Question Frequent BSOD's. & C Drive disk check after every restart. Suspect bad RAM.

GNTSquid

Distinguished
Nov 22, 2014
38
1
18,535
I have 32GB of RAM on 2x 16GB sticks. About 2 weeks ago I started getting frequent BSOD's but almost a different error code every time. I also get these disk error checks for the C drive after every restart whether its from a BSOD or not. It never says it finds errors. The most frequent error code has been about memory loss. I've got others about memory management and some others I cant remember the name of. For the most part I can use the computer normally, so its hard to say any one thing is causing the crashes.

Today I ran some tests outlined in this thread: How to test your RAM using memtest86 booted from a USB. I tested both sticks separately and got errors immediately with both. Earlier today I did Memtest64 within Windows and the built in Windows Memory Diagnostic test and got immediate errors on those as well.

I havent changed any of my BIOS settings in probably over a year. I think i've updated BIOS once back in late 2020 if I remember right. Are there more tests I should do, is it safe to say my RAM has gone bad? What are the odds two sticks would go bad, i've never had this happen before? Is the only fix to buy new RAM?

Link to a couple minidump files for the BSOD.

Specs are:

Windows 10 Pro 22H2
Asus ROG Strix X570 -E Gaming BIOS ver 2606
AMD Ryzen 9 3900X 12-Core Processor 3.79 GHz
32GB G.Skill Trident RAM
Asus Tuf Gaming RTX 3080 video card
 
Last edited:
I have 32GB of RAM on 2x 16GB sticks. About 2 weeks ago I started getting frequent BSOD's but almost a different error code every time. I also get these disk error checks for the C drive after every restart whether its from a BSOD or not. It never says it finds errors. The most frequent error code has been about memory loss. I've got others about memory management and some others I cant remember the name of. For the most part I can use the computer normally, so its hard to say any one thing is causing the crashes.

Today I ran some tests outlined in this thread: How to test your RAM using memtest86 booted from a USB. I tested both sticks separately and got errors immediately with both. Earlier today I did Memtest64 within Windows and the built in Windows Memory Diagnostic test and got immediate errors on those as well.

I havent changed any of my BIOS settings in probably over a year. I think i've updated BIOS once back in late 2020 if I remember right. Are there more tests I should do, is it safe to say my RAM has gone bad? What are the odds two sticks would go bad, i've never had this happen before? Is the only fix to buy new RAM?

Link to a couple minidump files for the BSOD.

Specs are:

Windows 10 Pro 22H2
Asus ROG Strix X570 -E Gaming BIOS ver 2606
AMD Ryzen 9 3900X 12-Core Processor 3.79 GHz
32GB G.Skill Trident RAM
Asus Tuf Gaming RTX 3080 video card
Do a bios clear and then boot memtest86 and test.
 
failed memory test does not necessarily mean that there is any problem with your RAM sticks. Your first assumption should be that your BIOS RAM timings settings are incorrect.
Generally, you would go into bios and reset the bios to defaults, update the BIOS to get the most current RAM timings for your motherboard then reboot and test the RAM on the default timings.
If the test passes with default timings then you can increase the RAM timing speed in the BIOS reboot and retest.
If the test fails with default timings then you decrease the RAM timings. Note: it is pretty common for BIOS to use a incorrect command delay timing (generally, the 5th RAM timing, often defaults to 1 T while it might need to be 2 T)

note: the more RAM you have then the slower the timings need to be. Also, the further the RAM sticks are to the CPU the slower the circuit has to run. These settings get tunned by the motherboard vendor after the motherboard is released to the public. fixes are in the bios update.

some vendors like asus attempt to automatically tune your RAM timings and will increase them out of range. You can turn this tuning off in BIOS

timings can also change over time as capacitors age and change the circuit timings. Old motherboards/power supplies can change their output voltages over time. voltages to cpu pins are converted to clock frequencies used to shuffle data between different levels of cache inside of the cpu. These also are RAM errors that result in bugchecks called by the CPU but they do not refer to your RAM sticks. (refers to the cpu internal cache RAM)
windows memory dumps can be useful to detect the source of the error.
 
