year old. Do you experience the same problem with the game now? If so, you could try another cable or plugging the monitor into a different port.

[--<[ Blei ]>--]

Originally posted by Kaihekoa:

What exactly do you mean by GPU temperature is limited to 62C by the manufacturer? The monitor turning off could have been because your drivers were nearly a year old. Do you experience the same problem with the game now? If so, you could try another cable or plugging the monitor into a different port.

I checked the GPU temperature (while doing a stresstest) with GPU-Z back then and noted that it never went higher than 62 deghree celsius.
The fanspeed seemed to be set to go as high as needed to make the heat not exceed 62 celsius.

i just wanted to check it out.
I set the max temerature to 75celsius (power limit went down to 83%) in MSI-A.
still pretty much 1800MHZ, which isn't that bad^^
I edited the fan-speed curve (100% fanspeed at 80celsius, just in case)

so, im going to play and check for possible graphic-bugs and...of course temperature

til now no bugs, themperature is at max. 75degree...GPU @ ~1800MHZ
Framerates also not that bad

seems to work....but i still have no clue what went wrong :/

[just.kamk /idle]

Originally posted by --< Blei >--:

Hi

When i bought my 1080 i did some stress tests and noted that the GPU was temperature limited at 62 degree celsius by the manufacture company (KFA2).

By default? Just no. Either you set it yourself, or it was pre-owned.

Then i switched to "advanced" at GPU-Z just to see this mess:
wtf is this? (picture)[ibb.co]

That is not the temperature sensor, but cards current settings!


Display turning off could also be caused by a faulty / crappy PSU.
And / or you did OC the card, insufficient GPU cooling, insufficient case cooling, etc. etc.

Use the sensor tab in GPU-Z, and activate logging.

[--<[ Blei ]>--]

Originally posted by kamk /idle:

By default? Just no. Either you set it yourself, or it was pre-owned.

it was a new card and i didn't change anything.
Maybe it was "accidentially" that the temperature stuck at exactly 62 degree celsius

And / or you did OC the card, insufficient GPU cooling, insufficient case cooling, etc. etc.
Use the sensor tab in GPU-Z, and activate logging.

No OC by me.
However, it's actually very hot outside atm (for german standard, ~30degrees celsius)...also inside my house it's just slightly less hot....maybe this played a role.

[Monk]

If you don't want to do a proper of just set temp and power limit to max and set your fans as loud as you are prepared to put up with, limiting your card to a lower temp for no reason won't bring any benefits,but will slow it down as it will ?united it's boost clock more so than a higher temp would.

Most 1080's should get 2000-2100MHz pretty easily. And mid 1900's - 2k just on its own turbo.

Edit.
Also, it's summer, stuff will run hotter, even under water my cards are up a bit.

[MarioTwins]

My 1070 hits 79 regularly.
I have not changed any clock speeds or fan profiles.
There ok to run a little warm

[_I_]

founders or blower style gpus alwayd run hot due to their design

case air enters the back of the card, goes over the gpus vrm area, then flows over the gpu heatsink and out the rear vents which ther are not many due to all the connectors on the card

after using a gtx 260 with blower, i decided never to use one of those again unless for top card in sli/cf

near 80 is warm for a 1070, check fans, vents, and airflow

[MarioTwins]

It gets plenty of air. It’s a zotac mini
So probably due to its size

[Malygos]

92 maximum is the normal shutoff point for pascal but while 82c isnt dangerous you are throttling by that point is this a mini card? Because if so they tend to run hotter than the full size cards. Idk about your case but it helped my card alot adding a fan to the bottom blowing up to the gpu i never go above 72c. You can also watch some videos on how to replace the thermal paste is easy and dropped a few degrees of my temps when i got my card from msi the thermal paste was already hard and cracking

[Monk]

Well, you aren't technically throttling on pascal under you hit the 90"s (I've had a couple of faulty cards and at which point their clocks drop to around 1500MHz), with pascal, for every 5-10c above about 30c you will 'step down' from the max potential boost clock, this is different to throttling, although, it sort of appears to do the same thing, the card slows down.

From ALOT of testing, the largest drops in potential boost clock happen at around 50c and then at around 80c, though, it will only ever reach its true potential right down in the 20's without LN2, mind you, even under water, I have had very limited testing with my temps in the 28-35 range as frankly I need more cooling to stay there than my 2 480mm rads can provide.

Edit.
As an idea of the difference, at around 30c my cards will boost to 2088MHz each, above that they tend to sit at 2076MHz, though silicon lottery plays a big part, I had one card (with faulty memory sadly) that would sit at 2150MHz on air at around 70c!!!, Though that mayor been one of the fastest 1080ti pieces of silicon in history, I shed a tear when I had to replace it lol

[Malygos]

At 82c the factory boost is gone you’re down to base clock and it will start going under base clock at 85c most boost clocks are in the 1800’s so to me losing 200mhz because of temps is throttling unless he has poor fan set up theres no reason to even be in the 80’s 92c is the shutdown point so if you hit 90c you are beyond throttling and if you play there consistently gonna knock the cards life span down

[Monk]

Base. Locks are around 1600 depending on model, anything above that is still the boost.
Throttling is when the card is trying to save itself from damage and the clocks will jump off a cliff to protect itself.

As long as a component is within operating temperature range, it's perfectly safe and it is very hard to say how it will affect its life, in reality it's maybe 10 years instead of 10 and a half etc.