[SOLVED] Windows 10: Need help reading full memory dump files (2400g machine with issues)

[Vule1901]

Hi everyone, since changing my ram from Geil 8gb kit 3000cl16 to Patriot (hynix cjr) 16gb kit 3733cl17 (running stock 2133 and 3200cl14 currently which passed 32 memtest86 and multiple other ram tests) I started having a lot of different issues (mostly video tdr errors with amd notification). To cut it short, no more driver issues (bcs i downloaded drivers for apu and chipset from motherboard site) but now, my system crashes in a way: Screen goes black and monitor cycles from pitch black to "backlight is on" a few times while system still runs (audio and calls go through), after which screen light up to a "white static screen" after which system buzzes with a bsod sound but with no blue screen, it either hangs (almost 5 mins when full dump is turned on) before restarting or restarts immediately (sometimes tehre is no memory dump /minidump).

I have gone through 3 psu changes, dozens of driver/chipset driver combinations, 7 bios revisions (from f41 to f60 currently) and multiple clean windows installs.
Ran the system at default bios settings and ran it overclocked too, it passes any test i put on it but crashes in random moments
Crashes do happen mostly randomly, but most frequent moments of crashes are when opening stuff.
The system in question is a amd ryzen 5 2400g on a Gigabyte B450m s2h motherboard with before mentioned 16gb kit, a 256gb samsung m.2 nvme ssd (970 evo oem) with an old wd green 240gb m.2 ssd
Windows event viewer always gives a critical error 41 with stopcode 278 and errors 6008 and 7000

Another thing to mention is that since I was having issues with gpu (apu) drivers I updated my windows and blocked updates using third party tool that I have been using for years before.
One of the reasons for doing so is bcs windows kept installing drivers over already installed gpu drivers from amd or gigabyte site.

In the link given at the bottom are multiple minidump (only one had nvme driver/controler report, which is the first time its been reported) files and a 16gb full memory dump file.
Since we have tried a lot of things (reddit kind of ghosted my posts ) I was recommended to reach out on this forum for someone with expertise to read the full memory dump file.

Thanks to everyone who tries!



The link: https://mega.nz/folder/awIQwAzT#DjuPTU4QlxkObipC0O_9gQ (I can keep uploading newer files if needed)

 

[Koekieezz]

Hi everyone, since changing my ram from Geil 8gb kit 3000cl16 to Patriot (hynix cjr) 16gb kit 3733cl17 (running stock 2133 and 3200cl14 currently which passed 32 memtest86 and multiple other ram tests) I started having a lot of different issues (mostly video tdr errors with amd notification). To cut it short, no more driver issues (bcs i downloaded drivers for apu and chipset from motherboard site) but now, my system crashes in a way: Screen goes black and monitor cycles from pitch black to "backlight is on" a few times while system still runs (audio and calls go through), after which screen light up to a "white static screen" after which system buzzes with a bsod sound but with no blue screen, it either hangs (almost 5 mins when full dump is turned on) before restarting or restarts immediately (sometimes tehre is no memory dump /minidump).

I have gone through 3 psu changes, dozens of driver/chipset driver combinations, 7 bios revisions (from f41 to f60 currently) and multiple clean windows installs.
Ran the system at default bios settings and ran it overclocked too, it passes any test i put on it but crashes in random moments
Crashes do happen mostly randomly, but most frequent moments of crashes are when opening stuff.
The system in question is a amd ryzen 5 2400g on a Gigabyte B450m s2h motherboard with before mentioned 16gb kit, a 256gb samsung m.2 nvme ssd (970 evo oem) with an old wd green 240gb m.2 ssd
Windows event viewer always gives a critical error 41 with stopcode 278 and errors 6008 and 7000

Another thing to mention is that since I was having issues with gpu (apu) drivers I updated my windows and blocked updates using third party tool that I have been using for years before.
One of the reasons for doing so is bcs windows kept installing drivers over already installed gpu drivers from amd or gigabyte site.

