GPU crashing

[pcme]

Hi, my pc arrived yesterday. I've been having problems with the GPU shortly after use. It initially ran like a dream for the first 2-3 hours before I got a driver timeout error & was forced to use task manager to close the game. ("We detected a driver timeout has occurred on your system. A bug report has been generated"). Loaded up the game again & got about a minute in before crashing with the same error. It's now an instant driver timeout on launching any game.

It came installed with the 20.12.1 Optional driver, I first tried using the AMD Utility Cleanup tool to remove that & install the 20.11.2 Recommended driver but the problem persisted.

I've since reinstalled windows but that didn't solve it either.

Since encountering the crashes I started monitoring the performance metrics in the Adrenalin software during. On some games where I can hit the lobby, I've noticed the GPU Utilization consistently spiking from 0 to 100 or close to it before getting the timeout error. The GPU temperature was usually in the mid 40s with the highest going just over 50. It was then I noticed Fan Speed at 0 RPM, I found a 'Zero RPM' option enabled by default on the Adrenalin software that I disabled. It dropped the GPU temperature to low 30s but timeouts continue.

I haven't got too adventurous inside the case, only going so far as to verify everything I could access was firmly connected.

Everything else runs extremely fast, it's just any load on the GPU now causes it to instantly crash.

Case	FRACTAL DEFINE 7 BLACK QUIET MID-TOWER CASE
Processor (CPU)	AMD Ryzen 9 5900X 12 Core CPU (3.7GHz-4.8GHz/70MB CACHE/AM4)
Motherboard	ASUS® ROG STRIX X570-F GAMING (USB 3.2 Gen 2, PCIe 4.0) - ARGB Ready!
Memory (RAM)	64GB Corsair VENGEANCE RGB PRO DDR4 3200MHz (4 x 16GB)
Graphics Card	16GB AMD RADEON™ RX 6800 - HDMI, DP - DX® 12
1st Storage Drive	2TB PCS 2.5" SSD, SATA 6 Gb (520MB/R, 470MB/W)
1st M.2 SSD Drive	1TB SEAGATE FIRECUDA 520 GEN 4 PCIe NVMe (up to 5000MB/R, 4400MB/W)
DVD/BLU-RAY Drive	NOT REQUIRED
Power Supply	CORSAIR 850W RM SERIES™ MODULAR 80 PLUS® GOLD, ULTRA QUIET
Power Cable	1 x 1 Metre UK Power Cable (Kettle Lead)
Processor Cooling	Corsair H115i RGB PRO XT Hydro Series High Performance CPU Cooler
Thermal Paste	STANDARD THERMAL PASTE FOR SUFFICIENT COOLING
Sound Card	ONBOARD 6 CHANNEL (5.1) HIGH DEF AUDIO (AS STANDARD)
Network Card	10/100/1000 GIGABIT LAN PORT (Wi-Fi NOT INCLUDED)
Wireless Network Card	WIRELESS INTEL® Wi-Fi 6 AX200 2,400Mbps/5GHz, 300Mbps/2.4GHz PCI-E CARD + BT 5.0
USB/Thunderbolt Options	MIN. 2 x USB 3.0 & 2 x USB 2.0 PORTS @ BACK PANEL + MIN. 2 FRONT PORTS
Operating System	Windows 10 Home 64 Bit - inc. Single Licence [KUK-00001]
Operating System Language	United Kingdom - English Language
Windows Recovery Media	Windows 10 Multi-Language Recovery Image - Unlimited Downloads from Online Account
Office Software	FREE 30 Day Trial of Microsoft 365® (Operating System Required)
Anti-Virus	NO ANTI-VIRUS SOFTWARE
Browser	Google Chrome™
Warranty	3 Year Silver Warranty (1 Year Collect & Return, 1 Year Parts, 3 Year Labour)
Delivery	2 - 3 DAY DELIVERY TO REPUBLIC OF IRELAND
Build Time	Standard Build - Approximately 13 to 15 working days

 

[Martinr36]

try uninstalling the driver using DDU, then install the latest driver from the AMD website

Display Driver Uninstaller (DDU) – Best Graphics Driver Removal Tool

Display Driver Uninstaller or DDU is a free graphics card driver removal tool for Nvidia and AMD graphics cards. DDU can completely remove video drivers if you cannot remove it from the windows uninstaller. Download Display Driver Uninstaller (DDU) for Windows. DDU is an effective and...

graphicscardhub.com

 

[pcme]

Hi Martin, thanks for the reply. I've just used DDU in safemode then installed the latest driver. Unfortunately problem is still there. Tested on 2 games, both crashed. I was only able to grab a screenshot of 1 since the Adrenalin software crashed with the 2nd game. This was just by hitting the Xcom2 lobby.