Last edited:

GNTSquid

Distinguished
Nov 22, 2014
38
1
18,535
Do a bios clear and then boot memtest86 and test.
failed memory test does not necessarily mean that there is any problem with your RAM sticks. Your first assumption should be that your BIOS RAM timings settings are incorrect.
Generally, you would go into bios and reset the bios to defaults, update the BIOS to get the most current RAM timings for your motherboard then reboot and test the RAM on the default timings.
If the test passes with default timings then you can increase the RAM timing speed in the BIOS reboot and retest.
I have done both of these things now.
  • Cleared CMOS with a screw driver and shorting the pins. Reset BIOS to default. Ran tests with memtest86 one RAM stick at a time. Immediate fails for both
  • Cleared CMOS by removing the battery for 30 seconds. Reset BIOS to default. Ran test with memtest86 one stick at a time. Immediate fails for both.
  • Next I updated BIOS to ver 4408 from Asus website. Ran tests with memtest86. Immediate fails again. Set to default BIOS settings again but I also had to change some fan settings so I would stop getting warnings about no CPU fan and setting AIO fan speed to auto based off CPU temp. Dont know if thats relevant.
After my 2nd test after removing the CMOS and browsing the internet looking for the BIOS update file I got another BSOD. I'm also still getting Disk scan and repair for the C drive. No errors with that ever seem to be found but its always doing it now.
Did the minidump files help? I've been saving these files from the memtests would those be helpful to look at?

If the test fails with default timings then you decrease the RAM timings. Note: it is pretty common for BIOS to use a incorrect command delay timing (generally, the 5th RAM timing, often defaults to 1 T while it might need to be 2 T)

note: the more RAM you have then the slower the timings need to be. Also, the further the RAM sticks are to the CPU the slower the circuit has to run. These settings get tunned by the motherboard vendor after the motherboard is released to the public. fixes are in the bios update.

some vendors like asus attempt to automatically tune your RAM timings and will increase them out of range. You can turn this tuning off in BIOS

timings can also change over time as capacitors age and change the circuit timings. Old motherboards/power supplies can change their output voltages over time. voltages to cpu pins are converted to clock frequencies used to shuffle data between different levels of cache inside of the cpu. These also are RAM errors that result in bugchecks called by the CPU but they do not refer to your RAM sticks. (refers to the cpu internal cache RAM)
windows memory dumps can be useful to detect the source of the error.
I dont understand a lot of what youre saying here. I dont really understand how to change RAM timings or what numbers or info i'd punch in. All I know is you do that in BIOS. How can I use this?
 
big problem : (you should asume this is the cause of your problems)
two copies of amdryzenmasterdriver.sys running:

you should use the asus utility uninstaller to uninstall one version. I would download microsoft autoruns64.exe find the drivers and delete the entries so they do not load. (bios armor crate might reinstall them until it is turned off)
Autoruns for Windows - Sysinternals | Microsoft Learn

most likely the first copy overclocks your system by some percent, then the second copy overclocks the system again by some percent then you get errors that windows memory manager is saving to your pagefile.sys. when you reboot the overclock is restarted but the errors that windows saved to the pagefile.sys are reloaded. You have to delete the pagefile.sys by turning off virtual memory and turning it back on since you will not know what errors were saved.



C:\Program Files\AMD\RyzenMaster\bin\AMDRyzenMasterDriver.sys Tue Mar 31 21:07:41 2020 (5E84138D)
fffff8058b5f0000 fffff8058b622000

C:\Program Files (x86)\ASUS\AI Suite III\DIP4\AMDRyzenMasterDriver\bin\AMDRyzenMasterDriver.sys Fri Sep 18 01:22:51 2020 (5F646E5B)
fffff805`7745

you have two copies of amdryzenmasterdriver running from different directories.




second bugcheck was due to memory corruption saved to the pagefile.sys. most likely the memory timings were wrong, causing data to be stored incorrectly in RAM then the windows memory manager copied the bad data in RAM and stored it in the pagefile.sys. Later the system tried to use/clean up the data and found it had been modified and called this bughceck.
The data gets written to disk after a long sleep and your machine was running for 1 day 18.5 hours.