In the link given at the bottom are multiple minidump (only one had nvme driver/controler report, which is the first time its been reported) files and a 16gb full memory dump file.
Since we have tried a lot of things (reddit kind of ghosted my posts ) I was recommended to reach out on this forum for someone with expertise to read the full memory dump file.

Thanks to everyone who tries!



The link: https://mega.nz/folder/awIQwAzT#DjuPTU4QlxkObipC0O_9gQ (I can keep uploading newer files if needed)

try this step by step (read until end):

• Disconnect from internet
• Uninstall gpu driver DDU (no safe mode needed, clean and do not restart).
• Uninstall all the processors on device manager (should be 8 on yours, also when it asks for restart, click on no) like this:
  
  
  
  
  
• Uninstall AMD Chipset Driver in Control Panel (skip if there is none)
  
  
• Restart the pc to bios, and downgrade/update the bios to the one that uses Agesa 1.2.0.3 (or the latest bios but not above 1.2.0.3, anything below AGESA 1.2.0.4, 1.2.0.5, 1.2.0.6). Then go to bios again after update and load default or optimized settings and enable xmp.
  
  
• boot up to windows and install the latest AMD Chipset driver here, reboot and make sure AMD Ryzen Balanced is used on power plan, then connect to internet.
  
  
• Install the latest/stable driver.
  
  *do this all offline until reboot after installing chipset driver, also you may reboot to bios after all of this to set the XMP (and previous settings you did) and make sure ram is on slot 2 and 4 if you use 2 sticks. Download needed files (drivers, also the one on highlighted word) before doing step 1, do the step by orders.

Also please give Thaiphoon burner read and zentimings read of your ram and post it here.

 

[Vule1901]

I have already used ddu and tried different combinations of chispet and gpu drivers, only drivers that dont cause video tdr issues are the oldest on gigabyte site. Keep in mind that I used to use all drivers without issue. Here is the link: https://mega.nz/folder/vs4nDJCA#OZuc_vwCtsvg9X-emXMcrA

Update: I followed steps, not I'm at f61 bios(1203b agesa), latest chipset driver and i did all steps 2 times just bcs i wanted to go "big balls" and try the newest whql driver, it kept crashing with video tdr report windows, like 1, then 4 in a 20 seconds span (image freezes, and then procceed to a black screen that cycles backlight as described in my post but it goes back to windows with amd error message). Altough it did pass timespy it crashed as soon as league of legends loaded (bots), and kept crashing with tdr error no matter what i was using my pc for (no system restarts). So I did all the steps again and changed to the oldest gpu driver on gigabyte site for my motherboard (mb_driver_638_apu_20.10.26_whql). I guess now it's time to wait to see whether a fail occurs (currently at 3200cl14, but 2133 still gave video tdr)

note: I added a txt file containing win report about dxg error to the mega link above

Update 2: The system crash described in the post just happened again during a timespy (it could have also happened while opening file explorer, its random), same windows event error messages I will add another full memory dump file to the mega link in the post as soon as i make some room (upload will take another 90min to 150min)

 

[Colif]

Screen goes black and monitor cycles from pitch black to "backlight is on" a few times while system still runs (audio and calls go through), after which screen light up to a "white static screen" after which system buzzes with a bsod sound but with no blue screen, it either hangs (almost 5 mins when full dump is turned on) before restarting or restarts immediately (sometimes tehre is no memory dump /minidump).

do any errors show in reliability history for the times the screen goes black? It will likely show hardware error as windows sees any unexpected restart as a hardware error.
anything in there about GPU driver not loading?

Do you get any errors using old ram? passing memtest just means ram is good, doesn't mean mb can use it effectively.

is the new ram on the compatibility list - https://www.gigabyte.com/Motherboard/B450M-S2H-rev-1x/support#support-doc (Your cpu is raven ridge)
there aren't many CL14 ram sticks on the Gigabyte listing.

