a gpu that overheats ALWAYS means some goop/pad issue?

[Deleted member 234478]

guy selling Zotac 3080 AMP Holo. says it gets hot. he didnt say it overheats

is the solution ALWAYS just clean goop pads? asking $375

if it doesnt shut down under some furmark, but gets hot that would be acceptable? 80c ?

I bought a used amd card that got hot and replaced thermals and it ran fine, but I dont know so asking

 

[Vya Domus]

I don't know what you mean by goop pads, you mean thermal pads ? It's not necessary to replace them, unless they're ripped apart, other than that you can always change the paste on the GPU.

80C seems fine for that particular card :

 

[kilo]

I don't know what you mean by goop pads, you mean thermal pads ?

I believe the author is describing thermal paste.

Edit: On a fourth read, maybe not?

@inquisitor1 It's also important to verify the thickness of the thermal pads you're replacing. Thick ones can be 2mm and thin pads 1mm (generally). Please be cautious when changing the pads out with the correct size or you may damage the chips.

 

[Vincero]

Lots of potential in terms of bad case airflow, etc., can impact GPU temperature.
Unfortunately the cooling curve on some cards is to wait for the temperature to rise somewhat before ramping up the cooling effort - essentially almost creating a scenario where the card cooling only really kicks in at higher temperatures. You can always apply a custom fan curve to keep temperature lower on most cards - I've rarely ever seen a card pushing near 100% PWM fan effort even when hitting their thermal limits (usually because a power limit stops further thermal output).

Cleaning contact surfaces/pads / blowing out dust from cooler / refreshing TIM with a quality product is never a bad thing.

 

[Deleted member 234478]

goop is the paste and pad, thermal pads. my bad.


ill ask snd a message. see what he says regarding its behaviour

if its crashing? ill ask what temps it gets and im doubting its ONLY 80

I wrote 80 as a guess. I have a feeling theres some info missing in the ad

 

[GerKNG]

Zotac uses low end thermal paste on their Turning/Ampere GPUs. i bet it's pumped out and dried it for a long time.
i'd buy a tube of thermalright TF8 or TFX (Tf7 is different paste that pumps out quite fast)
Or a Honeywell 7950/Thermal Grizzly PTM.

 

[TheinsanegamerN]

Lots of potential in terms of bad case airflow, etc., can impact GPU temperature.
Unfortunately the cooling curve on some cards is to wait for the temperature to rise somewhat before ramping up the cooling effort - essentially almost creating a scenario where the card cooling only really kicks in at higher temperatures. You can always apply a custom fan curve to keep temperature lower on most cards - I've rarely ever seen a card pushing near 100% PWM fan effort even when hitting their thermal limits (usually because a power limit stops further thermal output).

Cleaning contact surfaces/pads / blowing out dust from cooler / refreshing TIM with a quality product is never a bad thing.

Waiting to turns the fans on is actually a GOOD thing, thermal cycling is what destroys GPUs, not running hot. It's better for a card to sit at 80C for 10 hours then to cycle between 40C and 80C 10 times in 10 hours.
where they screw up, as you said, is not kicking the fans on fast enough, usually for noise reasons, shame there's no "cool temp" option in drivers. I always crank mine up to limit max temps but the zero RPM option says because its frankly better for the card itself.

goop is the paste and pad, thermal pads. my bad.


ill ask snd a message. see what he says regarding its behaviour

if its crashing? ill ask what temps it gets and im doubting its ONLY 80

I wrote 80 as a guess. I have a feeling theres some info missing in the ad

Zotac was once known for subpar heatsinks. Even with new paste the cooler may always suck. Also depends on case, modern cases runa lot hotter then older designs.

 

[Vincero]

Waiting to turns the fans on is actually a GOOD thing, thermal cycling is what destroys GPUs, not running hot. It's better for a card to sit at 80C for 10 hours then to cycle between 40C and 80C 10 times in 10 hours.
where they screw up, as you said, is not kicking the fans on fast enough, usually for noise reasons, shame there's no "cool temp" option in drivers. I always crank mine up to limit max temps but the zero RPM option says because its frankly better for the card itself.

Zotac was once known for subpar heatsinks. Even with new paste the cooler may always suck. Also depends on case, modern cases runa lot hotter then older designs.

I get the whole thermal cycling thing, so yes in that sense even the fan-stop modes is actually helpful somewhat to potential lifespan... I just don't necessarily agree that the thermal window should sit quite as high - at the end of the day it is additional thermal expansion, they aren't using superconductors which are largely unaffected by temperature, and all of the other board parts do essentially soak up some of that GPU heat aside from potentially cheapo VRMs which also need help staying cool but also loose efficiency and lifespan as they get hotter.

As for the noise aspect, of course that is a design consideration... but so is making a card with a 300W+ TDP - the fact they have morphed into these 3+ width slot coolers and are using the last half of the card heatsink/shroud to blow heat through into the case for whatever other cooling you have to deal with it leaves me with little sympathy towards the designers in that sense as, essentially, you 'need' more than 1 in/out case fans to deal with that which is also additional noise even if the card is now a bit quieter.
Yeah the blower style ones were louder, but even the older GTX Titan's with blower style were only 2 slot width (50% of which was lost to IO connectors)... I do wonder how much better / capable a 3-slot+ wide blower cooler would work - would still be louder but probably much more effective vs 2-wide... might still struggle with 300W TDP though.

Also, I'd go 50/50 with older/newer cases being better for managing heat. Generally a lot of the cheaper cases these days are much much better design in terms of trying to manage heat / airflow vs old cheap cases. More expensive cases generally were better off in that regard but now ironically are designed more for style in some cases. Definitely newer SFF profile cases are much more designed to cooling than before where the priority was 'small'.

 

[Deleted member 242843]

ALWAYS? No, but it's usually the next step in diagnostics once you have eliminated fans and air flow as the reasons for it getting hotter than it ought.

 

[Deleted member 234478]

ALWAYS? No, but it's usually the next step in diagnostics once you have eliminated fans and air flow as the reasons for it getting hotter than it ought.

so a hot /overheating card can ONLY be either the fan, some issue with the heatsink, goop/pads..

does it mean some issue with a component on the card? or mostly not related?

I wish by default I could set the card to run at 45% fan speeds no matter what. I hate fan cycling noise. once its like hair dryer then its quiet and it comes on when you least need it during an intense part of the game that distracts you

 

[Deleted member 242843]

so a hot /overheating card can ONLY be either the fan, some issue with the heatsink, goop/pads..

does it mean some issue with a component on the card? or mostly not related?

I wish by default I could set the card to run at 45% fan speeds no matter what. I hate fan cycling noise. once its like hair dryer then its quiet and it comes on when you least need it during an intense part of the game that distracts you

There can be other problems with the card, sure. Fans, heatsink, TIM, airflow are certainly the most common and most likely things. If a component is getting overly hot, then it stands to reason that the means by which said heat is removed are likely the issue, but that is certainly not all existing possibilities.

MSI afterburner works really well for controlling fan speed, voltages, clocks for gpu......can use it to define a custom fan curve for your gpu, or set it at a static speed (such as 45%).

Although it's a bit more than I am comfortable with for constant use, 80c isn't really all that hot when you are running furmark, which will tax your card harder than you ever will in all but the most extreme instances.

If it is too hot, find ways to cool it off. Don't over think it.

 

[Deleted member 234478]

great, thats excellent infor. much appreciated

you arent a dingus homer.