---------------
ROG STRIX X570-E GAMING
ROG Strix X570-E Gaming | ROG Strix X570-E Gaming | Gaming Motherboards|ROG - Republic of Gamers|ROG Global (asus.com)

there is a beta bios just released on 2023/02/09
the previous bios version release was
ROG STRIX X570-E GAMING BIOS 4408 dated 2022/05/31

your version is too old. you need to get at least the 4408 version and update the drivers. I expect that if you turn off armory crate then this bugcheck will go away. (still need to look at the older bugcheck)

Note: i would install the beta version to get the CPU security fixes. But note that you will also need to install the USB firmware fixes and update the various drivers. I run the AI Suite 3 Cleaner to uninstall the armory crate from windows. then I would not install the Utilities: Armoury Crate.

note: the updated bios will update the default memory timings for your actual motherboard. This can mean you will not have to manually adjust the timings yourself.

--------------
looked at the first memory dump.
single bit memory corruption

first you need to go into bios and disable armory crate. save the bios and
then boot windows and uninstall the asus utility.
reboot into windows then turn off windows virtual memory and turn it back on. This will delete your pagefile.sys.

try this and see if your system runs better, I will try to see if I can find the timing specs for
G-Skill F4-3600C16-16GTZRC

edit:
default memory timing for chip is:
DDR4-3600 CL16-19-19-39 1.35V
still need to check the command rate, often motherboard default is incorrect. (generally the 5th memory setting is not documented correctly)






BIOS Version 2606
BIOS Release Date 08/13/2020
Manufacturer ASUSTeK COMPUTER INC.
Product ROG STRIX X570-E GAMING
Processor Family 6bh - AMD Zen Processor Family
Processor Manufacturer Advanced Micro Devices, Inc.
Processor ID 100f8700fffb8b17
Processor Version AMD Ryzen 9 3900X 12-Core Processor
Processor Voltage 8bh - 1.1V
External Clock 100MHz
Max Speed 4650MHz
Current Speed 3800MHz
memory:
Part Number F4-3600C16-16GTZRC
Speed 3600MHz
Manufacturer G-Skill
Size 16384MB
 
Last edited:
  • Like
Reactions: GNTSquid

GNTSquid

Distinguished
Nov 22, 2014
38
1
18,535
big problem : (you should asume this is the cause of your problems)
two copies of amdryzenmasterdriver.sys running:

you should use the asus utility uninstaller to uninstall one version. I would download microsoft autoruns64.exe find the drivers and delete the entries so they do not load. (bios armor crate might reinstall them until it is turned off)
Autoruns for Windows - Sysinternals | Microsoft Learn

most likely the first copy overclocks your system by some percent, then the second copy overclocks the system again by some percent then you get errors that windows memory manager is saving to your pagefile.sys. when you reboot the overclock is restarted but the errors that windows saved to the pagefile.sys are reloaded. You have to delete the pagefile.sys by turning off virtual memory and turning it back on since you will not know what errors were saved.



C:\Program Files\AMD\RyzenMaster\bin\AMDRyzenMasterDriver.sys Tue Mar 31 21:07:41 2020 (5E84138D)
fffff8058b5f0000 fffff8058b622000

C:\Program Files (x86)\ASUS\AI Suite III\DIP4\AMDRyzenMasterDriver\bin\AMDRyzenMasterDriver.sys Fri Sep 18 01:22:51 2020 (5F646E5B)
fffff805`7745

you have two copies of amdryzenmasterdriver running from different directories.
Thats odd. It looks like overclocking was happening from the Ryzen Master program and the AI Suit 3. I never even used "Ryzen Master". I downloaded it a couple years ago just to look at it but never did anything with it. I've now uninstalled Ryzen Master, Armoury Crate, and AI Suite 3. Used the AI cleaner to get rid of AI Suite 3 but there was no option to remove Armoury Crate so I got rid of that with Add/Remove.

Running AutoRun shows no more AMDRyzenMasterDriver.sys file.

I've disabled Armoury Crate install in BIOS as well.