APU has to use system ram to run, its possible that is reason you getting tdr errors at certain timings.

 

[Vule1901]

@Colif
Most common errors are erro id 41 with stopcode 278 as critical, and error id 12, 6008, 7000 and only when radeon software is installed (and it fails with "tdr") error id 1001 (drivers from gigabyte dont include radeon software).
This ram xmp is 3733cl17, I went with it bcs me previous ram wasnt on the qvl list and (i had almost no issues besides few gpu drivers and riot vanguard mess) the price difference of this any any other 3000-3600mhz kit was like 8-15usd and i figured I wont be able to get high fhz but ill tighten the timings (it does boot with only one ram stick in only left slot (A) at xmp settings 3733cl17).
But as things are, I have scheduled an rma for the ram since its under a fresh warranty. If nothing changes I will have it sent today and include qvl lists for the mobo and see if they will replace it (xmp fails memtest86, ye it could be up to cpu but it might just help with passing the rma check, if they do it at all)

 

[Colif]

I didn't see Memtest failing ram?
If you have proof of purchase, all ram has a life time guarantee meaning you can rma it and get new. Generally a memtest fail is all it takes.

could try running Prime 95, it checks CPU & Ram. I don't know chances of it showing anything.
since APU is on CPU, maybe

https://www.guru3d.com/files-details/prime95-download.html
Prime 95 how to Guide: http://www.playtool.com/pages/prime95/prime95.html

it can take a while to run.

 

[Vule1901]

@Colif
I can run p95 for as long as i want to, big, smal ffts, anything passes, and then hours later i could be in a game, class, opening discord, file explorer, watching youtube, etc., and the system would crash as desribed (2133, 3200, other frequencies and different timings, loose/tight, higher voltages, lower). I sure hoped to resolve this without doing the rma (which can take weeks since i bought ram through a 3rd party reseller who bought it in hungary, a lot cheaper (hungary prices)than in stores i have access to)

 

[Colif]

Screen goes black and monitor cycles from pitch black to "backlight is on" a few times while system still runs (audio and calls go through), after which screen light up to a "white static screen" after which system buzzes with a bsod sound but with no blue screen, it either hangs (almost 5 mins when full dump is turned on) before restarting or restarts immediately (sometimes tehre is no memory dump /minidump).

errors with no dumps can mean 2 things
the drive its writing too is the problem
its a hardware error and windows doesn't know what cause is.
page file is too small (I don't think this is reason)

what you describe isn't a windows problem
I think windows only creates minidumps if you have it set to make them.
So if you have them set, it won't make a memory dump, but if you don't have it set, it won't make a minidump. It doesn't make both.

that hang time is odd. its writing to an nvme so it shouldn't take that long.

have you run magician on the Samsung drive and run diagnostic scans? smart? https://semiconductor.samsung.com/consumer-storage/support/tools/
i don't think the Samsung NVME drivers would help, but they are there too.
try scanning WDC drive with this - https://support.wdc.com/downloads.aspx?lang=en&p=279

I have asked a friend to look at dumps but he is asleep now.

 

[Vule1901]

I have tried a clean os on both drives separately (only one plugged at a time since we suspected that the wd is dying) it fails the same way (new drivers tdr error+ "the issue", gigabyte drivers only "the issue")

Minidumps were a default options and while that was set when the crash would happen it would take seconds to restart and on the occasion would there be no minidumps. On a recomendation i turned full dump (so we can post it here) and it takes some time to restart (prbbly bcs its "taking a full dump" 😅). Both drives have firmware up to date and are passing their software tests (nvme passes self tests in bios).
The reason we suspected the wd is the issue is bcs its almost 4 years old with ~60tb written to it (but removing it made no changes).
Thanks, cant wait to hear from You and Your friend regarding the dump file!

