My GPU's fans don't blow unless I'm doing some rendering or gaming, and as harmless as that sounds, my card is often warm, or even hot, to the touch. I'm looking through the driver UI to see if there's an option to keep the fan on, but there doesn't seem to be. I also don't think there's any options in the BIOS, but I haven't checked recently.
Anyone have solutions to this? Is this normal?
Other urls found in this thread:
ask ur mum she knows how to blow
Install MSI gaming app before your card fries itself.
I'm serious, I fried my MSI 1080Ti this way.
How hot is hot? The rx580 armor is designed to not spin until it reaches 60c
That shit is retarded.
Fans should always spin, because GPU temp is a bad metric.
I fried my 1080Ti's GDDR5X because I was playing old games and the gpu was basically sitting idle.
Go into global wattman in radeon settings and make a fan curve.
Found it, I'll try that thanks
that makes no sense. for old games, the gpu utilization must have been almost nothing, meaning no difference between playing those games and staring at your desktop wallpaper.
you're just a retarded buttfucker.
make a custom fan curve. my "idle" is 30 c
Indeed, the gpu was doing nothing, but the memory was doing stuff.
The thing is, if you're gonna do a no spin fan, you should have temp sensors on memory chips as well.
no you are retarded.
The GPU is completely fine at any temps below 90 something.
You probably fried 1080ti because Nvidia is shit.
Well I suspect cemu is what killed it.
Since I was doing long sessions, and the fans were definitly stopped.
Here's the result if you're curious.youtube.com
Go to 5:45
Is that the 1080ti that died?
Looks like it had a defective temp sensor
>full load
>reports 40C
>probably doesn't activate the fance since it thinks it's cold.
40°C is processor.
No 1080Ti goes to 3500Mhz.
This fixed the issue, zero rpm is bullshit. I will have to tinker a little, but it is already much better. Thank you!
I hope that is just some stock image because God help you otherwise.
That MSI heatsink sucks even on their Nvidia GPUs, but is absolutely inadequate for AMD-based cards.
gpu is top line, right?
temperature goes to around 70, utilization above 90%, and the last number is fan RPM, which is 1000.
so the fans were working up until the firestrike demo.
nope, look a 5:45. Fans are stopped.
It is, XFX made my card. I think it the curve and zero rpm bs is retarded, but now the fan idles at 20% and by 60c it's at 60%. Playing Rust while setting up, it was up at 86c while the fan was at like 30% or something.
Like I said, I'll tinker with it, but this is already an improvement.
I was saying the fans were spinning until the battle between ice guy and fire guy started. In the beginning of the video, you can see fan rpm.
You probably just bought cards with bad coolers or the fan control fucked up. AS long as the GPU is beloe 60°C it's near impossible for the VRAM to hit problematic temps.
I did this video to speed up RMA process, that still took a month because mining craze.
But this is default behavior, whith only nvidia drivers installed.
1080Ti didn't spin up much while playing Botw o cemu for long hours. It had me worried, but I thought it should be fine, right?
This kind of artifact are very symptomatic of defective memory.
In fact, I could reduce how long it took for them to appear by downclocking the memory.
I'm 99% sure that's what happened.
So by installing the MSI useless utility, I get the fans to spin even at idle by not changing any setting.
>MSI 180TI GamingX
Bad cooler.
>fried VRAM
Yes. Bad cooler.
Follow the posts backward.
I think they expect you to have 7 fans running on top of your GPU and CPU fans, so they try to conserve. Setting the idle on your fan speed to anything above being idle will automatically keep your GPU cool. Then start the curve from 40-45c onward in increments.
On mine, if the temperature is over 75c, the fan is at 100%. There should be no reason your GPU gets over 65-70.
>expect you to have 7 fans running on top of your GPU
What's the point of having them not spin, then?
It's not gonna be noisier, and you'll save, maybe, a Watt.
I have an RX570 armor and I keep it with Zero RPM disabled. With it enabled it reaches about 40C on idle and at full load it reaches 75C at 1500RPM. Disabled, on idle the card goes down to 26C. I guess I should keep zero RPM disabled.
Anyways, I'm a bit paranoid 1. Because I heard the cooler on this card is the worst possible and 2. Because I don't know if I really did the correct thing to measure temperature on full. See, I tested FurMark and not happy with it not utilizing enough core clock speed, I used Unigine Valley and stopped the fucker at a specific scene which made my card run at 100% all time.