ROG STRIX X570-E GAMING
ROG Strix X570-E Gaming | ROG Strix X570-E Gaming | Gaming Motherboards|ROG - Republic of Gamers|ROG Global (asus.com)

there is a beta bios just released on 2023/02/09
the previous bios version release was
ROG STRIX X570-E GAMING BIOS 4408 dated 2022/05/31

your version is too old. you need to get at least the 4408 version and update the drivers. I expect that if you turn off armory crate then this bugcheck will go away. (still need to look at the older bugcheck)

Note: i would install the beta version to get the CPU security fixes. But note that you will also need to install the USB firmware fixes and update the various drivers. I run the AI Suite 3 Cleaner to uninstall the armory crate from windows. then I would not install the Utilities: Armoury Crate.

note: the updated bios will update the default memory timings for your actual motherboard. This can mean you will not have to manually adjust the timings yourself.

--------------
looked at the first memory dump.
single bit memory corruption

first you need to go into bios and disable armory crate. save the bios and
then boot windows and uninstall the asus utility.
reboot into windows then turn off windows virtual memory and turn it back on. This will delete your pagefile.sys.

try this and see if your system runs better, I will try to see if I can find the timing specs for
G-Skill F4-3600C16-16GTZRC

edit:
default memory timing for chip is:
DDR4-3600 CL16-19-19-39 1.35V
still need to check the command rate, often motherboard default is incorrect. (generally the 5th memory setting is not documented correctly)
I am running BIOS ver 4408. I installed that this morning and decided to skip the newest version for now because its still in beta.

In BIOS I could not find where to check the RAM timings. I found something related to overclocking and RAM and I saw a bunch of numbers but there was far more numbers than just 16-19-19-39. There were dozens of numbers I could have changed and had letters CHA or CHM or something like that above each box. It didnt make a lot of sense to me so I left it alone.

Currently everything with the RAM is set to Auto and my RAM is running at a lower frequency (around 2800mhz) than its made for. It should be running at 3600mhz but to get that i have to activate DOCP which seems to do some overclock stuff.

I turned off and back on Windows Virtual Memory

After all this no BSOD's so far, but its never never at regular intervals to begin with so I think only time will tell if thats fixed. I am however still getting a disk check on the C drive after every restart or boot.
 
Last edited:
try to find the setting in bios to turn off armory crate.
it puts files on your drive directly from the bios (outside of the operating system) the files keep coming back until you turn it off in bios.
run your system with the lower RAM clock speed until you think it is stable. Then go into bios and set the memory profile to the proper speed and retest to see if it is stable.

the chkdsk of the c: will happen if the file system dirty bit is set
this bit get set when the shutdown does not complete.

you can use this tool to see if the file system dirty bit is set
fsutil dirty | Microsoft Learn
ie start cmd.exe as an admin then run
fsutil.exe dirty query

so after the system boots and check disk has run then run this command and see if check disk cleared the bit.

(ie chkdsk is not clearing the bit or the bit is getting set again at each shutdown)

both problems have different causes/fixes.
(but I would turn off armory crate first before I spent too much time figuring it out)
 

GNTSquid

Distinguished
Nov 22, 2014
38
1
18,535
try to find the setting in bios to turn off armory crate.
it puts files on your drive directly from the bios (outside of the operating system) the files keep coming back until you turn it off in bios.
run your system with the lower RAM clock speed until you think it is stable. Then go into bios and set the memory profile to the proper speed and retest to see if it is stable.

I did get rid of Armoury Crate, it is gone. Getting rid of it wasnt a problem. I'm still unsure what you want me to do with the RAM though.
Below are screens of the BIOS and the only place I saw that had anything to do with timing. But like I said there's way more numbers than just the 4 you mentioned for the timing. I'm not sure what any of this means or what the numbers should be. Note that this is what it is when set to default, I haven't touched any of this. Should I change any of these numbers?

1gdai5N.png


VWCpMWC.png


RVPcgfi.png



the chkdsk of the c: will happen if the file system dirty bit is set
this bit get set when the shutdown does not complete.

you can use this tool to see if the file system dirty bit is set
fsutil dirty | Microsoft Learn
ie start cmd.exe as an admin then run
fsutil.exe dirty query

so after the system boots and check disk has run then run this command and see if check disk cleared the bit.

(ie chkdsk is not clearing the bit or the bit is getting set again at each shutdown)
So I did the fsutil.exe dirty query as well as "fsutil dirty query c:" from the link there and got this in return. Still says the same after a shutdown and a restart.

9vV6Qok.png


Is it supposed to be dirty?
 
ok, the file system is marked dirty even after you have booted up. This means the chkdsk command did not fix the error in the file system on bootup. so try this:
start cmd.exe as an administrator and run
chkdsk c: /v /scan /forceofflinefix
this will run the check disk scan in verbose mode and scan your drive to find the problem then queue the fix for the reboot.
it should give you about 1 page of info and tell you what phase of the chkdsk that found the error.
give it a try, capture the screen, reboot and check to see if the drive is dirty after the reboot.

if I remember correctly the last time I saw this error it turned out to show a error in phase3 of chkdsk and I had to add the /sdcleanup switch to make it clean up bogus security descriptors.
chkdsk | Microsoft Learn


Ram timings: I am not the best person to help you with the RAM timings. I generally just have people turn off the bios aitweaker, and pick a profile then run memtest86 and prime95 see if the hardware is stable. biggest mistake I see in old versions of a BIOS is that it sets the command rate to 1T by default. It looks like cpu-z shows a command rate of 2T.
This means that the electronics for the address controller takes 2 clock cycles to set up the electronics and stabilize the address lines before they are read.
The more RAM you have the slower the circuit works and it might require 2T
before the address lines are stable. If the timing is wrong the address lines are read before they finish going from a 1 to 0 or 0 to 1 and the wrong address is read.
Just a quirk of the electronics is that it takes longer time to change from a 1 to a 0 than it takes to change from a 0 to a 1. It also takes longer to change these the further the RAM chip is from the CPU. Just FYI

also: RAM timings: yes there are a lot more than 4 ram timings.
basically, most vendors use the same names for the first 4 RAM timings so memory vendors tend to tell you what they are. Problem is each vendor uses different names for all of the rest of the timings. Years ago there were 12 timings for the chips. I don't know how many are specified today. The timings depend on the vendor of the chip, the quality of the chip, characteristics of the motherboard and other factors. It gets pretty hard to guess what settings to use.

I think this is why asus has the ai tweaker in bios it attempts to tweak the timings lets your system run, then on reboot it tweaks the timings again to be faster. lets the system run, if the system crashes it goes back to a previous setting that did not crash. This method has problems with the new features of windows. IE windows memory compression and sleep states causing data corruption due to memory timing error to be saved to the pagefile.sys and reloaded. Since windows looks for malware if any of the data corruption is in a checked part of the windows kernel, windows will assume it is malware and shut the system down by calling a bugcheck.

this is another reason why you tweak your memory setting, boot memtest86 and test for correct settings. (no windows running means no pagefile.sys being saved with bad/corruupted data due to bad timing settings)
Most people do not know to turn off virtual memory to delete the pagefile.sys
 
Last edited:

GNTSquid

Distinguished
Nov 22, 2014
38
1
18,535
ok, the file system is marked dirty even after you have booted up. This means the chkdsk command did not fix the error in the file system on bootup. so try this:
start cmd.exe as an administrator and run
chkdsk c: /v /scan /forceofflinefix
this will run the check disk scan in verbose mode and scan your drive to find the problem then queue the fix for the reboot.
it should give you about 1 page of info and tell you what phase of the chkdsk that found the error.
give it a try, capture the screen, reboot and check to see if the drive is dirty after the reboot.

if I remember correctly the last time I saw this error it turned out to show a error in phase3 of chkdsk and I had to add the /sdcleanup switch to make it clean up bogus security descriptors.
chkdsk | Microsoft Learn
So I did this and rebooted the computer and had a bit of a scary moment. The computer was stuck in a cycle of checking and repairing the C drive but never booting to windows. After running "chkdsk c: /v /scan /forceofflinefix" I restarted the computer and it again did the disk check. After it got to 100% though it hung there for a few minutes. I did a hard reset using the reset button on the case. It started again and did the disk check all over again. Got to 100% process complete and just sat there for a few minutes again with no changes. This time I did a hard shut down with the power button. Computer powered back up and after the Asus splash screen then asked if I wanted to restart to windows or trouble shoot. I did trouble shoot to start in Safe Mode. Computer crashed I got a BSOD that said "CRITICAL_PROCESS_DIED". Restarted computer again. Disk check again. It went through this time but again asked if I wanted to trouble shoot. This time I said no just restart and I FINALLY got in to Windows. I'm afraid to restart the computer again.

