Question Sudden BSOD while using Waifu2x Extension GUI

[ash_blake]

I was trying out Waifu2x and experimenting with different videos. Found a 8K video on YouTube and decided to upscale it's quality while keeping the resolution same. The app was probably extracting frames when the BSOD happened with "Unexpected Store Exception" code. My GPU is RX 6700 XT with 12GB VRAM. I had 18% VRAM reserved on Waifu2x settings so I doubt VRAM max utilization caused it. Wasn't paying attention to Adrenalin metrics unfortunately. My ram is 16GB and Processor is Ryzen 9 5900X.

Analyzing the dump showed this:

PROCESS_NAME: MemCompression

READ_ADDRESS: 000000000000aa00

ERROR_CODE: (NTSTATUS) 0xc0000005 - The instruction at 0x%p referenced memory at 0x%p. The memory could not be %s.

EXCEPTION_CODE_STR: c0000005

EXCEPTION_STR: 0xc0000005

Is this an issue with the app using way too much RAM? This is the first time I faced any BSOD after changing my faulty PSU 4 months ago so I am a bit worried and needs your opinion.

Here is the link to the memory dump file. Can you please analyze it for me?

https://www.dropbox.com/sh/l86f6sgvi62n086/AAAm9HSTauGeCtOHNWG8gQ4Ha?dl=0

 

[Colif]

what storage drives do you have?
The store in Unexpected Store Exception is used by the CPU to track file locations in virtual memory. The actual files are in one of 2 locations, in ram or on Page file. Page file is on C drive, normally

So I would check your C drive, most makers have software you can run to check their drives out.

So its not ram or VRAM, although it can just as easily be a driver error.

I will look at dumps but @johnbl can probably tell you more

if its moving a large file around, its likely the drive.


File: MEMORY.DMP (Mar 8 2023 - 06:09:28)
BugCheck: [UNEXPECTED_STORE_EXCEPTION (154)]
Probably caused by: memory_corruption (Process: MemCompression)
Uptime: 0 Day(s), 3 Hour(s), 57 Min(s), and 35 Sec(s)

i haven't done this in a few months so naturally windows 10 on my VM I use to convert these needs to update. I got that result above out and then it froze up. I will add more if its useful.

report - run as fiddle to read (mostly for me)
mostly virtual memory operations. Possibly one part of the file didn't match what was expected.

 

[johnbl]

overall I would update the amd chipset driver and
the raid driver from here:
MPG X570S CARBON MAX WIFI (msi.com)
best to do it with a bios update. most likely the root problem that caused the bugcheck.

here are other notes/issues from your memory dump
-------------
since you provided a memory.dmp file I can see more details and I am looking at why your video upscale tool did not complete. I am looking at a disk log and it has 2.8 million entries basically calling
PauseAdapter
then ResumeAdapter
all on port 2.
I guess I would ask if this is a raid disk, and if port 2 is a special port for your motherboard. It seems strange to making these calls over and over.
timer shows 11:09:28.338
11:09:28.346
.008 seconds for each cycle.
something is wrong with the drive/controller.
maybe update the BIOS, the CPU chipset, any raid driver and check for drive firmware updates using crystaldiskinfo.exe

it could explain why the upscale .exe is not getting done fast enough. debugger indicates it is a raid driver running.
there are repeated calls to storport!RaidAdapterPowerDownDevice
in the stream of log entries.
Device overheated? problems in sleep functions?
(something is wrong, maybe a raid adapter going to sleep and not waking correctly? might tell the system not to let your drive sleep. ie windows control panel, device manager, right mouse click on the adapter, see if there is a power management tab and uncheck the option to let the device sleep (just for testing))
------------------
I would looking at ffmpeg_waifu2xEX.exe and find out if I wanted it running (ie actually using it) and why it ended up using so much physical ram and pagefile.sys space. If you expect it to use a lot, I would make the pagefile bigger.
------------------
ffmpeg_waifu2xEX.exe looks like it was running and using a very large amount of memory. window memory compression tried to do some compressing on some of the 3 of the 80 threads and a fault was called.
debugger indicates that the .exe was using
10575600KB of memory space.
VirtualSize 26750 Mb
looks like the tool is some sort of video upscaler/editor
you could reduce the number of threads. I do not know why microsoft was trying to run memory compression on these threads. (other than the system was low on memory) could just be an app bug where one thread does not finish and eats up the ram and virtual memory.
later windows detects the limited free memory attaches to the process and tries run the memcompress on it. it still should not bugcheck, maybe the process did not complete due to some error and was hung?
process used a lot of kernel time and was switched in over 150k times. you have other suspect issues with the machine but this .exe was what lead to the bugcheck.