:::: 2nd dump file failed bcs crash happened after almost 2 hours of uploading... starting over::::

 

[Koekieezz]

I have tried a clean os on both drives separately (only one plugged at a time since we suspected that the wd is dying) it fails the same way (new drivers tdr error+ "the issue", gigabyte drivers only "the issue")

Minidumps were a default options and while that was set when the crash would happen it would take seconds to restart and on the occasion would there be no minidumps. On a recomendation i turned full dump (so we can post it here) and it takes some time to restart (prbbly bcs its "taking a full dump" 😅). Both drives have firmware up to date and are passing their software tests (nvme passes self tests in bios).
The reason we suspected the wd is the issue is bcs its almost 4 years old with ~60tb written to it (but removing it made no changes).
Thanks, cant wait to hear from You and Your friend regarding the dump file!

:::: 2nd dump file failed bcs crash happened after almost 2 hours of uploading... starting over::::

TDR Issues usually caused by either driver, unstable ram, and bios itself. The old way of this is downgrading waaay back to AGESA 1.0.0.3, but thats was an asrock issue, idk if this could happen on other vendors. Also sometimes amd driver are hit and miss as my friends amd gpu have that issue, trying to find the best one. you could repeat the steps given but using agesa 1.0.0.3 instead. Just update them to 1.2.0.3 or 1.2.0.7 (wait for it for a green light) When you want to upgrade the apu/cpu.

You could try to test the ram at stock speed and xmp speed with this TM5 with Absolut config (open tm5.exe, click on load config, choose absolutnew.cfg and apply, run tm5 as admin) while also running furmark at the heaviest settings at your resolution.See if there is any ram errors and instablity while also stressing your igpu. Usually 16gigs take 1hr and 30mins with this absolut config (3cycles)

 

[Vule1901]

(Un)fortunately i could not postpone the rma process anymore so hours ago i sent the ram kit away, when i receive ram back and if this isnt resooved i will try as you said regarding bios rollback and later on o fcourse plethera of stress tests.
Bad timing on my end (with this post) bcs i already postponed rma few times bcs i had "just a few more things to try before sending to rma".

Regarding the drivers, i tried 10, 20, 30 (i dont rly know anymore) versions and even more combinations (on multiple bios versions, stock/custom/xmp) of gpu and chipset drivers (the ones that used to work dont work anymore).

Legal period for resolving rma is 30 days, could be done in a week or full month.
Thanks though!

 

[Colif]

last 2 BSOD were TDR, both mentioning DirectX as usual

looking at drivers reminded me to update my VM.

you had Afterburner installed, it might have been cause. But it wasn't directly mentioned.

 

[Vule1901]

@Colif, correct, but, I had it installed maybe 2 full days before posting this post and uploadi g files (issues were the same before installing)

 

[Colif]

i don't know how useful it would have been on an APU, since its mostly for discrete cards, not igpu. I wonder what it reported. It can't overclock them as such as its on CPU. Or change fan speeds..

I just thought I mention it.
I think it is ram... guess we will find out.

 

[Vule1901]

Installed it bcs i read smth about a possible solution (cant remember the context), had smth to do with "kernel" option or smth..
And it reported correct info for clock and speed. (Now, the irelevant info 😅)While years back when i first installed it, the clock and utilization were reported at around 15000 and 2milion (cant remember which was % and which was MHz, epic silicon quality 😂)

Edit: only afterburner used to report such foolish numbers, hwinfo64, ryzen master and others didnt fool around with reporting random numbers.

 

[Colif]

thats a nice overclock you have on CPU there...

if cause isn't the ram, it might be the CPU.

 

[Vule1901]

The rma was declined bcs ram "passed 20h memtest and 24h prime blend" according to them (didnt get the info on their setup yet).
I guess ill start all over by going back to f1 bios and systematically go up the laddwr till smth work 😅

 

[Colif]

i would get PC looked at by a store before replacing anything else.

I didn't think they actually tested the ram. I assumed they just replaced it.