Windows 8 Windows 8.1 BSOD again and again

[iNarek94]

Hello,

Recently (a week ago) I have changed my motherboard, cpu and ram. At this moment I have the current configuration

Corsair 550W PSU
ASRock H170 Pro 4
Intel Core i5 6500
XFX R9 270X 4GB
2x4GB Corsair Vengeance DDR4 2133MHZ
120GB SSD Kingston V300
1TB WD HDD

After struggling with drivers I have managed to install windows 7. After around 5 days of using it and playing games like battlefield 4, gta 5 (which had no issues) I started receiving blue screens a few times and decided I should change to windows 10, knowing that motherboard and cpu didn't really support windows 7 (thought that would help). Now, I had windows 10 running for a day and didn't receive any BSOD but Gta 5 kept crashing after a few minutes of playing, and being unable to deal with compatibility / driver problems I switched to windows 8.1. Right after installation of windows, I got vcredist 2008-2013, installed latest amd drivers, msi afterburner, which, in turn, updated direct X and I also installed all windows updates that were due. Today I received 5 BSOD in a row, with around 3-5 minute intervals. I have ran memtest86 1-2 passes with no erros, just to quick check memory sticks, although they are new and I doubt they have any malfunction. I also removed my GPU and put it back again, checking that my PSU connectors were sitting firmly (I have heard this has solved a similar issue before, gpu's power supply was unstable and resulted in errors), so as of 3 hours I didn't get any BSOD, but I doubt this state will last longer than a day or maybe even less.

Mostly atikmpag.sys dxgkrnl.sys dxgmms1.sys are mentioned in BSOD, here is the minidump.

I really hope someone can help me solve this problem, as after 3-4 years I have finally managed to upgrade my system and instead of stable performance I get this.

 

[kemical]

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

Use !analyze -v to get detailed debugging information.

BugCheck 116, {ffffe0000762b4d0, fffff801fcd7abf8, 0, d}

Unable to load image \SystemRoot\system32\DRIVERS\atikmpag.sys, Win32 error 0n2
*** WARNING: Unable to verify timestamp for atikmpag.sys
*** ERROR: Module load completed but symbols could not be loaded for atikmpag.sys
Probably caused by : atikmpag.sys ( atikmpag+ebf8 )

Followup: MachineOwner

Hi,
almost all of your dump files were like the above apart from one which was a bugcheck 3B (also GPU related). Bugcheck 116 basically means the driver tried to reset but didn't respond within a given amount of time.
In cases like this we suggest running the DDU.
The Display driver Uninstaller has to be used in safe mode where it will uninstall not only the gpu driver but anything else left behind by previous installs.
DDU

Once removed please install the latest whql driver.
Link Removed

Try running Furmark for an hour or so. Please monitor temps using HWiNFO64
FurMark: VGA Stress Test, Graphics Card and GPU Stability Test, Burn-in Test, OpenGL Benchmark and GPU Temperature | oZone3D.Net

Hope the testing goes well. Please post any new dump files.

 

[iNarek94]

Thank you for replying, kemical, I have done the steps you mentioned, used DDU in safe mod, uninstalled AMD drivers and CCC, then manually removed AMD folder and after restarting reinstalled the latest drivers. After that I ran a stress test on 1080p, MSAA x2 settings for 1 hour with 100% gpu load and no errors or bsod. During gaming I usually have gpu under ~80% load and with temperature ranging from 60 to 84 or so. During this test the temperature rose up to 94C so I had to remove the side panel of my case in order to prevent any possible damage, the test kept running at around 86C again under 100% load and 100% fan speed. Right before reading your post I was playing gta 5 and got a red screen with a buzzing sound, PC restarted and I found another minidump, here it is. I should also mention that 3 of atikmpag.sys dxgkrnl.sys dxgmms1.sys crashes were received during casual web browsing, so I'm not sure what to think of :/

Also after getting a new system I started receiving "Aw snap" page crashes on google chrome quite frequently, not sure if this is of any help.

 

[kemical]

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

Use !analyze -v to get detailed debugging information.