Problem with that is, I don't know if it's even accurate at all. I tried another game recently that doesn't make my card go past 60C, but the usage still is shown as 95% to 100%. So am I not supposed to trust the core/memory clock speed + gpu usage?
Look backwards as my 1080Ti GamingX fried up its memory.
zero rpm is never good, unless you have memory temp sensors.
Yeah, that's the problem. No sensors for temperature other than one.
Install afterburner and set the fan curve to start at 40.
Zero RPM decreases noise and prolongs the fan's life span while still keeping everything resonably cool as long as you have a decent cooler on your card.
That's how it works user. Your card is fine, otherwise cards would be dying en masse.
I can't believe people are advising to make a custom curve and there's like a single person saying how this is normal behavior.
It works fine if you're only playing modern games, getting 80+% gpu usage.
Unfortunately, it will fuck up by not cooling your vram when GPU is bored.
And let me tell you, a 1080Ti is very bored at 1080p.
> Unfortunately, it will fuck up by not cooling your vram when GPU is bored.
That only describes that your card cheaped out on VRAM cooling. Most modern cards have enough heatsinks on the RAM modules to even cool it entirely passively. VRAM uses around 50W max, there's no fans spinning needed. Unless the card has a bad cooling design and fancurve.
Dude, the GamingX cooler is no joke, unlike the Armor.
There's a reason I bought this shit.
Just saying, the first one died in 2 month, and I've had the replacement for a year. Only never running 0rpm.
goddamn boomers
It's obviously a good cooler for the GPU but NOT FOR THE VRAM
It's just bad design and you had bad luck. That's a problem with YOUR cooler though, not with zero RPM.
I give up.
I'm saying it's fine 99% of the time, as GPU usage will be proportional to memory use.
But playing cemu threw it out the window.
I was getting 20-25% GPU use while memory was the bottleneck.
Therefore it didn't spin the fans.
I played for hours with fans not spinning.
Is it that hard to grasp?
>I played for hours with fans not spinning.
See, that's fine!!!! Because there was only little heat to dissipate. The VRAMs should be fine even when the fans are not spinning if the cooler is appropriate.
Ofc if the VRAM cooling is shit you need fans spinning. But on normal cards that's not an issue.
I have several Nvidia and AMD cards here, none have an issue with CEMU.
>AS long as the GPU is beloe 60°C it's near impossible for the VRAM to hit problematic temps.
*100C
vram is made for operation up to 104-125C
Must be why it fried up! because it's fine.
Dude, just look up 1080Ti GaminX reviews. It's praised every-fucking-where.
I was pretty hooked up with Botw at the time.
Terribad game if you ask me, but I was still hooked for some reason.
Was having 10+hours sessions, and/or just keeping it running in the background because it took minutes to launch it again.
You're obviously too dense to understand that GPU.cooling != VRAM cooling.
If you honestly think that your VRAM overheated while your GPU was chilling below whatever the zero rpm thershold is, then ok.
I mean you could just check VRAM temps in CEMU right now and see that it's fine even with no fans and you probably just got a lemon.
>check VRAM temps
except you can't do that, because there are no vram temp sensors.
I found that my r9 380 would idle at 70c just sitting at desktop with dual 1080p screens.
I'm not sure why sapphire thought this would be good idea in it default configuration.
I'm guessing it's not Sapphire's fault but AMD drivers that turns on 3D clocks when you connect two monitors.
It could also be your power settings in the driver options or the OS that triggered the card to never fully go idle.
NO.
You either run it hot or never let it get hot.
Thermal differences are what kill the card, if you can't keep the card at 30C all the time, which you CAN'T, you better have it 50C on idle and 70C on load, than 30C on idle and 70C on load.
If you're really this stupid you might as well throw your card into the trash right away. The additional heat on idle won't fuck up your caps or VRMs, but the thermal expansion of running it cool when idle WILL fuck up your whole chip.
Any proof on this?
I just set used my tuning software and set my fans to 80% on my gpu when I got it and havent touched it since.
It's basic physics. If there's a lot of expanding and contracting then it can (and will, in the long run) break solder points and possibly other things due to the differing thermal expansion rates of the materials.
This. I had a 260 running for 10 years at 50C idle and it's still working.
>r*tards itt
my msi 1060 is still running strong after 2 and a half years of running off idle at 30 degrees sometimes. you guys are just bad with tech.
in globalwattman.
set zerorpm to off.
Idk man, considering most computers go into sleep mode, they'll be cooling down to
>2(two) whole years
holy shit user, thats a golden sample