Windows 10 Thread stuck in device driver - RX480

[phufbq]

Hello,

I'm having trouble with random system crashes. I get loads of different varieties:

1. Thread stuck in device driver (BSOD) with dumps files.
2. Red screen of death.
3. Screen going black, but with motherboard CPU and DRAM lights blinking on Gigabyte AB-350-Gaming 3 mobo.

The dump files when I view them in BlueScreenView have two things highlighted in red:

atikmdag.sys
dxgkml.sys

I have tried two different AMD drivers for the GPU. One does seem more stable than the other (but neither are entirely stable).
Less stable - 17.7.2 (recommended by AMD Radeon Settings).
More stable - 17.6.2 (recommended/tested by MSI).

Hoping that someone will be able to help me debug what is going on.
Regards,
Phil

 

[Neemobeer]

Make sure to upload your dump files so they can be analyzed. How to ask for help with a BSOD problem

 

[phufbq]

Thanks, Neemobeer

Here is the zip archive with all the diagnostic files in it.

 

[Neemobeer]

Certainly looks like the thread is stuck in driver atikmdag
2: kd> .thread
Implicit thread is now ffff8103`71305080
2: kd> kb
# RetAddr : Args to Child : Call Site
00 fffff806`f9935ffb : 00000000`000000ea ffff8103`71305080 00000000`00000000 00000000`00000000 : nt!KeBugCheckEx
01 fffff806`f99360ce : ffffbf80`f5005c38 fffff806`fe4cb1f0 ffffbf80`f5005c38 ffffbf80`f5005d20 : dxgkrnl!TdrTimedOperationBugcheckOnTimeout+0x3f
02 fffff806`fe459ea0 : ffff8103`6e091000 00000000`00000000 ffffbf80`f5005d20 fffff806`fe4cb1dc : dxgkrnl!TdrTimedOperationDelay+0xbe
03 ffff8103`6e091000 : 00000000`00000000 ffffbf80`f5005d20 fffff806`fe4cb1dc 00000000`00002710 : atikmdag+0x59ea0
04 00000000`00000000 : ffffbf80`f5005d20 fffff806`fe4cb1dc 00000000`00002710 ffff8103`6eff0028 : 0xffff8103`6e091000

This could very well be a bad GPU and the thread could be waiting for the hardware to become ready/idle

 

[nmsuk]

I have a RX480 best drivers for it are 17.4.4 many are having your issue and the driver version I've mentioned seem to fix it for most.

 

[kemical]

*******************************************************************************
*                                                                             *
*                        Bugcheck Analysis                                    *
*                                                                             *
*******************************************************************************

Use !analyze -v to get detailed debugging information.

BugCheck 100000EA, {ffff810371305080, 0, 0, 0}

*** WARNING: Unable to verify timestamp for atikmdag.sys
*** ERROR: Module load completed but symbols could not be loaded for atikmdag.sys
Probably caused by : dxgkrnl.sys ( dxgkrnl!TdrTimedOperationBugcheckOnTimeout+3f )

Followup:     MachineOwner

DEFAULT_BUCKET_ID:  GRAPHICS_DRIVER_FAULT

Hi,
this is, as Neem's say's above, the driver endlessly spinning awaiting a response.
Whilst it can be a hardware issue I would first remove your driver using the DDU or Display driver Uninstaller to clear out any existing corruption by former installs:
DDU
(You'll need to boot into safe mode to use the DDU effectively)

As Rich say's above try the driver found here:
Radeon Software Crimson ReLive Edition 17.4.4 Release Notes

Can you also post the make and model of your PSU (power supply unit).
This driver is known for bsod's:
AppleCharger.sys Mon Oct 28 01:59:21 2013: GIGABYTE On/Off Charge driver. Known to cause bsod's on some systems. If the blue screen persists remove to test.

Lastly, all your dump files were the same bugcheck as above but what was also interesting the process running each time the crash happened was:

PROCESS_NAME:  GvLedServices.exe

GvLedServices.exe Apparently belongs to RGB Fusion which is a Gigabyte app (I've used it myself). Try removing this and see if the crashing stops.

Post back with any new dump files.

 

[phufbq]

Thanks for all the help.

I did the following:

1. Uninstalled GPU drivers using DDU and a safe boot.
2. Installed 17.4.4 drivers as requested.
3. Deleted the RGB Fusion software, and the charge on/off software (and most of the Gigabyte app center applications). There is now GvLedServices.exe running now.
4. Deleted the AppleCharger.sys, and the AppleChargeSrv.exe. They are in the recycling bin for now.

My PSU is a Corsair TX550M (550W Gold rated). I thought this would be sufficient.
It's powering:

• Gigabyte AB350-Gaming 3
• Ryzen 5 1600 (Stock settings)
• 2 x 8GB Corsair Vengeance RGB
• MSI Gaming X RX480 8GB
• One OCZTechnology SSD drive
• Case fans on Corsair 460X Crystal case. 3 x 120mm led fans and 1 x 120mm rear exhaust fan.
• One Corsair H100i v2 integrated water CPU cooler - 2 extra 120mm fans on this.

My GPU cannot successfully get through MSI Kombuster stress tests. I will see what happens with 17.4.4 and report back.

 

[kemical]

My PSU is a Corsair TX550M (550W Gold rated). I thought this would be sufficient.

According to Guru 3D 450W is sufficient so you should be fine:

This is Guru3D's generic power supply recommendation for the series:

• AMD Radeon RX 480 - On your average system the card requires you to have a 450 Watt power supply unit as minimum.

Ref:
MSI Radeon RX 480 GAMING X review

My GPU cannot successfully get through MSI Kombuster stress tests. I will see what happens with 17.4.4 and report back.

If this is the case then you could be looking a bad gpu.

Have you been overclocking the card at all?

 

[kemical]

@nmsuk actually owns a 480 so he might have further insights re your card?

 

[nmsuk]

The black screen, does it do this at idle? The other issues are sounding to me to be PSU related, though could just be a faulty GPU. If the card has a bios switch move that while the card is off and then boot and test.

 

[kemical]

I will see what happens with 17.4.4 and report back.

Any news?

 

[phufbq]

I'm happy to report that the above issue has been resolved. No issue with my GPU/PSU/CPU.

It must have been the GvLedServices.exe.

I've been able to install the latest AMD GFX driver now without any issue. Which is case anyone is interested was 17.9.1.

 

[kemical]

It must have been the GvLedServices.exe.

Well that's good news at least.

I had RGB Fusion (GvLedServices.exe) installed myself but didn't have any issues with it although I've since uninstalled it. These Aorus motherboards may have a multitude of LED's but you can only have one colour at any one time, so mine is permanently set to Deep red.