PSA: New AMD Crimson Drivers could be killing cards

I made this thread http://hardforum.com/showthread.php?t=1883496 this morning because I have one entire day trying to solve this issue with my card, every-time I restart/shutdown/turnON my machine Radeon Settings Overdrive, the past Catalyst Control Center keep setting my fan to manual 30% and it override my MSI afterburner custom fan profile, so in order to make it work I have to enter in Radeon Settings and turn on automatic fan control, if not then afterburner custom profile will not work, this is certainly annoying, gladly my boyfriend told me to always have AB OSD turned ON, this is how I noticed this nasty bug.
 
I can confirm.. this is bad issue with my Sapphire 390X because the card simply won't turn on the fans, it got locked at 20% which I think isn't enough to make the fans spin, after couple of fresh installation this thing its still happening and i've used Afterburner and Trixx to custom fan curves, nothing work actually to make the fans work.. installed previous Catalyst 15.7 still doesn't work.. I will just wait until the hotfix if doesn't work then RMA.
 
Dumb question... Is this happening to people without utilities like Afterburner installed? There is a common thread here with people using AB, I haven't seen much for complaints from people who had their card crashing and then went to investigate more. Wouldn't be the first time a tweaking utility has screwed up fan speeds...
 
AMD has tweeted that they are aware of the issue and will be releasing a hotfix tomorrow...it is recommended that AMD users revert to a previous driver until the situation is rectified
 
I think breaking cards with this is possible because a lot of the cards out there don't have actual heatsinks on VRMs/MOSFETs and instead rely on the airflow of the fan to cool those.

So if the user had pushed voltage and power, and the fan doesn't turn on, it's feasible that that can cause those components to go poof.

Does that sound feasible?
 
trandoanhung1991 said:
I think breaking cards with this is possible because a lot of the cards out there don't have actual heatsinks on VRMs/MOSFETs and instead rely on the airflow of the fan to cool those.

So if the user had pushed voltage and power, and the fan doesn't turn on, it's feasible that that can cause those components to go poof.

Does that sound feasible?
Click to expand...

If the user was pushing voltage and power, they are operating the card out of spec and responsible for monitoring it
 
Ocellaris said:
If the user was pushing voltage and power, they are operating the card out of spec and responsible for monitoring it
Click to expand...

Look at how many people that has Afterburner installed, with custom fan profile and still Overdrive overrides it.

I doubt many people have fan speed on their OSD. I certainly don't.
 
trandoanhung1991 said:
I think breaking cards with this is possible because a lot of the cards out there don't have actual heatsinks on VRMs/MOSFETs and instead rely on the airflow of the fan to cool those.

So if the user had pushed voltage and power, and the fan doesn't turn on, it's feasible that that can cause those components to go poof.

Does that sound feasible?
Click to expand...

Voltage and Power scale with frequency.
Frequency is reduced with throttling.

Think of frequency as resistance.
The lower the frequency, the higher the resistance.
Less current flows with higher resistance.
 
Temperature does affect the life span of semiconductors, pcb, components, and packaging, but most GPU's well chips/cards/electronics in general are tested and designed for certain range of temperature for normal use.
 
cocdod said:
actually, hawaii chips die at staggeringly high rates. most likely because of the terrible heat issues even with a decent cooler, i can't imagine what would happen when a fan fails on one.
Click to expand...

No they don't. Failure in the field for R9 cards is at 1.92%. Nvidia is becoming the new Apple and I see that their fanboys are following a simple trend.

Still getting a 780 Lightning for my collection tho :D
 
Higher failure rates wouldn't be AMD's fault, they just make the GPUs. Unless there's a problem with their ref boards which barely anyone actually uses.

http://hardforum.com/showthread.php?t=1861702

The real problem are the AIBs cutting corners on 'budget' AMD products to save money. If you were ASUS, MSI, and Gigabyte would you prioritize AMD over the company selling 4 times as many video cards? I believe the "We use military quality components" nonsense is marketing fluff. There were a lot of reports of VRM issues over the last gen or two of AMD cards, remember that? If you cheap out on cooling and board design, then the cards will fail more often. Then you have the obvious stuff like higher power consumption degrading components faster. If one card uses 50% more power than another card, all things being equal the less efficient card will fail sooner.

As for the fan bug killing video cards, a GPU running at 95~100C will survive, but VRMs running at 120C+ won't.
 
Very True, this is why I stated the method of testing could have been better.

We don't really have any idea of how the rx 3xxx series are for failure rates, but I would assume they are still close to the rx2xxx series since they are pretty much the same cards.

Yeah I don't see silicon at around 100 c being that bad plus they will throttle and will shut down, modern GPU's for the past 5 to 7 years have had shut down features for excessive temperatures, now temperature spikes that is different, ie, when you start up a 3d program or game and the fan doesn't spin up, that might cause issues there as the temperature will spike too quickly for the GPU to react. And this will also effect the vrm's too.
 
Ocellaris said:
Dumb question... Is this happening to people without utilities like Afterburner installed? There is a common thread here with people using AB, I haven't seen much for complaints from people who had their card crashing and then went to investigate more. Wouldn't be the first time a tweaking utility has screwed up fan speeds...
Click to expand...

