BSODs related graphics card and drivers

Hi. I have a R9 380 GPU and i suffer from some black screen issues/TDR crashes/red screens/BSODs. I had run memtest86, furmark, prime95 and other stress tests a lot of time; and they couldn't detect any defections. Most of the crashes are happened in the games but the last time driver crashed while browsing. Then couple of minutes later  VIDEO_SCHEDULER_INTERNAL_ERROR BSOD appear. Before this i had 4 BSODs in last months; 3 of them happened with blank red screen.

Unfortunately some of them are erased and i uploaded them:

Please run this report collecting tool so that we can provide a complete analysis: (from the pinned topic at the top of the forum)   http://www.bleepingcomputer.com/forums/t/576314/blue-screen-of-death-bsod-posting-instructions-windows-10-81-8-7-vista/
FYI - I don't often use the Perfmon report, so if it doesn't work please just let me know.
NOTE:  On problem systems it can take up to 20 minutes for the log files to complete.  Please be patient and let it run.

If you still have problems with it running, there's an alternate tool here (direct download link):  https://github.com/blueelvis/BSOD-Inspector/releases/download/1.0.5/BSODInspector-1.0.5.exe

NOTE:
Please zip up the (.ZIP) files - do not use .RAR or other compression utilities. 
.ZIP is the type file that can be uploaded to the forums.

Here's a page I made for Video TDR errors:  http://www.carrona.org/VideoTDR.html

There's only 3 memory dumps, but each one is a different error code.  Most often this is suggestive of a hardware problem - so please start with these free hardware diagnostics:  http://www.carrona.org/hwdiag.html

There's no need to repeat the tests if you've already run them recently - but please complete them all and let us know the results.

Start with the video tests (that you haven't run yet), as all 3 of the memory dumps have video in common.

Analysis:
The following is for information purposes only. The following information contains the relevant information from the blue screen analysis:
**************************Wed Dec  7 16:13:04.342 2016 (UTC - 5:00)**************************
Loading Dump File [C:\Users\john\SysnativeBSODApps\120816-14921-01.dmp]
Windows 10 Kernel Version 14393 MP (6 procs) Free x64
Built by: 14393.447.amd64fre.rs1_release_inmarket.161102-0100
System Uptime:0 days 3:53:11.071
Probably caused by :dxgmms2.sys ( dxgmms2!VidSchiProcessIsrCompletedPacket+e453 )
BugCheck 119, {1, c402e, c4031, ffffc7832eef5010}
BugCheck Info: VIDEO_SCHEDULER_INTERNAL_ERROR (119)
Arguments:
Arg1: 0000000000000001, The driver has reported an invalid fence ID.
Arg2: 00000000000c402e
Arg3: 00000000000c4031
Arg4: ffffc7832eef5010
BUGCHECK_STR:  0x119
DEFAULT_BUCKET_ID:  WIN8_DRIVER_FAULT
PROCESS_NAME:  System
FAILURE_BUCKET_ID: 0x119_1_DRIVER_REPORTED_INVALID_FENCE_ID_dxgmms2!VidSchiProcessIsrCompletedPacket
  BIOS Version                  2501
  BIOS Release Date             04/07/2014
  Manufacturer                  To be filled by O.E.M.
  Baseboard Manufacturer        ASUSTeK COMPUTER INC.
  Product Name                  To be filled by O.E.M.
  Baseboard Product             M5A97 R2.0
¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨``
**************************Sun Nov 13 15:46:32.008 2016 (UTC - 5:00)**************************
Loading Dump File [C:\Users\john\SysnativeBSODApps\111316-11562-01.dmp]
Windows 10 Kernel Version 14393 MP (6 procs) Free x64
Built by: 14393.447.amd64fre.rs1_release_inmarket.161102-0100
System Uptime:0 days 7:09:30.737
Probably caused by :dxgmms2.sys ( dxgmms2!VIDMM_GLOBAL::UpdatePageTable+360 )
BugCheck C2, {7, 0, 0, ffff820e2bffef60}
BugCheck Info: BAD_POOL_CALLER (c2)
Arguments:
Arg1: 0000000000000007, Attempt to free pool which was already freed
Arg2: 0000000000000000, Pool tag value from the pool header
Arg3: 0000000000000000, Contents of the first 4 bytes of the pool header
Arg4: ffff820e2bffef60, Address of the block of pool being deallocated
BUGCHECK_STR:  0xc2_7
PROCESS_NAME:  MirrorsEdgeCatalyst.exe
FAILURE_BUCKET_ID: 0xc2_7_dxgmms2!VIDMM_GLOBAL::UpdatePageTable
  BIOS Version                  2501
  BIOS Release Date             04/07/2014
  Manufacturer                  To be filled by O.E.M.
  Baseboard Manufacturer        ASUSTeK COMPUTER INC.
  Product Name                  To be filled by O.E.M.
  Baseboard Product             M5A97 R2.0
¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨``
**************************Fri Oct 28 10:38:39.364 2016 (UTC - 5:00)**************************
Loading Dump File [C:\Users\john\SysnativeBSODApps\102816-12812-01.dmp]
Windows 10 Kernel Version 14393 MP (6 procs) Free x64
Built by: 14393.351.amd64fre.rs1_release_inmarket.161014-1755
System Uptime:0 days 5:27:11.093
*** WARNING: Unable to verify timestamp for win32k.sys
*** ERROR: Module load completed but symbols could not be loaded for win32k.sys
Probably caused by :dxgmms2.sys ( dxgmms2!VIDMM_LINEAR_POOL::FindBlockRun+c0 )
BugCheck 1000007E, {ffffffffc0000005, fffff8092cf971a0, ffff9d00993bf098, ffff9d00993be8c0}
BugCheck Info: SYSTEM_THREAD_EXCEPTION_NOT_HANDLED_M (1000007e)
Arguments:
Arg1: ffffffffc0000005, The exception code that was not handled
Arg2: fffff8092cf971a0, The address that the exception occurred at
Arg3: ffff9d00993bf098, Exception Record Address
Arg4: ffff9d00993be8c0, Context Record Address
PROCESS_NAME:  System
BUGCHECK_STR:  AV
DEFAULT_BUCKET_ID:  WIN8_DRIVER_FAULT
FAILURE_BUCKET_ID: AV_dxgmms2!VIDMM_LINEAR_POOL::FindBlockRun
¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨``



3rd Party Drivers:
The following is for information purposes only. My recommendations were given above. The drivers that follow belong to software or devices that were not developed by Microsoft.  You can find links to the driver information and where to update the drivers in the section after the code box:

**************************Wed Dec  7 16:13:04.342 2016 (UTC - 5:00)**************************
AsIO.sys                    Mon Aug 23 21:53:02 2010 (4C7325FE)
HWiNFO64A.SYS               Tue Mar 31 05:51:32 2015 (551A6E24)
rt640x64.sys                Tue May  5 12:21:03 2015 (5548EDEF)
mbam.sys                    Tue Aug 11 13:35:19 2015 (55CA3257)
amdacpksd.sys               Thu Aug 11 04:33:48 2016 (57AC386C)
AtihdWT6.sys                Tue Sep 20 19:17:24 2016 (57E1C384)
atikmpag.sys                Tue Nov 15 15:36:17 2016 (582B71C1)
atikmdag.sys                Tue Nov 15 15:53:27 2016 (582B75C7)
¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨``
**************************Sun Nov 13 15:46:32.008 2016 (UTC - 5:00)**************************
atikmpag.sys                Tue Oct  4 21:29:28 2016 (57F45778)
atikmdag.sys                Tue Oct  4 22:51:46 2016 (57F46AC2)
¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨``
**************************Fri Oct 28 10:38:39.364 2016 (UTC - 5:00)**************************
GPU-Z.sys                   Wed Oct  6 12:14:37 2010 (4CACA06D)

Thank you for your reply and the advices that you have given.

I have ran nearly all the stress tests(i didn't mentioned in my first post but i tried video memory stress test, OCCT, memtestCL too). The only tests that i didn't try in the link you gave me are Hot CPU Tester Pro 4, artifact locator. I'll try them.

I couldn't add resmon's report i waited more than 20 mins and i cancelled the process. I also add Mirror's Edge Catalyst's own crash dump. It created on another crash(while system gave TDR and succesfully recovered). The two of red screens happened in this game.

 SysnativeFileCollectionApp.zip   2.02MB   3 downloads

 DESKTOP_HH247R6_12_8_2016_2_57_58_PM.zip   4.54MB   1 downloads

 CrashDump_2016.10.30_18.32.19.506.zip   50.2KB   1 downloads

My rig is ASUS R9 380 STRIX+AMD FX-6300+ASUS M5A97 R2.0 Mobo+Corsair Vengeance 2*4 GB 1600 mhz cl9+samsung 850 evo 250 gb ssd+seagate stdm1000 1 tb hdd, and i have Corsair CX600, which delivers 12V via one channel. It should be enough for this system, but somehow i have suspects about it.

Additional informations:

1/I have 3 PAGE_FAULT_IN_NON_PAGED_AREA errors which are only happened in Rise of the Tomb Raider DirectX 12 mode. DX11 have terrible performance(which is normal) but it never gave that error. This error always happened in the first 15 minutes of the game; if the system survive in that time, never gave these BSODs but sometimes gave DXGI ERROR DEVICE REMOVED error messages.

2/These crashes sometimes happen in particıular situations. For example if i play Batman:Arkham Knight @1080p it always crashes with black screen/TDR. It's not an uncommon issue on Tonga GPUs.

3/If you have time could you please review these articles:

http://www.pcinvasion.com/r9-380-black-screen-issues-amds-persistent-unsolved-bug

Looks to me it's hardware problem but none of the synthetic stress tests can trigger the fault. The crashes are very random.

Edited by numberofthebeast, 08 December 2016 - 07:46 AM.

The article about the black screen errors seems to be a bit biased towards this being only an AMD problem.

This is not the case.  Black screen errors occur on all platforms and with all video cards.
They have so many different causes that I no longer even try to troubleshoot them.

At work we do simple hardware diagnostics and then try System Restore.

If that doesn't work, then we suggest a backup and wipe of the system.

Now, just to add confusion to this problem, the video hardware tests that are most often used to test video cards are (IMO) no longer able to test everything in today's video cards.

This doesn't mean it's a problem with the tests - rather that the video card manufacturer's have added features, but they have no tests that will detect problems for these new features.

Still, the only reliable, 100% certain "test" is to replace a defective component with one that's known to be good - but this can get awfully expensive if you're trying to rule out multiple components.

In the event that you don't have a known good card to try, I would suggest purchasing one from a shop that will let you return it for your money back (if it's not needed).

Please uninstall MirrorsEdgeCatalyst - that's what that strange memory dump blames (CrashDump_2016.10.30_18.32.19.506.dmp is a user mode dump, not a kernel dump.  And I'm not familiar with user mode dumps).

Your UEFI/BIOS (version 2501) dates from 2014.  Please check at the manufacturer's website to see if there are any UEFI/BIOS updates available for your system.  If you are able to install the update through Windows (without booting from an external drive), then go ahead and update it.  WARNING - if the computer might shut down during this procedure, please don't do it, as this may physically damage the computer and prevent it from booting.
FYI - W8 and W10 communicate more with the UEFI/BIOS than previous versions of Windows, so it's important to ensure that the UEFI/BIOS is kept up to date (and that outdated UEFI/BIOS' may be the cause of some compatibility issues).

Although you appear to have a reasonable number of Windows Update hotfixes for this version of your OS, please double check for any new Windows Updates.  It only takes one update to cause a problem, so it's essential that you have all of them.  The actual number is not important.  Rather it's important that you checked manually, installed any available updates, and didn't experience any errors when checking or updating.

C: drive only has about 7% free space, D: drive has less than 8% free space, E: drive only has about 1% free space. .  Windows likes 15% free space in order to perform stuff "behind the scenes" without adversely affecting the system's performance.  Please free up 15% on ALL hard drives (you can get away with 10% on larger drives and won't notice a large performance penalty).  Low free space can cause BSOD's - but the actual amount depends on the files being used by the system.

Several TDR errors in the WER section of the MSINFO32 report - be sure that you try this procedure:  http://www.carrona.org/VideoTDR.html

Thanks for your very informative replies.

The black screens and signal losses can happen in every graphics card but the article has one interesting point: The people who had black screen problems with their Tonga Gpus, some of them RMAd their cards and get a new one(new card but the same model) and usually this didn't resolve their problems. Every card could be defective, in fact R9 series relatively high Dead on Arrival rate. Is my card faulty since the day one? Probably, but the problem is reseller couldn't replicate my problem when i RMAd it. 