note 3: owning process was ffmpeg_waifu2xEX.exe
Image: MemCompression attached to it and it led to the bugcheck.

ffmpeg_waifu2xEX.exe has a very large number of threads. three of the threads were in the process of getting a memory compression error. cpu core 13,17 and 20
cpu core was trying to trim the working set for ffmpeg_waifu2xEX.exe looks like it also was in the process of a fault.


note 2: you could also update the BIOS to a more current version (beta or last stable version) then do a clean windows install and not install the ryzenmaster driver and dump any malware at the same time.

note: you have a old version of the bios running that does not have the various security fixes and you have a old version of the ryzenmaster driver running.
(you should update both the driver and bios together)
current beta bios released on 2023-03-03

windows was trying to remove some data from compressed memory, the data had been modified and it ended up using a bogus
memory address ( 000000000000aa00 )
A address like this could be an indication of malware trying to infect your system. I would guess it is coming from your internet download manager but it does look like you have several replaced drivers.

I would, use autoruns to remove the download manager.
turn off the system virtual memory to delete the pagefile.sys.
run cmd.exe as an admin then run
dism.exe /online /cleanup-image /restorehealth

then reboot and turn on virtual memory again to create a new pagefile.sys.
at this point I would run a malware scan to see if i could find any malware on your system.



download and run microsoft autoruns64 from here:
Autoruns for Windows - Sysinternals | Microsoft Learn
find the driver:
\SystemRoot\system32\DRIVERS\idmwfp.sys Wed Dec 19 06:52:11 2018
and delete it. (internet download manager by by Tonec )

other suspect drivers:
\SystemRoot\System32\Drivers\Fs_Rec.sys unavailable (FFFFFFFE)
\SystemRoot\System32\drivers\volmgrx.sys unavailable (FFFFFFFE)
Msfs.SYS unavailable (FFFFFFFE)

some windows files have been modified:
\SystemRoot\System32\win32kfull.sys

note:
seems weird to have Intel Telemetry Driver and Intel Power Engine Plugin installed on a amd based motherbooard

AMDRyzenMasterDriver.sys Thu Jun 24 22:21:58 2021 running.
--------------
machine info:
BIOS Version 1.40
BIOS Starting Address Segment f000
BIOS Release Date 08/11/2022
Manufacturer Micro-Star International Co., Ltd.
Product MPG X570S CARBON MAX WIFI (MS-7D52)
Processor Version AMD Ryzen 9 5900X 12-Core Processor
Processor Voltage 8bh - 1.1V
External Clock 100MHz
Max Speed 4950MHz
Current Speed 3700MHz

 

[Colif]

op replied to me via PM

"1TB SSD, partitioned in 2 drives, C (191GB) and D. Checked device manager. All drivers showed to be working properly. I must add that my Adrenalin drivers are quite old, 22.5.1. Didn't update in ages after 22.10.2 wrecked havoc in my system lol and installed the old stable one after the PSU incident. I checked windows error checking tool (the one you can access from C/D drive properties) and it showed no error. Checked SSD health a few days ago with CrystalDiskInfo and it showed 100% healthy and no bad blocks. So what do you suggest I should do then? I must mention one thing. After the BSOD I rebooted my PC and used for another 5 hours, even used CPU/GPU intensive softwares like Cupscale. Nothing else went wrong. Although I must mention that ever since the Windows Update of February, I'm noticing that the Windows news and interest bar (the one that shows weather in the taskbar) is acting up every once in a while. Need to restart Windows Explorer to make it work properly whenever it happens. Not sure whether these two are connected or not."

but only relates to my answer. He couldn't post here, not sure why.
edit: seems it was caused by running ram in XMP. He tested in memtest, errors in XMP, none out of XMP.

