PSA: the crimson driver may lock your gpu fans at a random percentage!

DJS4000

DJS4000

Gawd
Joined
Nov 2, 2009
Messages
913
Just a heads up for fellow [H] users. There are multiple reports that the current crimson driver locks your gpu fan(s) at as low as 20%, causing thermal issues and, at worst, frying your card under sustained load.

Afterburner or other fan control tools seem to work, but have to be re-applied after each reboot.
 
Hot fix is coming on Monday, check around and there are a few threads with more info on this.
 
DJS4000 said:
Just a heads up for fellow [H] users. There are multiple reports that the current crimson driver locks your gpu fan(s) at as low as 20%, causing thermal issues and, at worst, frying your card under sustained load.

Afterburner or other fan control tools seem to work, but have to be re-applied after each reboot.
Click to expand...

modern GPUs are thermally protected, it would be almost impossible to fry one from a busted fan.

Its most likely just bozos looking for a handout...
 
Dont they test this shit before they release it?

WTF. AMD?
 
it was just defaulting to manual instead of auto if cnext crashed.

If it stuck at 20% it would just throttle. Anyone saying it fried their card is full of it.

it only effected a small subset of people

mine was fine.

shit happens.
 
The Mac said:
it was just defaulting to manual instead of auto if cnext crashed.

If it stuck at 20% it would just throttle. Anyone saying it fried their card is full of it.

it only effected a small subset of people

mine was fine.

shit happens.
Click to expand...

Actually it can burn up your VRMs and all sorts of shit. Believe it or not but throttling a GPU thats already designed to hot ash shit isnt going to do a fucking thing. Where do you think the threshold is set for throttling on these already hot cards? They are already pushing the limits of how much heat silicone can take.

Go set your fan to 0% for a while and play some games so when your card goes tits up i can tell you you are full of shit...

Can your for once just calm down on the zealous love for a product and think rationally for a minute? What handout is there even available? Like these guys want to lie just to sit and wait for a fucking RMA? Where do you come up with this shit?
 
the vrms are thermally protected.

there would have to a hardware defect for them to blow.

I come up with this shit with my circuit design experience and my BSEE.

Stop spreading fud.
 
Last edited:
The Mac said:
the vrms are thermally protected.

there would have to a hardware defect for them to blow.

I come of with this shit with my circuit design experience and my BSEE.

Stop spreading fud.
Click to expand...

This.

VRMs can take a heck of a lot more heat than the GPU, and when the GPU throttles, the VRMs drastically reduce their output voltage... so yeah. I highly doubt someone's card has been fried. BUT I'm willing to be proven wrong.
 
The Mac said:
the vrms are thermally protected.

there would have to a hardware defect for them to blow.

I come up with this shit with my circuit design experience and my BSEE.

Stop spreading fud.
Click to expand...

Sorry but like i said drop your fan speed and prove it. You can only throttle so much before shit goes... I believe a few edge cases lost their hardware. Play for a few hours with your fans set low like that and your hardware is history. This has been shown plenty of times now. Go youtube videos where people run processors with no heat sink until poof. Processors have had thermal protection for much longer than GPUs. Its not foolproof FFS. Stop claiming everything you dont want to hear as FUD...

I give zero fucks about your experience with circuit design or your BSEE. Claiming that thermal protection is some 100% no fault protection from failure is a really fucking stupid thing to say even more so for an electrical engineer to make such a dumbass claim.
 
Disposed said:
Sorry but like i said drop your fan speed and prove it. You can only throttle so much before shit goes... I believe a few edge cases lost their hardware. Play for a few hours with your fans set low like that and your hardware is history. This has been shown plenty of times now. Go youtube videos where people run processors with no heat sink until poof. Processors have had thermal protection for much longer than GPUs. Its not foolproof FFS. Stop claiming everything you dont want to hear as FUD...

I give zero fucks about your experience with circuit design or your BSEE. Claiming that thermal protection is some 100% no fault protection from failure is a really fucking stupid thing to say even more so for an electrical engineer to make such a dumbass claim.
Click to expand...

I'm looking for the like button....;)
 
Disposed said:
Sorry but like i said drop your fan speed and prove it. You can only throttle so much before shit goes... I believe a few edge cases lost their hardware. Play for a few hours with your fans set low like that and your hardware is history. This has been shown plenty of times now. Go youtube videos where people run processors with no heat sink until poof. Processors have had thermal protection for much longer than GPUs. Its not foolproof FFS. Stop claiming everything you dont want to hear as FUD...