At some point it also gave me an "Automatic Repair" window. Said Press Advanced options to try other options to repair your PC or Shut Down.
Log file: E\Windows\System32\Logfiles\Srt\SrtTrail.txt .
I did find a SrtTrail.txt file on the C drive in the same file path as above. Here's a link to it if it helps.
Now I do have an E drive but it only contains work stuff. Mostly its storage for files but there's some custom software on it, never used though. And there's no Windows\System32 folder on it. It does have a WindowsApps folder but it wont let me open it says I dont have permission and I cant seem to give myself permission.

Checked the disk and its still dirty.

This is the result of the forceofflinefix command.
ltAfATT.png



Why is this happening? I didnt download anything weird or change any hardware or settings. Aside from the normal software updates or new games the computer is the same as its been for a year now. These problems seemed to start on their own one day.

EDIT UPDATE:

Ran the forceofflinefix again and got a different result...or more results. Also running the "chkdsk /spotfix" as suggested in the command window just tells me it cant because the volume is in use by another process and asks if I want to check the next time the system restarts.

wPi6klT.png
 
Last edited:
This is progress, some of your files have clusters that are marked as free.
windows will not be able to repair this and it is why it fails the chkdsk.

I would expect this to be a bug in the firmware of the drive. But it could also be a bug in the storage controller or the storage controller driver, or the cache system or drive firmware has not finished processing its background queue.
(but the repairs might be in queue to the drive firmware)

try this first:
boot into the bios and leave the computer on for a hour or two. The drive firmware will start running 5 minutes after the system is idle. You may find the drive firmware was starved from running due to sleep settings but in bios it will finish all of the windows repair operations that are pending.
(this is about write endurance of solid state drives)

then boot into windows and run your check disk scan to see the list of bad files and see if they are fixed.

if they are not fixed after the bios wait, then you will have to delete each bad file one by one. Then figure out why they were corrupted in the first case.


I would try the following:
google how to turn off lazy writes for your drive and turn it off for now.
download and run crystaldiskinfo.exe and check the firmware version of the drive and google to see if it is the current. Then install the chipset support drivers directly from the CPU vendor.

even if there was no bug in the firmware, bios or chipset you can still get this type of error if you installed on a quick formated drive. most drives are quick formated now since no one want to wait hours for a full format.

current versions of windows now assumes a quick format and then runs a background system process that checks each cluster for proper function. if it finds a bad cluster it tries to copy the data over and over until it gets a good copy. When it gets a good copy it moves it to a new cluster and marks the old cluster as bad. (and your drive size shrinks by the size of the cluster that is marked as bad) This process can take days after a windows install and is hindered by the sleep setting for a drive. The process starts 5 minutes after idle and goes until the drive is told to sleep (at maybe 10 minutes)
you can speed the process up by telling window not to sleep your drive for a few days. if you look at resource monitor it will look like the system process is running 100% at idle and is using a lot of resources until this processing is done.
 
  • Like
Reactions: GNTSquid
you can use file explorer and find each bad file from the check disk report and delete the file
this will remove the problem that chkdsk could not fix. this will be find for files that are logs and you do not care about. But the files that are DLL have code and the dll will be missing after you delete them. you may have to reinstall the nvidia physic dll after deleting the file. same with the microsoft ai code.

you can also delete the files from the command line like this
del "c:\users\kyle\appdata\roaming\discord\code cache\js\55aboo408de0b172_0"

edit: I guess i would be in favor of just deleting all of the cross linked files and files with free clusters. I think it is better to have a missing .dll file or java script than to have a .dll that has part of a txt file log in it by mistake. just easier to fix later when the file is reported as missing rather than getting a access violation because some log got cross linked into a .dll file and run as code.