 

[ash_blake]

overall I would update the amd chipset driver and
the raid driver from here:
MPG X570S CARBON MAX WIFI (msi.com)
best to do it with a bios update. most likely the root problem that caused the bugcheck.

here are other notes/issues from your memory dump
-------------
since you provided a memory.dmp file I can see more details and I am looking at why your video upscale tool did not complete. I am looking at a disk log and it has 2.8 million entries basically calling
PauseAdapter
then ResumeAdapter
all on port 2.
I guess I would ask if this is a raid disk, and if port 2 is a special port for your motherboard. It seems strange to making these calls over and over.
timer shows 11:09:28.338
11:09:28.346
.008 seconds for each cycle.
something is wrong with the drive/controller.
maybe update the BIOS, the CPU chipset, any raid driver and check for drive firmware updates using crystaldiskinfo.exe

it could explain why the upscale .exe is not getting done fast enough. debugger indicates it is a raid driver running.
there are repeated calls to storport!RaidAdapterPowerDownDevice
in the stream of log entries.
Device overheated? problems in sleep functions?
(something is wrong, maybe a raid adapter going to sleep and not waking correctly? might tell the system not to let your drive sleep. ie windows control panel, device manager, right mouse click on the adapter, see if there is a power management tab and uncheck the option to let the device sleep (just for testing))
------------------
I would looking at ffmpeg_waifu2xEX.exe and find out if I wanted it running (ie actually using it) and why it ended up using so much physical ram and pagefile.sys space. If you expect it to use a lot, I would make the pagefile bigger.
------------------
ffmpeg_waifu2xEX.exe looks like it was running and using a very large amount of memory. window memory compression tried to do some compressing on some of the 3 of the 80 threads and a fault was called.
debugger indicates that the .exe was using
10575600KB of memory space.
VirtualSize 26750 Mb
looks like the tool is some sort of video upscaler/editor
you could reduce the number of threads. I do not know why microsoft was trying to run memory compression on these threads. (other than the system was low on memory) could just be an app bug where one thread does not finish and eats up the ram and virtual memory.
later windows detects the limited free memory attaches to the process and tries run the memcompress on it. it still should not bugcheck, maybe the process did not complete due to some error and was hung?
process used a lot of kernel time and was switched in over 150k times. you have other suspect issues with the machine but this .exe was what lead to the bugcheck.


note 3: owning process was ffmpeg_waifu2xEX.exe
Image: MemCompression attached to it and it led to the bugcheck.

ffmpeg_waifu2xEX.exe has a very large number of threads. three of the threads were in the process of getting a memory compression error. cpu core 13,17 and 20
cpu core was trying to trim the working set for ffmpeg_waifu2xEX.exe looks like it also was in the process of a fault.


note 2: you could also update the BIOS to a more current version (beta or last stable version) then do a clean windows install and not install the ryzenmaster driver and dump any malware at the same time.

note: you have a old version of the bios running that does not have the various security fixes and you have a old version of the ryzenmaster driver running.
(you should update both the driver and bios together)
current beta bios released on 2023-03-03

windows was trying to remove some data from compressed memory, the data had been modified and it ended up using a bogus
memory address ( 000000000000aa00 )
A address like this could be an indication of malware trying to infect your system. I would guess it is coming from your internet download manager but it does look like you have several replaced drivers.

I would, use autoruns to remove the download manager.
turn off the system virtual memory to delete the pagefile.sys.
run cmd.exe as an admin then run
dism.exe /online /cleanup-image /restorehealth

then reboot and turn on virtual memory again to create a new pagefile.sys.
at this point I would run a malware scan to see if i could find any malware on your system.



download and run microsoft autoruns64 from here:
Autoruns for Windows - Sysinternals | Microsoft Learn
find the driver:
\SystemRoot\system32\DRIVERS\idmwfp.sys Wed Dec 19 06:52:11 2018
and delete it. (internet download manager by by Tonec )

other suspect drivers:
\SystemRoot\System32\Drivers\Fs_Rec.sys unavailable (FFFFFFFE)
\SystemRoot\System32\drivers\volmgrx.sys unavailable (FFFFFFFE)
Msfs.SYS unavailable (FFFFFFFE)

some windows files have been modified:
\SystemRoot\System32\win32kfull.sys

note:
seems weird to have Intel Telemetry Driver and Intel Power Engine Plugin installed on a amd based motherbooard

AMDRyzenMasterDriver.sys Thu Jun 24 22:21:58 2021 running.
--------------
machine info:
BIOS Version 1.40
BIOS Starting Address Segment f000
BIOS Release Date 08/11/2022
Manufacturer Micro-Star International Co., Ltd.
Product MPG X570S CARBON MAX WIFI (MS-7D52)
Processor Version AMD Ryzen 9 5900X 12-Core Processor
Processor Voltage 8bh - 1.1V
External Clock 100MHz
Max Speed 4950MHz
Current Speed 3700MHz

Hi! I'm sorry for late reply. While creating the account I probably turned on notification for everything and it got flooded so didn't notice your response.

As Colif mentioned, I got in touch with him since I couldn't post my reply here. I used the upscaler many times before but never for a 8K video. It ran fine in past within 4K limits.

My RAM: https://www.corsair.com/us/en/Categ...geance-RGB-PRO-SL-White/p/CMH16GX4M2D3600C18W

My SSD: https://www.samsung.com/us/computin.../980-pro-pcie-4-0-nvme-ssd-1tb-mz-v8p1t0b-am/

Now here is the issue I found with my RAM operating at 3200mHz without XMP:
Log Name: System
Source: Microsoft-Windows-MemoryDiagnostics-Results
Date: 08/03/2023 05:52:16 PM
Event ID: 1102
Task Category: None
Level: Error
Keywords:
User: SYSTEM
Computer: Ash
Description:
The Windows Memory Diagnostic tested the computer's memory and detected hardware errors. To identify and repair these problems, contact the computer manufacturer
Event Xml:
<Event xmlns="http://schemas.microsoft.com/win/2004/08/events/event">
<System>
<Provider Name="Microsoft-Windows-MemoryDiagnostics-Results" Guid="{5f92bc59-248f-4111-86a9-e393e12c6139}" />
<EventID>1102</EventID>
<Version>0</Version>
<Level>2</Level>
<Task>0</Task>
<Opcode>0</Opcode>
<Keywords>0x8000000000000000</Keywords>
<TimeCreated SystemTime="2023-03-08T11:52:16.6268364Z" />
<EventRecordID>25264</EventRecordID>
<Correlation />
<Execution ProcessID="8668" ThreadID="8664" />
<Channel>System</Channel>
<Computer>Ash</Computer>
<Security UserID="S-1-5-18" />
</System>
<UserData>
<Results xmlns="http://manifests.microsoft.com/win/2005/08/windows/Reliability/Postboot/Events">
<LaunchType>Manual</LaunchType>
<CompletionType>Fail</CompletionType>
<MemorySize>16308</MemorySize>
<TestType>10</TestType>
<TestDuration>1192</TestDuration>
<TestCount>12</TestCount>
<NumPagesTested>4152916</NumPagesTested>
<NumPagesUnTested>1749</NumPagesUnTested>
<NumBadPages>7</NumBadPages>
<T1NumBadPages>0</T1NumBadPages>
<T2NumBadPages>0</T2NumBadPages>
<T3NumBadPages>0</T3NumBadPages>
<T4NumBadPages>0</T4NumBadPages>
<T5NumBadPages>1</T5NumBadPages>
<T6NumBadPages>1</T6NumBadPages>
<T7NumBadPages>0</T7NumBadPages>
<T8NumBadPages>1</T8NumBadPages>
<T9NumBadPages>0</T9NumBadPages>
<T10NumBadPages>0</T10NumBadPages>
<T11NumBadPages>3</T11NumBadPages>
<T12NumBadPages>1</T12NumBadPages>
<T13NumBadPages>0</T13NumBadPages>
<T14NumBadPages>0</T14NumBadPages>
<T15NumBadPages>0</T15NumBadPages>
<T16NumBadPages>0</T16NumBadPages>
</Results>
</UserData>
</Event>

My PC failed to turn on with XMP on at max RAM advertised speed which is 3600mHz. So I turned XMP on but set frequency to 3200mHz. It turned on just fine, and I ran another Memory Diagnostics test.