I forgot the ask about the disk spaces, because most of my BSODs happened when my system was short of free space on C: drive. I suspected about it before but TDR errors don't have any correlation with low disk space. I want to ask another question: Should i disable virtual memory? Some people suggested that because according to their observations most of the BSODs somehow related to virtual memory; even the real cause is hardware failure or drivers. 

I think TDRs and BSODs are caused by different problems. Could be multiple defects on my system. Or maybe the real cause is my PSU? I have Corsair CX600 and looks like it's sufficient for my rig; but i don't know how to test it. 

I'm gonna test the card on one of my friends' system but the most difficult thing is probably i am not be able to replicate the same conditions. These crashes are very random and they have very interesting patterns. Obviously there was a bug(even the card or drivers) but it can only be triggered by certain conditions. If we don't detect any problem while gaming that won't be mean that my card is not defective. 

Edit reason: Correcting some typos and telling additional details.

 

Edited by numberofthebeast, 10 December 2016 - 12:03 PM.

I primarily deal with stock systems, and avoid changing settings as much as is possible.

It is my opinion that if the component doesn't work properly at stock settings (all software issues aside) - then it's defective.

If the seller cannot replicate the problem, then there's a couple of possibilities:

- the card is damaged, but the replication testing process isn't accurate

- that the card is not damaged, but hardware circumstances in your system are different than those of the seller's test apparatus (either hardware or compatibility issues).

- that the card is not damaged, and that the problems are software related (single problem)

- that the card is not damaged, and that the problems are software related (multiple interacting problems)

There are 2 types of BSOD errors (talking about how problems affect them).

There are those that stem  from a single cause.

There are those that stem from multiple interacting problems.

There is a difference in how graphics drivers interact with the system and how other drivers interact with the system.  The graphics drivers actually have a lot more access to the stuff in the kernel space than most other drivers do.

The TDR is a special sort of error that spits out different BSOD's depending on the event that generated it (see this page for more details on TDR: 

Disk space - IMO lack of free space should not cause a problem.

In reality, I have seen several cases where it does cause BSOD's

My personal theory is that it's caused by 3rd party drivers that can't handle storage space/virtual memory problems.

As it's not easy to prove this problem, I always check free space in the reports and suggest fixing it if you are having BSOD's or other errors.

Finally, you have to understand that vendors make video cards in order to make money.

IMO it's why there aren't new video card tests - simply because it costs too much to develop a test for a card that only has a 3 year (or less) support life cycle.

Also, there are other, business related, decisions that affect their support.  For example, it's cheaper to RMA a card than it is to test it thoroughly.

Thanks for your reply; i know why every tests(even the manufacturer's tests) are insufficient and this problem may caused by design error(every hardware have bugs; and most of the graphics card bugs can patch via software-for example Skylake Prime95 issue is fixed by BIOS update- and i know the privileges of graphics driver. And i'm nearly sure that my issue is caused by multiple reasons(combination of a lot of factors) My final questions are

1/should i disable or manually configure the virtual memory? 

2/What do you think Clockblocker's very positive effect on my situation and the some people's claim that this probably somehow related to downclocking/power saving system? I observed that tool really have an effect when it's running. Are there any additional power efficiency/saving settings that i have to tweak inside Windows/registry/BIOS? 

1) it won't hurt to disable or manually configure virtual memory - although the system may not work right then (but it's easy to change back)

Long ago I went without virtual memory on a system of mine (that had lot's of RAM) - all was well until I went to play a game.  Without virtual memory the game wouldn't run - so I enabled the virtual memory and haven't looked back since then.

2) As I said previously, I don't like to tweak systems in order to make them work properly.  It's my opinion that, if you need to tweak to make it work, then something is broken.

You pay for a part that's supposed to work - why should you have to fiddle with it to make it work properly?  Would you buy a car like this?

But, if it works for you, I'd say it's probably OK to use it.  You can find tweaks all over the web for tweaking power settings in Windows.
The only "tweak" that I use is to set my system (when plugged in) to settings that relate to it being Always On.  Beyond that, I don't even mess with the Fast Startup option anymore (haven't since W7)

Lastly, I'd suggest posting in some of the overclocking forums - as those people have much more experience with tweaking settings thatn I do.

Good luck!