GPU losing signal

metaton · Saturday at 22:49

Hi everyone,

I am in desperate need for some help, I am at the end of my wits.

I had this setup and it worked for more than 6 months without problems. Since last Friday my GPU started to shut down after booting into windows. (couple minutes in). I had an MSI tri oc 5090 rtx (fans stopped, signal lost) I could hear audio playing. Since then I ordered an Asus rog Astral 5090 rtx and that does the same thing (the MSI card has been RMA -d I thought it is hardware failure) except fans running full speed. The error is not always present, mostly happens on cold boot, if I manage to make it work (with reboots, unplugging hdmi using a different display device GPU works under maximum load for hours and hours. No issues. Until the next cold boot, when it starts again and it is doing it repeatedly.

My setup: Ryzen 9800 X3DMobo: MSI tomahawk X870E (Bios version 2.A90)Case: Hyte 70 iRam 32GB Kingston FuryGPU 5090 rtx ROG Astral.PSU:Corsair AX1600i 1600W 80+ Titanium Modular Digital ATX

I tried so far:

Using a different display (Samsung 4k tv, different HDMI cable) still happens.
Using integrated graphics: No issues, works perfectly fine.
removed Nvidia driver with DDU, fresh driver install.
second GPU (I would imagine both GPU-s have the same hardware failure is extremely unlikely.)
Anybody has any idea what is this?

Personality+ · Saturday at 22:56

Motherboard most likely / PSU

metaton · Saturday at 22:58

Thank you, is it a hardware failure you think? Or something which can be solved by cmos reset or bios update? I was thinking I can rule the PSU out as when the GPU works and wont shut down it works fine for hours, even under maximum load?
(Also which part of the mobo?) I got a riser (came with the hyte case) could be the riser the culprit?

Tetras · Saturday at 23:04

Is there anything in the event viewer (time stamped from) when it loses signal?

metaton said:
Since last Friday my GPU started to shut down after booting into windows. (couple minutes in).

Is it always a couple of minutes?

If so, it sounds very suspicious of a software problem, like there is something loading in the background that is breaking the driver. Did you install or update anything recently?

metaton said:
Since then I ordered an Asus rog Astral 5090 rtx and that does the same thing (the MSI card has been RMA -d I thought it is hardware failure) except fans running full speed. The error is not always present, mostly happens on cold boot, if I manage to make it work (with reboots, unplugging hdmi using a different display device GPU works under maximum load for hours and hours. No issues. Until the next cold boot, when it starts again and it is doing it repeatedly

I assume that you checked the power connector is clean and undamaged? Black screen issues are often power connectors, though, you said if you can make it work then it doesn't happen until the next reboot, which doesn't sound like the power connector.

metaton said:
removed Nvidia driver with DDU, fresh driver install.

How easy is it for you to obtain/try a second drive with a completely fresh install?

Did you try a much earlier driver rather than the most recent?

metaton said:
Hyte 70

Are you using a riser?

metaton · Saturday at 23:10

Tetras said:
Is there anything in the event viewer (time stamped from) when it loses signal?

Is it always a couple of minutes?

If so, it sounds very suspicious of a software problem, like there is something loading in the background that is breaking the driver. Did you install or update anything recently?

I assume that you checked the power connector is clean and undamaged? Black screen issues are often power connectors, though, you said if you can make it work then it doesn't happen until the next reboot, which doesn't sound like the power connector.

How easy is it for you to obtain/try a second drive with a completely fresh install?

Did you try a much earlier driver rather than the most recent?

Are you using a riser?

1 It is always a couple of minutes. If it survives the boot then it is stable.
2.checked the connector, it was flawless. Also the MSI card did not have any burns or smell. Now I am using the asus adapter.
3. It is a bit of a hassle, I need to do a complete windows reinstall, I dont have a spare drive, I was waiting to see what you guys say, if it is clearly a hardware issue I did not want to do a reinstall first.
4. I used an older nivida driver which was working fine for months. Until last Friday. Now I am using the latest nvidia driver, I removed the old one with the DDU.
5. Yes I am using the hyte riser which came with the case.

Personality+ · Saturday at 23:12

Try connecting the GPU directly with mobo

metaton · Saturday at 23:13

Event viewer comes up with this: Event 41: The system has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed, or lost power unexpectedly. -Could have been me turning off the pc after losing signal or is this the reason while lost signal?

Tetras · Saturday at 23:15

metaton said:
-Could have been me turning off the pc after losing signal or is this the reason while lost signal?

