PSA: AMD Crimson drivers disable/lower fanspeed on certain GPU's causing them to fry.

[Nuckels]

The latest crimson driver will lock your fanspeed in overdrive, causing them to not or spin at a very low RPM and potentially frying the GPU.
This does not occur with all AMD gpu's but as of right now there is no accurate list of which gpu's do and do not have this issue.
Read more about it here: https://np.reddit.com/r/Amd/comments...an_speed_when/ and on the rest of the AMD reddit.

I coudn't find any topic on it yet seeing as I remember reading here that people had issues with them.
So bewary of upgrading them, how this got past QA is beyond me.

GG AMD you dun goofed.

Edit: I rebooted in safe mode and ran the latest DDU to uninstall them, and installed a older version and no longer have said issue.

[BicycleMafioso]

how this got past QA is beyond me.

For the same reason that nVidia shipped self-frying drivers with the GTX 590 and how the first (1? 2?) of the Windows 10 launch drivers fried Kepler generation cards.

Also, taking this with a grain of salt atm, Reddit appears to be down right now. Will comment later when I can actually read what's up and see if any cards has actually fried yet.

[Nuckels]

For the same reason that nVidia shipped self-frying drivers with the GTX 590 and how the first (1? 2?) of the Windows 10 launch drivers fried Kepler generation cards.

Also, taking this with a grain of salt atm, Reddit appears to be down right now. Will comment later when I can actually read what's up and see if any cards has actually fried yet.

I have ran them myself and noticed my GPU going up to 70degrees in csgo where I normally ran at 45degrees, had a fanspeed of 21%.
Some googling later and there are a bunch of people raging their GPU has shit the bed.

https://www.reddit.com/r/Amd/comment...ans_dont_work/

Reddit works for me btw.

Edit: just happy I noticed in csgo and not in GTA 5 =_='.
Edit2: It seems to only occur if you got overdrive enabled tho.

[Deleted]

Edit2: It seems to only occur if you got overdrive enabled tho.

I got Overdrive enabled for TW3, played it and temps are fine.

So might be just very specific cases.

[Artorius]

Mine is currently at 37°C, with those drivers for 3 days and I haven't noticed anything wrong yet.

[Butler to Baby Sloths]

AMD Radeon Graphics @AMDRadeon

We are aware of low fan speed reports on select GPUs with Radeon Software Crimson Edition. We intend to publish a hotfix on Monday.

https://twitter.com/AMDRadeon/status/671058196547706880

Hotfix coming tomorrow.

[Deleted]

On my second machine...I have the complete opposite. When I boot, my fan speed is set to 49%...I have to manually go into the overdrive settings and turn it to auto everytime I boot or it sounds like a jet. Nothing i've tried has fixed it yet.

[Evildeffy]

Question:

Those of you having this... are you on Windows 10?

Because I found a disturbing (and repeatable) issue in Windows 10 regarding drivers.
Any dual GPU card (I tried this on a HD4870x2 / HD5970 / HD6990 / HD7990 / GTX590 / GTX690 - Didn't have the R9 295X2) when installed on Windows 10 by it's MS Update/Native drivers would often ignore the GPU ID number of which has to be first (it does NOT do this on W7/W8.1) and set GPU 2 as the primary GPU.
In doing this the card basically loses it's fan control function and it'll either stick to it's lowest possible setting or turn off entirely, because fan control is linked to the control board of the primary GPU.

To fix this requires a lot of patience, DDU and reinstalling the driver multiple times and then pray Windows 10 doesn't revert drivers again.

Over all of these cards it's rather easy to burn out the cards if your fan doesn't adjust and it's got something to do with Windows 10's hardware ID recognition.
Since Windows 7 and Windows 8.1 would always identify the GPUs properly.

This issue sounds extremely familiar to me and I believe your issues are actually more due to Windows 10 than the drivers itself.

[Evildeffy]

http://wccftech.com/amd-crimson-driv...ottling-issue/

Driver update released, problem fixed apparently.

[Vash The Stampede]

So it seems the drivers were responsible for dead video cards. Who's going to replace the dead video cards now? AMD or card manufacturers.

[Evildeffy]

Just like the nVidia case: No-one unless you can prove it was caused by the drivers.

Good luck with that.

Though the case isn't as wide spread as the nVidia case.

[Wries]

Some month ago I had a scary moment where my 980 refused to ramp up the fan speed. Unknowingly I sat there while temps climbed. Until they reached 90 degrees, apparently, because at that point some sort of failsafe must have set in and a hurricane was all of a sudden let loose inside the PC, prompting me to check the MSI/AB graphs.

Soo yeah. They should have that.. Failsafe shutdowns, thermal throttling. Stuff like that. Have chips actually been fried? Anyone here on mmo-c affected first hand?

Just like the nVidia case: No-one unless you can prove it was caused by the drivers.

Good luck with that.

Though the case isn't as wide spread as the nVidia case.

If it's within warranty, the cause shouldn't matter that much. And many vendors extend themselves to 3+ years nowadays.

[Butler to Baby Sloths]

Yeah, there should be a thing that sets fan speed to 100% if the temperature climbs excessively fast, and not just the GPU, but the VRM and memory which aren't in direct contact to 4 fucking massive heatpipes (and thus rely wholly on airflow).

Or perhaps a failsafe that turns off the card if the fans stop spinning.

[Vash The Stampede]

One thing that's bothered me was how computers report their temperature, or the lack of reporting it. The thing could sit there happily cooking away without any warning what so ever. Intel does this too but they've done a good job throttling the CPU to prevent damage. Neither AMD or Nvidia have any implementation to do this with their GPUs and on top of that no alarms go off to alert you either. That's a practice that needs to change.

[Artorius]

One thing that's bothered me was how computers report their temperature, or the lack of reporting it. The thing could sit there happily cooking away without any warning what so ever. Intel does this too but they've done a good job throttling the CPU to prevent damage. Neither AMD or Nvidia have any implementation to do this with their GPUs and on top of that no alarms go off to alert you either. That's a practice that needs to change.

Quite honestly the GPU should shut itself down to 100mhz when reaching critical temperature. Intel CPUs do.

[Remilia]

Wonder if there's any possibility to do like an over temperature protection like in PSUs, where there's no software that can intervene it at all.

[Artorius]

Wonder if there's any possibility to do like an over temperature protection like in PSUs, where there's no software that can intervene it at all.

Well, they could just take the power off completely. Windows would blue-screen but whatever.

[Deleted]

http://wccftech.com/amd-crimson-driv...ottling-issue/

Driver update released, problem fixed apparently.

Still hasn't fixed my fan speed setting itself to a specifically high level upon boot...

[Evildeffy]

Still hasn't fixed my fan speed setting itself to a specifically high level upon boot...

You sure you don't have stuff like MSI Afterburner installed?

[Deleted]

You sure you don't have stuff like MSI Afterburner installed?

Nope. Problem doesn't happen when I revert back to the old drivers. Only happens when using any of the crimson drivers.