• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

RX 9070 Random System Freezes / LiveKernelEvent 141 / AMD_WATCHDOG / WHEA 18 - After months of troubleshooting I'm out of ideas

W

winnlonzo

New Member
Joined
Aug 4, 2026
Messages
1 (1.00/day)
Hi everyone,


I'm looking for help because I've been troubleshooting this issue for several months and I've reached the point where I don't know what else to test.


I'll try to include absolutely everything I've done so nobody has to guess.



My PC​


  • GPU: AMD Radeon RX 9070 (non-XT)
  • Previous GPU: Radeon RX 6600
  • CPU: Ryzen 7 5700X
  • Motherboard: MSI MAG B550 Tomahawk
  • BIOS: Latest available (E7C91AMS.AK1)
  • RAM: 32 GB (2x16) Lexar Thor DDR4 3200 CL16 (XMP)
  • PSU: Cooler Master MWE Gold V3 850W (brand new)
  • Windows 11
  • 3 monitors:
    • 2x Titan Army 180 Hz
    • 1x 60 Hz monitor


The issue​


The RX 9070 randomly crashes.


Sometimes while gaming.


Sometimes while watching YouTube.


Sometimes while simply sitting on the Windows desktop doing absolutely nothing.


The crashes are completely random.


Sometimes after 2 hours.


Sometimes after 24 hours.


Sometimes after several days.



What happens during the crash​


The sequence is almost always identical.


  1. Entire PC freezes.
  2. Audio loops / glitches.
  3. Discord freezes.
  4. YouTube freezes.
  5. Mouse and keyboard stop responding.
  6. Both 180 Hz monitors lose signal.
  7. The 60 Hz monitor sometimes briefly turns green before also losing signal.
  8. Sometimes after 20 seconds to 5 minutes the GPU driver recovers.
  9. Windows keeps running in the background.
  10. I usually have to reboot because the monitors never wake back up.

Sometimes Device Manager shows the GPU normally.


Sometimes it disappears.


Sometimes Windows reports Code 22 or Code 31 until another reboot.



Windows Errors​


I consistently get combinations of:


LiveKernelEvent


  • 141
  • 193
  • 1b0
  • a1000001
  • a2000002

AMD creates:


  • AMD_WATCHDOG
  • AMD_REPORT_UM

I also have dump files such as:


AMD_WATCHDOG-xxxxx.dmp


WATCHDOG-xxxxx.dmp


AMD_REPORT_UM-xxxxx.dmp



Event Viewer​


The most interesting error is:


WHEA-Logger Event ID 18


Reported by component:
Processor Core


Machine Check Exception


Cache Hierarchy Error


Processor APIC ID: 3


This WHEA appears when the RX 9070 crashes.


I don't normally see WHEA errors otherwise.



The strange part​


The GPU eventually comes back.


After another reboot Windows detects it again.


Sometimes Device Manager says "This device is working properly."


Then everything works...


Until the next random crash.



What I've tested​


Two different RX 9070​


The first RX 9070 was returned.


I received a brand new replacement.


Exactly the same problem.



RX 6600​


I reinstalled my old RX 6600.


It works perfectly.


No crashes.


No freezes.


No driver recovery.


No random black screens.



PSU​


I suspected the PSU.


Bought a brand new Cooler Master MWE Gold V3 850W.


No improvement.



PCIe slot​


Tested another PCIe slot.


No improvement.



BIOS​


Updated BIOS to latest version.


Loaded Optimized Defaults.


Tested:


  • PCIe Auto
  • PCIe Gen3

No difference.



Resize BAR​


Tested:


Enabled


Disabled


No improvement.



Above 4G Decoding​


Enabled


Disabled


No improvement.



RAM​


Lexar Thor 3200 CL16


Tested:


XMP enabled


XMP disabled


No improvement.



Drivers​


Clean installation multiple times.


DDU used.


AMD Adrenalin reinstalled several times.


Different driver versions tested.


No lasting improvement.



Windows​


Windows fully updated.


Chipset drivers updated.


Motherboard drivers updated.


No change.



Monitors​


Originally using:


2x 180 Hz


1x 60 Hz


Tested with only one monitor connected.


Exactly the same issue.



Electrical issue​


At one point I suspected a mini-fridge connected to the same power strip because it caused USB devices to briefly disconnect.


I completely separated:


PC + monitors


from


Mini-fridge and other devices


using two independent Legrand power strips.


No improvement.



What I noticed​


The crash is not simply:


"display driver stopped responding"


The entire system freezes.


Audio loops.


Windows freezes.


Everything becomes unresponsive.


Sometimes Windows recovers after several minutes.



AMD Driver versions​


Tested several Adrenalin versions including the newest available.


Still happening.



Community research​


I've found multiple users with:


RX 9070


RX 9070 XT


LiveKernelEvent 141


AMD_WATCHDOG


Black screens


Random freezes


Sometimes even WHEA Cache Hierarchy Error.


So I'm clearly not alone.



What I'm thinking​


At this point I see several possibilities:


  1. Faulty MSI B550 Tomahawk motherboard
  2. PCIe / motherboard issue only exposed by the RX 9070
  3. CPU / motherboard interaction
  4. RX 9070 driver bug
  5. Some compatibility issue between RDNA4 and my platform

I don't really believe it's:


  • PSU
  • RAM
  • Windows installation
  • the GPU itself

because:


  • two RX 9070 behave identically
  • RX 6600 is perfectly stable


Next plan​


At the moment the RX 9070 is back in its original box.


I'm planning to buy another motherboard (probably X570S) and keep absolutely everything else identical.


Same:


CPU


RAM


PSU


SSD


Windows


GPU


This way I'll only change one component.



Questions​


Has anyone experienced exactly this?


Especially:


  • RX 9070
  • Ryzen 5000
  • B550/X570
  • LiveKernelEvent 141
  • AMD_WATCHDOG
  • WHEA Event 18 Cache Hierarchy Error
  • GPU eventually recovering after several minutes

Could this still realistically be the motherboard?


Or is there anything else I haven't thought of after months of testing?


Any help would be greatly appreciated.


Thanks!
 
winnlonzo said:
Could this still realistically be the motherboard?
I think it probably is. Good idea ordering an x570 to test that.

Another thing to try would be to install Bazzite and see if the issues are happening. Then you could 100 percent rule out OS issue.
 
If you see "WHEA 18 Cache Hierarchy" you absolutely have a CPU issue. I've never seen it be anything else and I've RMA'd at least 15 CPUs for this issue, AMD have never once disputed it and always just dispatched me a brand new CPU. I assume they test the CPUs I send off to them, and based on that assumption that means that all of them have been accepted as RMAs because they do in fact test as faulty.

I'm just guessing here, but I suspect if the 8-lane RX 6600 is fine and the 16-lane RX 9070 is not, then you have a defective PCIe lane on your CPU's I/O die.
  • Have you re-seated the CPU just to rule out a bad connection on one of the pins?
  • Do you have another AM4 CPU you can test with?
  • Is your CPU under 3 years old (retail box) or 1 year old (tray/OEM) for a potential RMA?
 
VuurVOS said:
Like @Let3r stated it might be an OS issue, did you try a clean install? Did you try a CMOS clear and disabling Precision Boost Overdrive?
OP has updated to the latest BIOS which is in almost all circumstances a CMOS clear. For MSI "Load Optimized Defaults" is typically a safe-mode with JEDEC timings and strict CPU TDP/PPT/EDC limits adhered to.
 
just saying that is the same


MSI calls it 4g decoding in the gpu section

--

MSI is bad in regards of radeon 9070.

Feel free to look up my posts here.

ASUS PRime radeon 9070 no post screen, needs password typed without a screen to load the operating system which loads teh firmware and reinitalises the card. Or a reset of the motherboard to reinitalise the graphic card
Card returned

acer predator bifrost 9070, fine wiht one monitor. but two monitors quite often, does not boot with the smae motherboard firmware.
for some reason, twho different edid and firmware asus proart monitors, one on hdmi and one on dp, the motherboard does not find the gpu or the screens quite often.

msi has issues with the RADEON 9070.

i wired a 2 meter cable and an external reset switch to reset my motherboard quite fast when it hangs during cold boot. sometimes the motherboard does not even react on the reset. needs power off button - long pressed than another cold boot

msi has definitely problems with the uefi firmware. https://www.msi.com/Motherboard/X670E-GAMING-PLUS-WIFI/support

--

You did not write if you are using Stock settings. Your tomahawk board afaik should be the same as my previous owned and sold msi b550 gaming edge wifi. stock values, you may touch and chagne secure boot and resizeable bar, tpm settings, but do not touch anything else. i consider your motherboard one of the better ones, what i would buy

-- when your fridge is connected to the same fuse than it does not matter than you are using different power strips. you need to use another phase of your house. usually in europe 3 phase. unplug the fridge or when that is not possible move the fridge to a phase which is not used for your computer. that also implies for all the light switches etc.

--

A clean windows install should be the way to go with a proper storage, no bad sectors when you sorted out your fridge that it does not use the same fuse and the same phase as your computer.

-- am4 is much easier to memory test. get that free bootable iso and let it bootup and memory test over the night. or get a compiler toolchian and proper setup to stress test your cpu and dram. that should show fragile cpu and dram, regardless the root cause, hardware damage, bad connection, wrong uefi settings because some guy stated these are sane values, etc.
 
Last edited:
Chrispy_ said:
OP has updated to the latest BIOS which is in almost all circumstances a CMOS clear. For MSI "Load Optimized Defaults" is typically a safe-mode with JEDEC timings and strict CPU TDP/PPT/EDC limits adhered to.
CMOS clear AND disabling Precision Boost Overdrive. System can be unstable with PBO enabled (even with default TDP PPT EDC limits).
By disabling PBO you will eliminate a source of instability.
 
WHEA-Logger Event ID 18

CPU internal cache/memory adjacent failure.

Test, replace the cpu.
Double down, replace memory and cpu.

The GPU driver can be restarted at any time.

Windows key + Shift + Ctrl + B simultaneously.

18 months.

All the information, + the parts you've invested in, Whea errors in general aren't good.

But AMD has always had that weird clock errata thing going on.
 
You must log in or register to reply here.

Similar threads

N3utro
OCCT VRAM test immediately crashing on a 9070 XT
Replies
0
Views
316
N3utro
N3utro
H
  • Question Question
RTX 3080 black screen only while gaming - LiveKernelEvent 141 / nvlddmkm
Replies
0
Views
84
hussainaseeri
H
P
RX 9070 not detected on Gigabyte B650M Gaming WiFi (works in other system)
Replies
3
Views
395
Ripcord
R
A
RX 570 renders normally but HDMI/DP output is intermittent — “No EDID read” and empty DDC/I2C bus
Replies
4
Views
137
Angell2
A
K
RX 5700 XT No post
2 3
Replies
61
Views
663
Valser
V