Here is the result:

Log Name: System
Source: Microsoft-Windows-MemoryDiagnostics-Results
Date: 09/03/2023 11:21:29 PM
Event ID: 1102
Task Category: None
Level: Error
Keywords:
User: SYSTEM
Computer: Ash
Description:
The Windows Memory Diagnostic tested the computer's memory and detected hardware errors. To identify and repair these problems, contact the computer manufacturer
Event Xml:
<Event xmlns="http://schemas.microsoft.com/win/2004/08/events/event">
<System>
<Provider Name="Microsoft-Windows-MemoryDiagnostics-Results" Guid="{5f92bc59-248f-4111-86a9-e393e12c6139}" />
<EventID>1102</EventID>
<Version>0</Version>
<Level>2</Level>
<Task>0</Task>
<Opcode>0</Opcode>
<Keywords>0x8000000000000000</Keywords>
<TimeCreated SystemTime="2023-03-09T17:21:29.6954271Z" />
<EventRecordID>26095</EventRecordID>
<Correlation />
<Execution ProcessID="9040" ThreadID="9036" />
<Channel>System</Channel>
<Computer>Ash</Computer>
<Security UserID="S-1-5-18" />
</System>
<UserData>
<Results xmlns="http://manifests.microsoft.com/win/2005/08/windows/Reliability/Postboot/Events">
<LaunchType>Manual</LaunchType>
<CompletionType>Fail</CompletionType>
<MemorySize>16308</MemorySize>
<TestType>10</TestType>
<TestDuration>1192</TestDuration>
<TestCount>12</TestCount>
<NumPagesTested>4152916</NumPagesTested>
<NumPagesUnTested>1749</NumPagesUnTested>
<NumBadPages>3</NumBadPages>
<T1NumBadPages>0</T1NumBadPages>
<T2NumBadPages>0</T2NumBadPages>
<T3NumBadPages>0</T3NumBadPages>
<T4NumBadPages>0</T4NumBadPages>
<T5NumBadPages>1</T5NumBadPages>
<T6NumBadPages>0</T6NumBadPages>
<T7NumBadPages>0</T7NumBadPages>
<T8NumBadPages>0</T8NumBadPages>
<T9NumBadPages>0</T9NumBadPages>
<T10NumBadPages>0</T10NumBadPages>
<T11NumBadPages>2</T11NumBadPages>
<T12NumBadPages>0</T12NumBadPages>
<T13NumBadPages>0</T13NumBadPages>
<T14NumBadPages>0</T14NumBadPages>
<T15NumBadPages>0</T15NumBadPages>
<T16NumBadPages>0</T16NumBadPages>
</Results>
</UserData>
</Event>

Number of bad pages vary in both cases, not sure whether it means anything or not. Memtest also found error in both cases.

Set the RAM to auto, aka 2133mHz and both Diagnostics tools and memtest showed 0 error.

I'm utterly confused now. Did the BSOD happened because running Waifu2x at 8K made it utilize the ram to 100% which means it accessed the bad pages of the ram and crashed? Is my RAM faulty? Or is the issue something else entirely?

I don't think device overheated but can't be sure as I wasn't paying attention to the metrics. Waifu2x usually makes my GPU temperature less than 70°C but CPU temp 65-69° C. Sleep function...if you meant Sleep Mode, it's off. High Performance is selected with disk never going to sleep.

I don't use Ryzen Master itself. Some drivers must've came with Adrenalin but that's it, which itself is outdated. Considering how unpredictable AMD drivers are I'm a bit concerned about updating and farther breaking my system tbh. Virtualization is off iirc. As for other stuffs that came with it, the OS was setup by the company from where I purchase the PC components. When I claimed warranty on PSU back in November they set it up again from scratch since Adrenalin 22.10.2 almost broke it completely, making it unstable. They also partitioned my SSD and made it a clean installation from scratch. Afaik they checked all components then and only found a faulty PSU, not sure how true that is.

Since I didn't do the memtest when I created this thread, it lacked information. Which I hope I provided now. Let me know what you think.

Edit 1: Did a full scan of the system last month using Windows Defender, which is always up-to-date. It found nothing.

 

[johnbl]

memtest86 is the standard for testing your memory hardware. it runs on your machine but without having windows running.

while windows is running and you are running your upscaler I would be monitoring your system resources.