note; appdata directory is a hidden directory, if you use file explorer to delete the file you have to make sure hidden files/directories are visible.

this was one of the files. the quotes are required since one of the directories has an embedded space in it.

some of your other files are cross linked, IE 2 files linked to the same cluster. Windows does not know which file truly owns the cluster. You can guess and delete one of the files in the pair but you might guess incorrectly. hopefully you can find a firmware update that fixes the problem then just delete the software and reinstall and be good. I would keep checking my system with the chkdsk verbose switch to see if you still get more corrupted files, I guess it would also be good to know if the files are only being corrupted during a install or if they get corrupted over time.

edit: I guess I should also point out that the file corruptions may have happened while running old bios versions/ chipset versions and you have already installed the fix but are looking at the old corruptions. delete the files and check for more corruptions to see if the problem is fixed.
 
Last edited:
  • Like
Reactions: GNTSquid

GNTSquid

Distinguished
Nov 22, 2014
38
1
18,535
try this first:
boot into the bios and leave the computer on for a hour or two. The drive firmware will start running 5 minutes after the system is idle. You may find the drive firmware was starved from running due to sleep settings but in bios it will finish all of the windows repair operations that are pending.
(this is about write endurance of solid state drives)
I did this. Let it sit in BIOS for about 1h 45min but it never went in to sleep mode. It just sat there seemingly not doing anything. Eventually I exited into Windows where it did another scan and repair.

then boot into windows and run your check disk scan to see the list of bad files and see if they are fixed.

if they are not fixed after the bios wait, then you will have to delete each bad file one by one. Then figure out why they were corrupted in the first case.
check disk came back with 3 queued .log files for offline repair. Two were identical and related to ASUS Armoury Crate HAL something or other. The other was related to Adobe Creative Cloud that I need for work. None of the previous flagged files from the earlier check disks were listed. Deleted them and the check disk came back empty with no queued files for offline repair!

Did a reboot and no scan and repair! Started up without issue, finally!



I would try the following:
google how to turn off lazy writes for your drive and turn it off for now.
download and run crystaldiskinfo.exe and check the firmware version of the drive and google to see if it is the current. Then install the chipset support drivers directly from the CPU vendor.
I couldnt find any info on lazy writes except in this Microsoft help doc that is way over my head. I dont understand anything being talked about or how to put it to use. So lazy write could still be on. I dont know how to turn it on or off.

Crystal Disk says my C drive firmware is: 111110WD
My C drive is a Western Digital_Black SN750 500GB M.2 SSD - To update its firmware WD required me to download their software called "Dashboard". Doing so shows the firmware being the same as whats in Crystal Disk and that the firmware is already the most up to date available.

you can speed the process up by telling window not to sleep your drive for a few days. if you look at resource monitor it will look like the system process is running 100% at idle and is using a lot of resources until this processing is done.
Should I always have no sleep for my C drive? What power/sleep settings should I have to avoid this happening in the future? Since I built the computer I usually let it sit idol. Only really turning it off or rebooting it to clean it of dust or because an install requires it.

As for drivers and keeping those up to date. Should I avoid using Armoury Crate, it seems like that was the source of these issues as a lot of the files being mentioned in the check disk were from ASUS? RGB was managed through Armoury Crate and its "Aura Sync" software and since I dont have Armoury Crate right now my computer is lit up like a bag of Skittles.

I still need to do another RAM test and mess with those timings. You said the most current are DDR4-3600 CL16-19-19-39 1.35V should I look at the timings listed in CPU-Z and punch those in the BIOS settings? Should I change the clock speed first?
 
google for and download microsoft autorun64.exe
take a look at what is in the windows task scheduler. Asus might put in a task to reinstall its utility even after you remove it.
You should delete the asus armory crate files, reboot and make sure they do not come back.

if asus is putting files on the disk from the bios it could corrupt the disk file system.
(and the files come back after rebooting, assuming that it does not reinstall the files via a the task scheduler)

after you get it so there are no more armory create, you can try and change memory timings and running memtest86 to confirm them.

lazy writes can be access from windows control panel device manager by right clicking on the storage drive and looking at the policy tab. I think there is a checkbox for lazy writes.
 