BugCheck D1, {fffff8020f57a438, 2, 0, fffff801ff559cc0}

*** WARNING: Unable to verify timestamp for win32k.sys
*** ERROR: Module load completed but symbols could not be loaded for win32k.sys
Probably caused by : memory_corruption

Followup: memory_corruption

Hi,
the above dump file is basically caused by drivers using improper addresses. Unfortunately no driver was mentioned in the dump so it might be an idea to run the driver verifier.
However, this:

During this test the temperature rose up to 94C so I had to remove the side panel of my case in order to prevent any possible damage, the test kept running at around 86C again under 100% load and 100% fan speed.

This is a little too warm and could be 'part' of the problem.

Airflow is also important and sometimes a little cable routing can do wonders.

Can you tell us if your using a stock cooler or a third party version with the GPU?

Have you tried downclocking the gpu? If not then try running the card at around 100mhz lower.

Sorry I forgot about the verifier.
The verifier can induce a bsod loop on start up so make sure you have a way of getting into safe mode to turn it off. A recovery disk/USB is really handy. Creating a USB recovery drive - Windows Help

Driver verifier
How do I fix a Blue Screen in Windows using Driver Verifier?
Post any resulting dump files.

 

[iNarek94]

Hi,

I am using stock cooler, and, as you suggested, I improved my cable management and relocated one of the case fans near the gpu, now the temps are lower, running furmark with a closed case and 90% gpu fan use at around 85C. However, when I tried to change core clock in radeon settings, my displayed shut down and I had to restart the PC. Every single time that I tried to access radeon settings my display would shut down again so I uninstalled it, leaving the drivers only.

The error I found in event event logs stated the following

Faulting application name: RadeonSettings.exe, version: 10.1.1.1633, time stamp: 0x57509bb4
Faulting module name: MSVCR120.dll, version: 12.0.21005.1, time stamp: 0x524f83ff
Exception code: 0xc0000005
Fault offset: 0x000000000003c3f9
Faulting process id: 0xe30
Faulting application start time: 0x01d1cd77294b11ff
Faulting application path: C:\Program Files\AMD\CNext\CNext\RadeonSettings.exe
Faulting module path: C:\Program Files\AMD\CNext\CNext\MSVCR120.dll
Report Id: 81986999-396a-11e6-8269-d0509997b3e2
Faulting package full name:
Faulting package-relative application ID:

I also got one bsod with the same "atikmpag.sys dxgkrnl.sys dxgmms1.sys" when I uninstalled radeon settings and didn't restart yet.

 

[kemical]

I also got one bsod with the same "atikmpag.sys dxgkrnl.sys dxgmms1.sys" when I uninstalled radeon settings and didn't restart yet.

Please post any new dump files.

access radeon settings my display would shut down again so I uninstalled it, leaving the drivers only.

So from what you say above it sounds like you installed your gpu drivers through device manager and then installed the Radeon Software Crimson suite separately?

If this is the case then try using the DDU and then installing the full driver and not via device manager which as i pointed out isn't the most desirable of methods for exactly this reason. (unless of course if you prefer running without the manager)

 

[iNarek94]

So from what you say above it sounds like you installed your gpu drivers through device manager and then installed the Radeon Software Crimson suite separately?

Not really, I just ran the uninstaller and chose to remove software only, leaving the drivers installed. I would love to upload the dump file, but I have already installed windows 8.1 again, and as windows automatically downloaded and installed a 2014 version driver and software from radeon, I started getting those crashes again immediately, the only difference is PC doesn't restart and I get the "driver stopped responding and recovered" message from radeon. In the event log it looks like this "Display driver amdkmdap stopped responding and has successfully recovered."

 

[BIGBEARJEDI]

Looks to me like your PSU is underpowered and not a quality brand. A 4GB video card needs more juice; I'd replace with at least a 700Watt PSU such as Seasonic or Thermaltake. You might download the free HWINFO64 diagnostic and post back your voltage rails.

<<<BIGBEARJEDI>>>

 

[iNarek94]

I don't think Corsair is not a quality brand, also, if PSU was the case, I should have at least noticed it while doing a 1 hour stress test with a 100% gpu load, nevertheless I attached 2 screenshots of voltages.

To be honest, I am more biased towards searching for the problem in memory timings.

 

[kemical]

Your PSU should be plenty. Your card requires a 500W PSU and whilst Corsair maybe isn't the best they certainly aren't the worst.