I might reduce the apps thread count, I am not sure if some app thread is not finishing, ties up a bunch of resources then microsoft memory compression kicks in and gets an access violation compressing the data for the thread.

still wonder why your drive controller is pausing and restarting 125 times a second. (I would focus on this)

if I were to guess, I think the drive controller is pausing and restarting all the time even when your app is not running. you would never know it until something failed.

tell me if you are running a raid setup.

if you want you can try this:
google how to force a memory dump using a keyboard. make the registry settings.
set your system to make a kernel dump.
boot your system but do not run your upscale conversion app.
let the system run for a hour or so, then force a kernel dump on the working system.

I will look at the internal storage controller log and see if it is turning off/on 125 times a second with out any load on it.

this way you will know if it is due to running the app or happening all the time without you knowing.

--------------
generally, amd will find and fix problems first
they would roll out fixes in the amdryzendriver.sys first. Then give motherboard vendors bios updates to incorporate into the vendors bios. (takes like 3 months after the handoff for the vendor to provide the bios update)
Microsoft also quickly gets a lot of error reports from windows error reporting. AMD will also give microsoft some fix and microsoft might make a file change and push it out via windows update.

the whole process is petty screwy since what gets loaded depends on the CPU microcode update level.
and these updates come from different sources.
bios, ryzendriver, and microsoft
microsoft microcode does not load if the bios or ryzen driver has already had the update level.

the windows debugger command to show the cpu microcode level do not work on AMD cpus, just on intel cpus

 

[ash_blake]

memtest86 is the standard for testing your memory hardware. it runs on your machine but without having windows running.

while windows is running and you are running your upscaler I would be monitoring your system resources.

I might reduce the apps thread count, I am not sure if some app thread is not finishing, ties up a bunch of resources then microsoft memory compression kicks in and gets an access violation compressing the data for the thread.

still wonder why your drive controller is pausing and restarting 125 times a second. (I would focus on this)

if I were to guess, I think the drive controller is pausing and restarting all the time even when your app is not running. you would never know it until something failed.

tell me if you are running a raid setup.

if you want you can try this:
google how to force a memory dump using a keyboard. make the registry settings.
set your system to make a kernel dump.
boot your system but do not run your upscale conversion app.
let the system run for a hour or so, then force a kernel dump on the working system.

I will look at the internal storage controller log and see if it is turning off/on 125 times a second with out any load on it.

this way you will know if it is due to running the app or happening all the time without you knowing.

--------------
generally, amd will find and fix problems first
they would roll out fixes in the amdryzendriver.sys first. Then give motherboard vendors bios updates to incorporate into the vendors bios. (takes like 3 months after the handoff for the vendor to provide the bios update)
Microsoft also quickly gets a lot of error reports from windows error reporting. AMD will also give microsoft some fix and microsoft might make a file change and push it out via windows update.

the whole process is petty screwy since what gets loaded depends on the CPU microcode update level.
and these updates come from different sources.
bios, ryzendriver, and microsoft
microsoft microcode does not load if the bios or ryzen driver has already had the update level.

the windows debugger command to show the cpu microcode level do not work on AMD cpus, just on intel cpus

I'm not entirely sure whether mine is a raid setup, as I'm not that familiar with the term. I doubt it is tho since I didn't do any of that, and from what I gathered from a quick Google search makes me believe mine isn't one.

As for Drive Controller, do you mean Storage Controllers on device manager? I have two there, one from Microsoft, another is NVM Express Controller.

 

[johnbl]

I'm not entirely sure whether mine is a raid setup, as I'm not that familiar with the term. I doubt it is tho since I didn't do any of that, and from what I gathered from a quick Google search makes me believe mine isn't one.

As for Drive Controller, do you mean Storage Controllers on device manager? I have two there, one from Microsoft, another is NVM Express Controller.

try to go into bios and look at the configuration for your drive and see what mode the drive controller is set to.
should be under pide/sata
some motherboard will have special ports with special meanings.
I tried to look at your motherboard manual but when I downloaded it the first pages were in english but changed to german and I could not read it.

edit:
I looked up the name of the amd am4 raid driver and it would be
rcraid.sys I do not think you had it loaded so your system should not have been in raid mode.

edit 2:
found the bios manual for your board>
AMDX570BIOS.pdf (msi.com)
controller should be set to AHCI

do you have the setting for
Storage Hierarchy enabled?