Last edited:

GNTSquid

Distinguished
Nov 22, 2014
38
1
18,535
google for and download microsoft autorun64.exe
take a look at what is in the windows task scheduler. Asus might put in a task to reinstall its utility even after you remove it.
You should delete the asus armory crate files, reboot and make sure they do not come back.

if asus is putting files on the disk from the bios it could corrupt the disk file system.
(and the files come back after rebooting, assuming that it does not reinstall the files via a the task scheduler)
There's no trace of Armoury Crate that I can find, its disabled in BIOS, and the task scheduler doesnt have anything related to Asus or Armoury Crate in it.

after you get it so there are no more armory create, you can try and change memory timings and running memtest86 to confirm them.

lazy writes can be access from windows control panel device manager by right clicking on the storage drive and looking at the policy tab. I think there is a checkbox for lazy writes.
Looking at the policy tab I see these two options, but nothing about lazy write.

uCCwEv7.png
 
There's no trace of Armoury Crate that I can find, its disabled in BIOS, and the task scheduler doesnt have anything related to Asus or Armoury Crate in it.


Looking at the policy tab I see these two options, but nothing about lazy write.

uCCwEv7.png
sorry, I think the enable write cache is the setting is the correct term
(rather than lazy writes)
What is Lazy Write? (computerhope.com)

when lazy writes are enabled, the file system is lied to and is told the data is flushed to disk when it hits the cache RAM in the disk.
works most of the time unless the power goes out or you get a bugcheck with data in the cache and not on the actual disk yet.

turning off the cache write the file system will not get this signal until the data is really on the disk or ssd. takes a bit longer but prevents file corruption if the power goes out or any non complete shutdown. most likely you can turn on the write cache after the system is stable. IE you stop getting corruptions.
you also turn it off if you suspect bugs in the ssd. (if there are bugs and no firmware updates)

I guess I should point out that if their is a bug in the ssd and write cache is turned off then it might happen that windows shutdown will not complete. Windows will not shutdown if it thinks it will loose data. it will just hang there waiting for the signal from the ssd. often the screen will already be blanked out.
but the cpu fans don't turn off and people force the power off via the power switch. then the drive is dirty and chkdsk runs on the next boot again.
 
Last edited:

GNTSquid

Distinguished
Nov 22, 2014
38
1
18,535
Set latencies to 16-19-19-39
DDR voltage to 1.35V
Command Rate 1T.

Set DDR frequency to 2133mhz, test.
Increase frequency gradually and test after each step.
2400,2600,2800,2933,3000,3200,3400,3600.

When test fails, then return to previous settings that worked.
This will take a while, but if it fails at say 2800 does that mean I'll have to keep my RAM at 2600 from here on out and not get the full speed out of it?
 

GNTSquid

Distinguished
Nov 22, 2014
38
1
18,535
Set latencies to 16-19-19-39
DDR voltage to 1.35V
Command Rate 1T.

Set DDR frequency to 2133mhz, test.
Increase frequency gradually and test after each step.
2400,2600,2800,2933,3000,3200,3400,3600.

When test fails, then return to previous settings that worked.
So I ended up doing a test and it failed right away. I was also starting to get a lot of other issues. The C drive scan and repair was back and Windows Defender completely dissapared one time until I did a restart.

Tried disk cleaning and all that again but had no changes. Eventually I completely reinstalled Windows 10 from scratch and got 3 BSOD's during the process all of which talked about memory management. On the 3rd attempt I finally got Windows installed but got one more BSOD after that. Did a Windows Memory Diagnostic and it said it detected memory issues after the full scan.

I even tried upgrading to Windows 11 to see if that would fix something and I can't even do that, I get an error every time I try to install. I'm starting to suspect its a Motherboard issue but I dont know how I can prove that since I dont have a 2nd machine I can test hardware on? Seems kind of unlikely I would have a failed hard drive AND two sticks of bad RAM all at the same time though.