I give zero fucks about your experience with circuit design or your BSEE. Claiming that thermal protection is some 100% no fault protection from failure is a really fucking stupid thing to say even more so for an electrical engineer to make such a dumbass claim.
Click to expand...

There's no comparison between a cpu running now heatsink, and this situation. The cpu situation you listed is time related, the overtemp circuitry doesn't have enough time to react with the mass of a hs to absorb some of the heat and slow the temp rise.

In this case, you've still got a heatsink and a turning fan. I'm trying to find the post now, but a user over on reddit replicated the issue by locking his fan speed at 20% and running furmark.

The gpu temps rose until they hit 95c, and then it started to throttle, and it kept throttling until temps were low enough. On top of that, he stated that his VRM temps were lower than they were during normal operation.

https://www.reddit.com/r/Amd/comments/3ur175/just_did_some_quick_testing_with_the_crimson_fan/


https://www.reddit.com/r/Amd/commen...me_quick_testing_with_the_crimson_fan/cxh1sp1
 
Last edited:
KazeoHin said:
This.

VRMs can take a heck of a lot more heat than the GPU, and when the GPU throttles, the VRMs drastically reduce their output voltage... so yeah. I highly doubt someone's card has been fried. BUT I'm willing to be proven wrong.
Click to expand...

XFX 7950s/7970s were notorious for inadequate vrm cooling. I had personally die one on me due to overheating vrms.
 
TurboGLH said:
There's no comparison between a cpu running now heatsink, and this situation. The cpu situation you listed is time related, the overtemp circuitry doesn't have enough time to react with the mass of a hs to absorb some of the heat and slow the temp rise.

In this case, you've still got a heatsink and a turning fan. I'm trying to find the post now, but a user over on reddit replicated the issue by locking his fan speed at 20% and running furmark.

The gpu temps rose until they hit 95c, and then it started to throttle, and it kept throttling until temps were low enough. On top of that, he stated that his VRM temps were lower than they were during normal operation.

https://www.reddit.com/r/Amd/comments/3ur175/just_did_some_quick_testing_with_the_crimson_fan/


https://www.reddit.com/r/Amd/commen...me_quick_testing_with_the_crimson_fan/cxh1sp1
Click to expand...


Cool story... Use your head, out of thousands of cards sold you're telling me there's no chance something went wrong and a card failed because of this? You really believe it's more likely that they are all lying? Sorry but I've seen enough cards go tits up for dumb reasons to believe in the macs magical foolproof thermal throttling.
 
Disposed said:
Cool story... Use your head, out of thousands of cards sold you're telling me there's no chance something went wrong and a card failed because of this? You really believe it's more likely that they are all lying? Sorry but I've seen enough cards go tits up for dumb reasons to believe in the macs magical foolproof thermal throttling.
Click to expand...

Well that's no argument. Out of the 1000's of cards sold there are always some that fail. Isn't it more than likely that these cards were faulty in some way and were going to fail soon anyway. You also don't know what state these cards were in, they could have been in a bad case, over volted, custom bios etc.

If a card failed because of this issue than it had some underlying problem.

There is Thermal Throttling on all modern GPUs.
 
Disposed said:
Cool story... Use your head, out of thousands of cards sold you're telling me there's no chance something went wrong and a card failed because of this? You really believe it's more likely that they are all lying? Sorry but I've seen enough cards go tits up for dumb reasons to believe in the macs magical foolproof thermal throttling.
Click to expand...

I imagine that some cards died, but If they crapped out from running at 94c, the operating temp for reference 290(x), then I suspect the card had other, underlying, issues.

I don't buy the implication that there is a widespread death of cards, simple as taht.
 
magoo said:
Dont they test this shit before they release it?

WTF. AMD?
Click to expand...

Someone obviously didn't hear about the Nvidia driver that actually turned the cooling fans completely OFF a few years ago. I think I'd rather have 20% airflow on my card than 0%.
 
Creig said:
Someone obviously didn't hear about the Nvidia driver that actually turned the cooling fans completely OFF a few years ago. I think I'd rather have 20% airflow on my card than 0%.
Click to expand...

I ran that driver, it never turned my fans off. Most likely an overhyped discovery that mainly affected non reference cards - Just like this one.
 
You must log in or register to reply here.
Back
Top