I should have at least noticed it while doing a 1 hour stress test with a 100% gpu load,

Exactly which is one of the reasons you were asked to run it.

To be honest, I am more biased towards searching for the problem in memory timings.

Why?
As you've reinstalled 8.1 and got the old driver back, remove using the DDU as it automatically changes your update settings to not download a gpu driver
What concerns me is you tried to change the clocks and kept losing the display, remember that?

Can we try and just deal with one thing at a time?

So when you changed the clocks was it a big change you made or just small increments?

You could also try underclocking the card using a utility like AMD's Link Removed or MSI's Link Removed

 

[iNarek94]

I downclocked core from 1050MHz to 950 and I used Overdrive first and then, after uninstalling amd's software I used Afterburner. I have just checked my memory clocks through HWInfo and CPU_Z, I don't know why, but they are running at 1066MHz 15-15-15-36 (Is this because it displays half-speed?). The reason of my doubts was the solution found in this topic

 

[kemical]

I have just checked my memory clocks through HWInfo and CPU_Z, I don't know why, but they are running at 1066MHz 15-15-15-36.

What do you normally use to set the RAM speed? As your probably aware when you buy fast RAM you have to set the clocks yourself. Usually the stick will default to something like 1333MHz and you need to use XMP to run the RAM at the correct speed or just can clock it manually.
You say yours has dropped to 1066mhz?

Have you tried setting the RAM up to it's proper speed?

 

[iNarek94]

What do you normally use to set the RAM speed? As your probably aware when you buy fast RAM you have to set the clocks yourself. Usually the stick will default to something like 1333MHz and you need to use XMP to run the RAM at the correct speed or just can clock it manually.
You say yours has dropped to 1066mhz?

Have you tried setting the RAM up to it's proper speed?

I haven't ever touched RAM's part since getting it, because bios showed RAM running at it's nominal 2133MHz speed, it appears that cpu_z and other programs show half the speed (double data rate, in my case 2133/2 ~= 1066) and I didn't know that. I have just reinstalled AMD's drivers, played with clock values and had no crashes. Now I'm completely lost.

 

[kemical]

it appears that cpu_z and other programs show half the speed (double data rate, in my case 2133/2 ~= 1066) and I didn't know that.

Ah ok understood.

I have just reinstalled AMD's drivers, played with clock values and had no crashes. Now I'm completely lost.

So you reduced the clocks and the crashes have stopped.. That's good yes? I've now dealt with a number of users all with the same gpu and the only solution is to under clock the card.
How much did you drop the clocks by, just 50MHz?

 

[iNarek94]

Ah ok understood.

So you reduced the clocks and the crashes have stopped.. That's good yes? I've now dealt with a number of users all with the same gpu and the only solution is to under clock the card.
How much did you drop the clocks by, just 50MHz?

by 100MHz, and, unfortunately, another crash happened now, so the problem isn't solved yet, but I will keep monitoring different clock values. The thing is, I have been using this gpu for 1.5 years, always running at stock clocks and never had issues, but at that time I had AMD A10 5800K cpu, which was bottlenecking the gpu and, as stated in my original post, wouldn't let it use it's potential. Also I have been running programs, games on windows 10 on my old build without any errors, which take place now with my current build. So the gpu may be the problem, but cpu/mobo/ram could be too. I will post updates if I solve this problem, thank you a lot for your help. If you have any other suggestions, please share them with me.

 

[iNarek94]

Ok, as you stated, I have found other topics where people with R series are having either high clock or low voltage problems, which sounds logical, only, in my case I have been getting crashes like that during casual web browsing too, gpu's clock was at 600MHz, so these two facts don't really match. I feel like I'm approaching a dead end.

 

[kemical]

Is it possible to try the card in a different machine? Perhaps a friends?

 

[iNarek94]

Is it possible to try the card in a different machine? Perhaps a friends?

I still have my old rig parts, so I can rebuild it and run some tests.

 

[kemical]

Might be a good idea? The more testing you can throw at it the better.

 

[RichM]

Have you got any other ram try in it? I find it suspicious in looking at the tested ram for that board only 1 Corsair model is shown and that is "Value Ram" so I would suspect the ram itself is incompatible.