I do not have afterburner installed, both of my 7970s are at 20% fan all the time. Since both have full coverage water blocks on them and have yet to see more than 40c, I didn't know it was a bug until I saw all these threads pop up.

obviously, fans at 20% was of great concern to me :p
 
My MSI Afterburner seems to be holding up. Increased my fan speed to 85% to make sure it was working. Sounded like a Shop Vac. Knock on wood!
 
razor1 said:
Yeah I don't see silicon at around 100 c being that bad plus they will throttle and will shut down, modern GPU's for the past 5 to 7 years have had shut down features for excessive temperatures, now temperature spikes that is different, ie, when you start up a 3d program or game and the fan doesn't spin up, that might cause issues there as the temperature will spike too quickly for the GPU to react. And this will also effect the vrm's too.
Click to expand...

If a card with a dead/slow fan can spike "too quickly" for the GPU fail-safes to react, that fail-safe simply isn't effective. If they need a fail-safe for the VRM temps, that should be on the card as well.
 
Ocellaris said:
If a card with a dead/slow fan can spike "too quickly" for the GPU fail-safes to react, that fail-safe simply isn't effective. If they need a fail-safe for the VRM temps, that should be on the card as well.
Click to expand...

Control systems aren't perfect. And transients/steps are the hardest things to manage.
 
TaintedSquirrel said:
Higher failure rates wouldn't be AMD's fault, they just make the GPUs. Unless there's a problem with their ref boards which barely anyone actually uses.

http://hardforum.com/showthread.php?t=1861702

The real problem are the AIBs cutting corners on 'budget' AMD products to save money. If you were ASUS, MSI, and Gigabyte would you prioritize AMD over the company selling 4 times as many video cards? I believe the "We use military quality components" nonsense is marketing fluff. There were a lot of reports of VRM issues over the last gen or two of AMD cards, remember that? If you cheap out on cooling and board design, then the cards will fail more often. Then you have the obvious stuff like higher power consumption degrading components faster. If one card uses 50% more power than another card, all things being equal the less efficient card will fail sooner.

As for the fan bug killing video cards, a GPU running at 95~100C will survive, but VRMs running at 120C+ won't.
Click to expand...

Sapphire still sells the most AMD cards and the author of that study didn't use their cards.
 
cageymaru said:
My MSI Afterburner seems to be holding up. Increased my fan speed to 85% to make sure it was working. Sounded like a Shop Vac. Knock on wood!
Click to expand...

No AB installed, never touched the fan speed control in Crimson and all is working just fine using GPU-Z to monitor fan speed and temps.
 
Out of boredom I went back to my old HD7970 reference model this weekend (before I heard about this issue).

I have had no fan issues with the Crimson driver, even after changing the settings. They ramp up and down as necessary.
 
I may have had this issue with my card without even realizing it. My Fury never goes over 65C on 20% fan speed when gaming, so it's a non-issue for me.

In related news, the new drivers make my card run semi-passively very nicely now. The last Catalyst drivers would heat up to whatever temp I had set for the fan cutoff at idle and then continuously turn the fans on and off, hovering around the cutoff point. The new drivers make the card idle passively at just under 40C.
 
Watercooling is prohibitively expensive here. A custom loop will cost me around the cost of a 980.
 
testing methodology is invalid.

Their sample doesnt even include AMDs largest OEM. Sapphire.
 
trandoanhung1991 said:
Watercooling is prohibitively expensive here. A custom loop will cost me around the cost of a 980.
Click to expand...

I just slapped an AIO Corsair unit on mine that I had sitting around gathering dust. The kit to install it with was $30. Maybe Corsair makes an Nvidia kit also?
 
cageymaru said:
I just slapped an AIO Corsair unit on mine that I had sitting around gathering dust. The kit to install it with was $30. Maybe Corsair makes an Nvidia kit also?
Click to expand...

There's a kit called the kraken G10 that does basically the same thing, not by Corsair. Can use with various AIOs though.
 
Dayaks said:
There's a kit called the kraken G10 that does basically the same thing, not by Corsair. Can use with various AIOs though.
Click to expand...

Kraken G10 is $30 for just the bracket, still need a water cooler to use with it.
 
And remember when Kyle and other Nvidiots laughed when Reddit found the 3.5/0.5GB 970 issue, That you couldnt believe reddit....

Now we see reddit find a HUGE AMD bug that was killing cards....Those same people are now posting reddits links about AMD having issues....

O the Irony.
 
DASHlT said:
And remember when Kyle and other Nvidiots laughed when Reddit found the 3.5/0.5GB 970 issue, That you couldnt believe reddit....

Now we see reddit find a HUGE AMD bug that was killing cards....Those same people are now posting reddits links about AMD having issues....

O the Irony.
Click to expand...

Well you dont bite the hand that feeds you in this case thats nvidia. Heres something intresting nvidia so called superior drivers killing displays. https://forums.geforce.com/default/...g-samsung-and-lg-notebook-lcd-display-panels/
 
Deathroned said:
Well you dont bite the hand that feeds you in this case thats nvidia. Heres something intresting nvidia so called superior drivers killing displays. https://forums.geforce.com/default/...g-samsung-and-lg-notebook-lcd-display-panels/
Click to expand...

What I find odd is how the only real common factor here is the Samsung display present on these models is the only display out of a plethora of available displays on the market that are failing, yet everyone is blaming Microsoft and Nvidia?!

No other display on the market is exhibiting the same issue? Desktop or laptop? I run the drivers at 4k/80Hz over DP 1.2 without a single issue.

Wouldn’t it make sense to consider an issue with the Samsung's display firmware causing the fault if that's the only one real common theme here and not a single other make/model of display is having an issue under the same OS/drivers?
 
You must log in or register to reply here.
Back
Top