 

[Martinr36]

0k, check that the card is properly seated

 

[SpyderTracks]

Hi Martin, thanks for the reply. I've just used DDU in safemode then installed the latest driver. Unfortunately problem is still there. Tested on 2 games, both crashed. I was only able to grab a screenshot of 1 since the Adrenalin software crashed with the 2nd game. This was just by hitting the Xcom2 lobby.

Try uninstalling the driver again with DDU.

Then install 20.11.3 rather than the latest one, there are reports that was more stable.

https://www.amd.com/en/support/previous-drivers/graphics/amd-radeon-6000-series/amd-radeon-6800-series/amd-radeon-rx-6800-xt

 

[pcme]

0k, check that the card is properly seated

It looks pretty secure, not a lot of wiggle room when trying to adjust it. I can only see 1 set of wires from this side, I think to the power (8 wires) & that was firmly connected.

Try uninstalling the driver again with DDU.

Then install 20.11.3 rather than the latest one, there are reports that was more stable.

https://www.amd.com/en/support/previous-drivers/graphics/amd-radeon-6000-series/amd-radeon-6800-series/amd-radeon-rx-6800-xt

Hi Spydertracks, do you mean the 20.11.2 driver? I can only see 2 options under my card. It was the first driver I installed after encountering the issue but I can try it with DDU too.

https://www.amd.com/en/support/graphics/amd-radeon-6000-series/amd-radeon-6800-series/amd-radeon-rx-6800

 

[SpyderTracks]

It looks pretty secure, not a lot of wiggle room when trying to adjust it. I can only see 1 set of wires from this side, I think to the power (8 wires) & that was firmly connected.


Hi Spydertracks, do you mean the 20.11.2 driver? I can only see 2 options under my card. It was the first driver I installed after encountering the issue but I can try it with DDU too.

https://www.amd.com/en/support/graphics/amd-radeon-6000-series/amd-radeon-6800-series/amd-radeon-rx-6800

Really sorry, I'd linked to the 6800XT incorrectly, my mistake.

Yes, it would be the 20.11.2 one in that case, try it without installing Adrenaline as that may well immediately update it:

 

[pcme]

Really sorry, I'd linked to the 6800XT incorrectly, my mistake.

Yes, it would be the 20.11.2 one in that case, try it without installing Adrenaline as that may well immediately update it:

No problem, thanks for the advice, that's a clever trick

Unfortunately no joy though, game going unresponsive at lobby still and throws me back to the desktop

 

[SpyderTracks]

No problem, thanks for the advice, that's a clever trick

Unfortunately no joy though, game going unresponsive at lobby still and throws me back to the desktop

Hate to say it, but the next thing to try would be a clean install of windows just to be certain it's not some software conflict.

At this point, it is starting to sound like a hardware issue, but best to check the windows installation before jumping to that conclusion.

If you follow method one from this guide: https://www.howtogeek.com/224342/how-to-clean-install-windows-10/

You can create a bootable USB from here: https://www.microsoft.com/en-gb/software-download/windows10

At the very first stage where it asks where to install windows to, delete all existing system partitions. This will erase the whole system partition, so if you've got anything saved on the C drive, make sure to copy it off first of all.

 

[pcme]

Hate to say it, but the next thing to try would be a clean install of windows just to be certain it's not some software conflict.

At this point, it is starting to sound like a hardware issue, but best to check the windows installation before jumping to that conclusion.

If you follow method one from this guide: https://www.howtogeek.com/224342/how-to-clean-install-windows-10/

You can create a bootable USB from here: https://www.microsoft.com/en-gb/software-download/windows10

At the very first stage where it asks where to install windows to, delete all existing system partitions. This will erase the whole system partition, so if you've got anything saved on the C drive, make sure to copy it off first of all.

Thanks I'll give that a try, I had only used the reset windows option before. The windows 10 media is creating on a USB now.

Quick question from that guide while it's still setting up.

On the screenshot it will ask for product key, where will I find that? I've checked my PCS emails/welcome pack and don't think I've been given it, or can I just leave that field blank & windows will detect it's valid?

 

[SpyderTracks]

Thanks I'll give that a try, I had only used the reset windows option before. The windows 10 media is creating on a USB now.

Quick question from that guide while it's still setting up.

On the screenshot it will ask for product key, where will I find that? I've checked my PCS emails/welcome pack and don't think I've been given it, or can I just leave that field blank & windows will detect it's valid?

You won't need a product key, the key is actually linked to the hardware ID of the machine. When it asks for a key, there will be an option to continue without entering a key, just select that and it will automatically register with Microsoft server when you're next connected to the internet.

