AMD Radeon Crimson Driver

No issues with fan on my R9 290X Vapor-X but I run MSI AF with custom fan profile :)

So far I haven't ran into any problems with this driver, but I don't play any new games either!
 
I went back to the non Crimson divers. I was getting this weird stuttering indoors in fallout 4 with both Crimson drivers, rolled back to 15.11.1 non Crimson fixed the issue.
 
found this:
PC hardware manufacturer AMD recently released brand new drivers for their Graphics Cards called Crimson which offered improved performance in certain games notably Counter Strike: GO. After a couple of days users are reporting bugs and glitches with the new drivers with the major one killing numerous GPUs. The Crimson Drivers are supposedly limiting fan speed of the Graphics Cards which are causing them to overheat and die before the user can realize it. Drivers are overiding user setting and setting fan speeds to just 20% which is not suitable especially while playing demanding games. For those who have the Crimson Drivers installed and are looking for a fix can look into the temporary solution listed here which turns the fan to auto and change speeds as the temperature of your GPU rises. The solution is temporary because it the settings reset with every reboot.
 
Why do you need a temporary solution for the fan problem? Why not install the fixed drivers?
 
Fan speed was not a problem here, it rarely goes above 20%. I don't have time tonight, but when I get off work tomorrow I will reinstall Crimson and monitor the clock speeds.
 
OK, tested in fallout 4 at the Old North Church, anywhere there are dead ghouls on the ground my card downclocks to around 620 MHz, turn away from them, back to normal speed.
 
Whats normal speed? Also I now feel like I should have waited to install the crimson drivers but for some reason I always give AMD the benefit of the doubt.
 
1050 MHz, the drop happens at the same time it starts stuttering. I play at 4K, but this stutter seem to happen at any resolution. This is going to sound strange, but it almost seem like the stutter is causing the downclock, like how vsync will cause a card to downclock if it can maintain the frame rate at a lower frequency.

edit: Its like when you have something that is CPU limited, so the card never hits full speed.
 
Last edited:
@Sinistar
What is your CPU doing when those ghouls appear? Anything unusual going on with the load on your various threads/cores?

If the stutters are going to cause the downclock, the CPU would likely be maxing out at that point, although there could be some other form of bottlenecking too I suppose. Also, if the stutters are the root cause, lowering the rez ought to increase the downclock more than running at a very high rez (as there's less work for the GPU to do...)
 
yeah, I didn't think of checking the CPU usage before I uninstalled the driver. Will check that out in a while.
 
Not only do the clocks drop, but GPU usage drops near 0%.
No CPU spikes.
Yes, at lower resolution the downclock drops below 400 MHz
If I drop down to 720P the downclock gets near 300 MHz but the game remains smooth because it can maintain 60 FPS.
 
Last edited:
Hey I have a weird stuttering with FO4 with my notebook. It has 860M and a i7 4810QM. It stutters most indoors for some reason. I haven't checked for weird GPU power management behavior....
 
With the Crimson drivers something is causing GPU usage to drop to near 0%, then the clocks go down from the low usage.
Catalyst drivers GPU usage stay between 95 to 100 percent, and clocks never drop in 4k res.
 
This isn't confirmed, just hearsay, but apparently the original news about dead cards is false, and failsafe's on the cards have worked normally - the fan speed has gotten stuck, which the hotfix was for, but there's no confirmed dead cards
 
This isn't confirmed, just hearsay, but apparently the original news about dead cards is false, and failsafe's on the cards have worked normally - the fan speed has gotten stuck, which the hotfix was for, but there's no confirmed dead cards


Effectively, it will be really hard to kill a card nowadays, due to overheatts ... 10 years ago ( way more in reality ), when i was doing overclocking, it was frequent that we was kill the core, the ram, or vrm, due to overheat, now, the failsafe, kick, and it power off the gpu way before the temperature overheat can damage component.
 
This isn't confirmed, just hearsay, but apparently the original news about dead cards is false, and failsafe's on the cards have worked normally - the fan speed has gotten stuck, which the hotfix was for, but there's no confirmed dead cards
It would be a relief if this latest issue reaffirmed that hardware has kept with the times. The memory has faded somewhat with time, but I think I could still recognize the scent of a fried AMD Thunderbird. That is an era we needn't go back to.
 
It would be a relief if this latest issue reaffirmed that hardware has kept with the times. The memory has faded somewhat with time, but I think I could still recognize the scent of a fried AMD Thunderbird. That is an era we needn't go back to.
It hasn't been that long since parts have been frying, NVIDIA has had 3 driver sets frying cards, latest of which was in 2013
 
It hasn't been that long since parts have been frying, NVIDIA has had 3 driver sets frying cards, latest of which was in 2013
I considered it an indictment of the graphics cards when that occurred.
Perhaps there are reasons for why it happened, due to design cycles or some other internal cause. Given the decade of advanced warning, I have not seen a reason why it would be considered excusable.
 
Just got back from a business trip, tried the new 15.12 crimson drivers, downclocking problem still exist, back to catalyst drivers.
 
Back
Top