Graphics Cards

[dqpb]

As of about a week and a half ago, my GPU has started getting up to 100C+ whenever I run 3D games, shutting itself off when it gets to about 105C. I have the same problem as the person who made this thread at the end of March, and especially this person who replied two months later. That is, one of my fans isn't working and fan speed is reporting as 0 RPM (or, oddly, sometimes 11 RPM) regardless of whether the working fan is running. My GPU temperature also creeps slowly upward (currently at 82C) even when I only have an internet browser open, like now.

The OP of that thread above apparently fixed their problem by updating the Radeon Software, but I'm already on the current version (21.5.2).

Other stats, in case they're helpful:

• CPU: Intel Core i5-4690K @ 3.5GHz
• RAM: 8GB 1600 MHz DDR3 memory
• Nothing is overclocked

[Anonymous]

Nicht die Adrenalin Oberfläche benutzen, sondern nur den Treiber 21.6.1 WHQL installieren und mit den MSI Afterburner die Lüfterkurve anpassen und prozentual erhöhen. Dort kann man auch die Core Voltage verringern, was auch zu weniger Hitzeentwicklung beiträgt.  Mit einem Benchmark  CPU Test, ich benutze den FurMark, kannst du die Temperatur und Maximum überwachen, FSP und Core Clock vergleichen mit deiner Volteinstellung, ob die FSP einbricht. Im Grafiktool kannst du 5 Einstellungen abspeichern und den Grafikmaschinen der Games anpassen. Im FurMark kannst du auch die Daten deiner Grafikkarte des Herstellers mit der Standardeinstellung vergleichen. Wenn das nicht hilft, sind entweder die Lüfter  oder das BIOS n.i.O.?

[Geforsikus_2021]

Try to remove the video card driver via DFU and reinstall it after restarting the PC. How much the fan rotates, look under the video card, how the video card behaves during the downtime of the PC and during the Internet browser(frequency, load on the video card), there is either a glitch driver, or you have caught a miner, in simple tasks if everything is normal with the video card, it will not warm above 60, even the coolers will stand. If you even have a map heated up to 80 degrees in the browser, it means that the map is loaded with some kind of program or (a miner virus). There are hwinfo programs or the same amd adrenaline, there is monitoring(albeit meager) better than hwinfo or gpu-z. Your task is to check how the video card behaves, and then solve the problem by either reinstalling the driver(if the card behaves normally in idle time), or scanning the system for viruses(if the card is loaded even in simple tasks).

[dqpb]

Sorry for the delayed reply; I didn't have much free time yesterday and trying your suggestions took a while.

@Anonymous: Adrenalin 21.6.1 wouldn't install because it's not compatible with my GPU. Also, changing the fan curve in MSI Afterburner, my fan doesn't seem to be able to run at anything other than 0% or 100%—and even then, only one of the two fans runs. I'm nervous about messing with the CPU voltage curve and running FurMark (I don't understand how the former would affect my GPU temperature and I suspect the latter would crash my computer), but I'll come back to those if necessary.

@Geforsikus_2021: A Malwarebytes scan didn't find anything, so I tried uninstalling the GPU driver, rebooting, and reinstalling. I watched my GPU and both of the fans shuddered a bit as it was rebooting, but I still can't get one of them to move.

[Geforsikus_2021]

Considering that the video card is far from new, try to disassemble it, clean it if the fans are spinning well on it (that is, you can spin them with your finger and they are willingly spinning), then disconnect the fan power connector itself from the video card and stick it back again (maybe the connector has oxidized). Try again using speedfan to adjust the fan curve. Disassemble, clean, re-apply thermal paste, if the thermal pads are very worn out, try to replace (buy new ones) In theory, during downtime, the card should not be particularly heated, but work in a semi-passive cooling mode. Maybe the cooler has failed or the controller itself, which is responsible for the fan rotation curve, or the wire contact(oxidized) has no contact with the coolers.

[dqpb]

Before spending any money, I tried disconnecting and reconnecting the fan power cord. That doesn't seem to have changed anything, so I ordered thermal paste, which should come between August 20th and 25th; and replacement fans, which should come between August 25th and September 17th. I'll let you know what happens when I put those in.

[DDubya]

I have an R9 380 also and I just sorted out some overheating issues myself this week. In my case, both fans were turning but one of them was a bit reluctant. 

First thing I did is pull the card out and clean it.  My R9 380 was made by XFX and it has 2 small fans that are sucking air through a fairly low profile heatsink.  Dust was building up under the fans and in the heatsink, which I'm sure would have affected airflow and cooling. On the XFX card there are two tabs around each fan and you can actually pop them out quite easily. With other cards you'd just need a screwdriver to remove the fans.

It seemed like my card was packing dust between the back of the fans and the heatsink.  Much dustier in there than my case fans or CPU heatsink.  I guess the XFX R9 380 must double as an air filter...

After cleaning, my slower fan looks like it's now turning the same as the other fan. However, if you clean your card out and your bad fan still won't turn, I'd definitely replace the fans.  I don't think it's a driver issue.  Fans fail sometimes, and the card is never going to cool with one dead fan. 

I don't know if this forum allows links to shopping sites so I won't post.  But for my XFX card I found replacement fans for $10 (for both fans) on AliExpress.  If you also have an XFX card, the model number for their R9 380 replacement fans is FDC10U12S9-C.  Get the right replacement for your specific manufacturer.

I haven't ordered replacement fans myself yet because they still seem to be working better now, but after cleaning I still wasn't thrilled with my GPU temps.  Sometimes it just made more noise than I'd like with the card being 80c and the fan running 60%-70% for a long time.  Occasionally the temp was steadily increasing and I was seeing 85-90c before something bad happened.  I decided to tweak the card voltage and I'm pretty happy with the results. 

I did some benchmarking with Novabench before under-volting and again after.  After the voltage adjustment, the temps dropped down to the 65-70c range during the high intensity part of the test, and 40-45 when idle.  Benchmarks were the same or actually better in some cases after reducing voltage.  I think I was running into some throttling before with those high temps and now I get better performance while running cooler at lower voltage.

I'm not a tuning expert, so I may have over corrected when reducing.  I may try gradually increasing the voltage to find out what the ideal settings are for me.  I'm sure if I go too far I'll start to overheat again.  However, these are the current good settings:

This is in the Performance --> tuning section of Adrenaline 20.11.2.

[dqpb]

As luck would have it, my R9 380 is also made by XFX! I took it out and cleaned as much dust out as I could, but I checked my thermal paste and it's looking old and crusty. I'll order some new paste and get back to you once I use it.

[alex221]

 attempt