Also, once you're into windows, run windows updates repetitively until there are none left to apply. Then double check device manager and make sure all devices are installed. Then manually install 20.11.2 gpu drivers again and only install one game to test. Fingers crossed it will be a much better experience.

 

[pcme]

You won't need a product key, the key is actually linked to the hardware ID of the machine. When it asks for a key, there will be an option to continue without entering a key, just select that and it will automatically register with Microsoft server when you're next connected to the internet.

Also, once you're into windows, run windows updates repetitively until there are none left to apply. Then double check device manager and make sure all devices are installed. Then manually install 20.11.2 gpu drivers again and only install one game to test. Fingers crossed it will be a much better experience.

I have 6 partitions available, do I delete all before installing onto Drive 1 then?

Name - Type

Drive 0 Partition 1 - MSR (Reserved)
Drive 0 Partition 2: New Volume - Primary
Drive 1 Partition 1: SYSTEM - System
Drive 1 Partition 2 - MSR (Reserved)
Drive 1 Partition 3: Recovery - Recovery
Drive 1 Partition 4: Windows - Primary

 

[SpyderTracks]

I have 6 partitions available, do I delete all before installing onto Drive 1 then?

Name - Type

Drive 0 Partition 1 - MSR (Reserved)
Drive 0 Partition 2: New Volume - Primary
Drive 1 Partition 1: SYSTEM - System
Drive 1 Partition 2 - MSR (Reserved)
Drive 1 Partition 3: Recovery - Recovery
Drive 1 Partition 4: Windows - Primary

Just delete all the Drive 1 partitions and install to the remaining space. Drive 0 is your data drive which can be left as it is.

 

[pcme]

Just delete all the Drive 1 partitions and install to the remaining space. Drive 0 is your data drive which can be left as it is.

Still no good. Thanks for your help on the install instructions.

 

[SpyderTracks]

Still no good. Thanks for your help on the install instructions.

Are all devices accounted for in device manager?

 

[ubuysa]

Are the any minidumps in C:\Windows\Minidumps or a kernel dump is C:\Windows\Memory.dmp? If so upload any you find to the cloud and post a link here.

 

[pcme]

Are the any minidumps in C:\Windows\Minidumps or a kernel dump is C:\Windows\Memory.dmp? If so upload any you find to the cloud and post a link here.

Hi Ubuysa, there is no minidumps folder or memory.dmp file. I did see a memory.dmp file yesterday but I think it was deleted with the windows reinstall. I did only test 1 game since reinstalling windows though. Would you like me to reinstall the drivers with the Adrenalin software & launch a few games to generate a new file?

 

[ubuysa]

Hi Ubuysa, there is no minidumps folder or memory.dmp file. I did see a memory.dmp file yesterday but I think it was deleted with the windows reinstall. I did only test 1 game since reinstalling windows though. Would you like me to reinstall the drivers with the Adrenalin software & launch a few games to generate a new file?

Don't do anything that might make things worse.

If you get any further BSODs do upload the dumps.

 

[pcme]

Don't do anything that might make things worse.

If you get any further BSODs do upload the dumps.

I've only got 1 BSOD this whole time. The error didn't stay up long enough to grab a screen but was something along the lines of video_failure_something.

Every other time it's crashed it's just been the 'driver timeout' error on the AMD bug report tool that throws me back to the desktop. A lot of the time the Adrenalin software has crashed with it. Without the Adrenalin software it just sends me back to the desktop too.

The game that's caused the crash has always gone not responding while still hearing the audio progress in the background. For some of those I've had some heavy stuttering on my cursor too until I ctrl-shift-esc to end task on the game. As soon as task manager has opened the stuttering is gone.

 

[ubuysa]

I've only got 1 BSOD this whole time. The error didn't stay up long enough to grab a screen but was something along the lines of video_failure_something.

Every other time it's crashed it's just been the 'driver timeout' error on the AMD bug report tool that throws me back to the desktop. A lot of the time the Adrenalin software has crashed with it. Without the Adrenalin software it just sends me back to the desktop too.

The game that's caused the crash has always gone not responding while still hearing the audio progress in the background. For some of those I've had some heavy stuttering on my cursor too until I ctrl-shift-del to end task on the game. As soon as task manager has opened the stuttering is gone.

Kernel dumps (C:\Windows\Memory.dmp) are only written by the kernel when the system halts - a BSOD. Minidumps (in C:\Windows\Minidumps) are taken by drivers, Windows system components, and even by applications, and the failure is usually recovered and the system continues.

Thus you may find minidumps even if the system doesn't BSOD. Those would be useful too.