Yeah, they're normally worthless, you'd need to look before that.

metaton · Saturday at 23:15

Also I got administrative events: distributedcom error 10016

metaton · Saturday at 23:17

And event 1000 Application error
Faulting application name: nvcontainer.exe,
Faulting module name: NvBackend64.dll, version

Tetras · Saturday at 23:18

Personality+ said:
Try connecting the GPU directly with mobo

Changing PCIE gen in bios can also help with a bad riser, worth trying for free/easy.

metaton said:
Also I got administrative events: distributedcom error 10016

probably nothing

metaton said:
And event 1000 Application error
Faulting application name: nvcontainer.exe,
Faulting module name: NvBackend64.dll, version

Was that close to the black screen?

metaton · Saturday at 23:18

Personality+ said:
Try connecting the GPU directly with mobo

Thanks it is a bit difficult as I have a riser and the hyte i70 case only have vertical slots, I cant fit the 5090 horizontally.

metaton · Saturday at 23:21

Tetras said:
Changing PCIE gen in bios can also help with a bad riser, worth trying for free/easy.

probably nothing

Was that close to the black screen?

I only see this once. But this one is frequent: Event 7009 A timeout was reached (45000 milliseconds) while waiting for the asComSvc service to connect.

metaton · Saturday at 23:23

Tetras said:
Changing PCIE gen in bios can also help with a bad riser, worth trying for free/easy.

probably nothing

Was that close to the black screen?

Which one I should change it to?

Tetras · Saturday at 23:25

metaton said:
I only see this once. But this one is frequent: Event 7009 A timeout was reached (45000 milliseconds) while waiting for the asComSvc service to connect.

Hmm, this appears to be related to Asus software.

Doesn't sound likely to cause a black screen, but if it happens every time then that's worth checking out.

Can you list all the apps you use that can interface with the board/graphics card?

Anything that loads on start-up is relevant, e.g. monitoring software (like afterburner), anything with an overlay, board/graphics manufacturer software including RGB/ARGB, driver updaters, mouse/keyboard utilities, hardware accelerated apps (like Discord).

I'd try to disable everything that starts when the PC runs, to rule software out of why that problem happens after 2 minutes every boot.

metaton · Saturday at 23:30

MSI center, Asus GPU tweak, Nvidia app, L-connect (Lian-li fans) NZXT cam, Sound blaster command (?), Hyte nexus, Wallpaper engine - I guess this is all. Do you think it could be software related?

metaton · Saturday at 23:31

None of this is recent install, except the GPU tweak for the new GPU, but the errors started before that.

Tetras · Saturday at 23:34

metaton said:
Do you think it could be software related?

If it happens nearly bang on 2 minutes every boot, then that is my best guess. That something is loading with Windows that is breaking the driver.

metaton said:
None of this is recent install, except the GPU tweak for the new GPU, but the errors started before that.

Aye, but there's a few reasons it could happen suddenly: bad auto update (most apps auto update nowadays), conflicting windows update or a background driver update.

metaton · Saturday at 23:35

Tetras said:
If it happens nearly bang on 2 minutes every boot, then that is my best guess. That something is loading with Windows that is breaking the driver.

Aye, but there's a few reasons it could happen suddenly: bad auto update (most apps auto update nowadays), conflicting windows update or a background driver update.

I checked what I had on the 22nd was windows security updates. That is the same day errors started happening. I am not sure if it could be related or not.

Tetras · Saturday at 23:41

metaton said:
I checked what I had on the 22nd was windows security updates. That is the same day errors started happening. I am not sure if it could be related or not.

This is what Google AI overview says, for whatever that's worth

Recent nvcontainer.exe crashes following May 2026 updates are primarily caused by conflicts between the new NVIDIA App and legacy GeForce Experience installations, or by critical ETW kernel pool exhaustion linked to the built-in FrameView SDK. [1, 2]

Immediate Workarounds

Uninstall GeForce Experience: Having both the old GeForce Experience and the new NVIDIA App installed simultaneously is known to cause nvcontainer loop crashes. Completely uninstall the old application. [1, 2]

Clean Reinstall Drivers: Completely remove your current graphics drivers using Display Driver Uninstaller (DDU) in Windows Safe Mode, then perform a fresh installation of the latest NVIDIA Drivers. [1, 2]

Disable the Service: If crashes continue, you can safely disable the "NVIDIA Display Container LS" service via services.msc. This resolves the crash without impacting gaming or GPU compute performance. [1]

For a visual guide on performing a clean driver reinstallation using DDU to clear out corrupted software states: