Unable to play TW3 after 15.7 Catalyst update

[Lisolo]

I updated to the 15.7 drivers right when they came out and I have a Sapphire 290x, though an Intel CPU. So far I've had no issues playing the game with new drivers, mostly around 58 to 60 fps with all settings on high and no Hairworks. I have my CPU OC'd pretty hard, 3.7 up to 4.7 and my 290x is OC'd from the factory but then I have pushed it up to 1300 mhz on the clock and 1500 on the memory. I remember having issues a few drivers and patches back where I had to remove my extra OC on the graphics card.

Always sucks to not be able to play the game you've been waiting for. Hope your reinstall clears up the issue!

 

[biebo]

15.7 here, no problems at all. You should download furmark and run a stress test. It's the best way to see if your GPU is alright.

 

[ketxxx]

So I saw that there were updated crossfire profiles for TW3 in the 15.7 Catalyst Drivers. I don't use Crossfire, however I am a believer in the "Update to the latest drivers regardless" mindset. I updated to 15.7 and everything was smooth. I got my system home (I work overnights and play during downtime at work) and launched TW3 for another fun filled session. Right off the bat the Raptr service crashed. Figured it was no big deal and continued to play. It wasn't until I got to Skellige that things got weird.

Diagonally connected squares started flickering across my screen. Occasionally at first, enough to be a slight annoyance. Then black squares would flicker that took up 1/4 of the screen real quick. The diagonal squares stated getting worse. My game would crash, citing that "Display driver amdkmdap has stopped and successfully recovered". I have tried the following fixes (All GPU driver installations were performed using DDU):

Restarting system - No change
Reinstalling 15.7 - No change
Rolling back to 15.6 - No change

Which led me to believe it was an issue with the game files, so:

Verified game files using GOG galaxy - No change
Reinstalled game - No change

Now heres where it gets weird. The flickering doesn't persist in the inventory areas. So the bestiary, alchemy, etc are fine. This would make you think it's possibly the GPU and a VRAM streaming issue, however don't count the weird out just yet. The flickering ONLY occured IN Skellige, however now it's progressed to the point where five minutes of playtime, regardless of area, causes intense square flickering and then black screen and loss of signal.

TL;DR: Weird flickering/artifacts that are seemingly caused by 15.7, however persisted after rollbacks, are NOT present in the menus, and ONLY occur in one specific area of the game leaving the rest of the game unaffected.

System Specs (All hardware is stock clocking):
Windows 7 x64 Ultimate
AMD FX 8350
Radeon R9 290X
16 GB RAM
15.7 Catalyst Driver

Any thoughts?

Edit: After extensive testing with many other GPU intensive games, I have found that this issue ONLY affects TW3. Unreal Tournament 3, State of Decay YOE, Just Cause 2, Ghost Recon Phantoms, Skyrim, Oblivion, Saints Row 2, 3, 4 are not affected by this issue.

---------- Updated at 06:45 PM ----------

I managed to find a Raptr video of my game back when it was playable, and the flickering only occurred in Skellige.

http://plays.tv/video/55a1631974131f06b8

Now all that happens is the diagonal squares star up fast, then the video card shuts off.

Keep in mind, NO OTHER GAME is affected by this, and I was able to play TW3 with 0 issues until I installed the 15.7 driver update

I haven't read all 3 pages so I'm going to assume you still have this problem.

1. Download Display Driver Uninstaller, when launched select the option to let it clean and restart.
2. Re-install Catalyst 15.7

Fixed? If not, read a bit more.

Delete user.settings and its backup if one is present, re-launch Witcher and re-apply your graphics settings. If problems persist shoot me a PM a lot of people seem to be having issues with Witcher so I'm using my background as a IT engineer to skit around threads clearing up what I can in my spare time.

 

[pagefault404]

Furmark caused the system to crash the same way the witcher does. I am looking into VRAM stress testing applications but at this point I am simply going to RMA the card. I work in the IT field and I have a problem with not knowing the root cause of issues. It's easier to RMA a piece of hardware and be done with it, but because this issue only affects the hardware when a specific application is running, naturally I want to know why. It's the same issue that I have when people use reinstalling windows as their end all troubleshooting. It may fix your issue in the short run, but if you don't understand what happened and why, it can just happen again.

The same thing goes if I RMA this card. This issue could appear again, then I am right back to square one. I really wish I could pick the brains of the developers who designed the architecture. There might be an issue with the stream processors that a simple change in CCC settings could fix and make this go away. Things the developers would possibly be able to help me troubleshoot.

Because games that use Nvidia technology are overloaded with tessellation to mess with ATI card users, maybe using custom tessellation settings could help with this issue. I don't know.

On the one hand, I really just want to RMA the card and be done with it. On the other, I want to crack this open so as to help others who may end up in my position.

Then again, maybe it's just my cards VRAM jacked up by the way the witcher uses it.

This rabbit hole is deep.

 

[ketxxx]

On second look of that video (couldn't see properly first go round due to its quality) the issue may be a heat related one, dismount the GPU cooler, give it and the card a good clean out of any dust, clean off the old TIM and apply some Arctic Ceramique. If you are overclocking the card reset it to defaults as well.

 

[Kinzley]

On second look of that video (couldn't see properly first go round due to its quality) the issue may be a heat related one, dismount the GPU cooler, give it and the card a good clean out of any dust, clean off the old TIM and apply some Arctic Ceramique. If you are overclocking the card reset it to defaults as well.

Agreed, i do pc maintenance and i usually observe this when gpus are overheating or they are defective. If you can still RMA do so.

 

[ketxxx]

Same here thats usually when I see it happen too although usually that black sorta checkerboard effect is either green or redish pink in my experience never seen the effect jet black like that before... which means the problem may be with some of the traces on the PCB and not the GPU, VRMs or memory overheating but lets hope its just overheating nobody likes having to RMA some hardware where its toasted itself.

 

[pagefault404]

I'll have to RMA before I pull the cooling block off. If I remove the cooler, that voids the warranty. I did make some possible headway in another area.

I changed Tessellation to override application settings and turned it down to 8x (The old hairworks framerate dump fix) and the game ran for a good five minutes after starting a new game. However I am troubleshooting this remotely and teamviewer sucks to play games through. However the system was stable until my girlfreind (my local POC) plugged my xbox controller in. I'm not sure why but maybe the controller is causing issues, so I had her unplug it, I am instaling the DLC, leaving tessellation at 8x and rerunning the game from my save at bald mountain, as the cutscene with the crones is a guaranteed crash. That will be the real test.

 

[ketxxx]

Don't worry about voiding a warranty just for removing the heatsink, a lot of manufacturers know thats going to happen and just don't care, likewise just don't mention you removed the heatsink either. The chances of you actually harming the card by removing the heatsink, cleaning it and the card then applying some fresh TIM is slim at best to none. Another way to look at it is if it is indeed overheating downtime will be around 30 mins instead of several weeks waiting on a RMA which its worth reminding people are not brand new cards from RMA unless you have a very posh extended warranty you paid a premium for somewhere in the price but merely repaired stock of other peoples faulty cards that have been RMAd.

 

[pagefault404]

Which is another reason why I want to avoid an RMA.

So after making the changes I mentioned, the crashes stopped becoming terminal. There are still the checkered square flickers, however now the game just crashes, and I can recover my OS from there. From here, it looks like a driver issue. Windows logs show errors from amdkmdap stopping and being recovered. This is definitely a step in the right direction as before this issue was terminal. Now I can recover my OS and actually troubleshoot. So far this is what I have learned about this issue:

Persists through driver updates and downgrades even using DDU
Persists after a windows reinstall
Persists after removing all user settings and saves, starting from a fresh witcher config and save folder
Changing tessellation from "AMD optimized" to "Override application settings - 8x" has extended the life of the application
Unplugging my rock candy xbox 360 controller has also assisted in extending the life of the application
Both above changes have changed the error from a terminal system halt to a crash to desktop, exposing the amdkmdap driver as the possible root cause of the issue.

However, researching the troubleshooting for amdkmdap, all anyone recommends is reinstalling drivers. This has been done previously and did not resolve the issue, so it looks like for the moment I have reached an impasse.

 

[ketxxx]

Not quite a impasse yet, from the info you have given it looks like your issue is either heat related or what would seem to be the more likely - your PSU starting to develop some kind of fault, the clue here is you saying unplugging your controller helped a bit. Its only a miniscule amount of power saved but if that really helped stability the PSU definitely isn't out of being the culprit. Disconnect all the hardware you can from the system to lessen the power load, set any OCd hardware back to stock speed and voltages then see if that improves your woes any further, if it does, its a step closer to narrowing down the PSU as the culprit.

 

[pagefault404]

So I downclocked my GPU 25% in CCC overdrive and it seems to have fixed the issue. So I am starting to agree with you that this is looking like either a bad OC or the PSU. Which would explain why things got better when I dropped tessellation, unplugged the controller and downclocked the card. The PSU is an off brand (pre built system manufacturers cutting corners) so I will shoot for this stability and report back if this does not work. All while saving for a new PSU.

Current suspected root cause: Overclocking power draw/possible faulty PSU

 

[pagefault404]

So I tested this when I got home from work. No change. Underclocked to 50%, still unstable. Unplugged all peripherals. No change.

My next course of action is to wait for 1.07 and see if that does anything for me.

 

[pagefault404]

So I found the Heaven and Valley benchmarks from Unigine software, and I noticed that Heaven was tessellation heavy. I ran Heaven and sure enough it crashed. So I revisited the downclocking solution, except I didn't stop at downclocking the core. I downclocked the memory as well.

I dropped the memory by 100MHz from the stock rate of 1250 to 1150 and started with a -25% downclock from 1000MHz to 750MHz.

Stable. Benchmark completed.

I raised the core clock from 750 to 900 and it started crashing again. So I dropped the clock to 850 from 900 and it's stable again. I can make it through both benchmark engines repeatedly at max settings.

Heaven Benchmark

Valley benchmark

So my resolution to this issue was to downclock my GPU, meaning it's possibly a power issue. I have yet to test this on the witcher 3 itself. I will test these system settings and report back.

 

[SoloCreep]

For anyone who overclocks their GPU's, keep one thing in mind. Overclocks stability can vary from one driver update to the next, sometimes they become unstable. While one card "brand/type" will not have any OC stability problems from updates, others may indeed need to adjust their clocks.

 

[pagefault404]

So I was able to play the game for two hours. Stable. No issues.

I had to downgrade to 15.6 and lower my clock rate. No matter what I changed my clock rates to in 15.7, the results were the same. Until I get my power supply replaced with something more reliable, I will keep these clocks.