[SOLVED] Nvidia drivers getting corrupted?

[Joel_228]

AMD 5800x
Corsair Vengeance 32GB 18cl running at DOCP
Zotac Trinity 3090
Strix X570-F Mobo

It all started when I was hearing crackling in my speakers sometimes. The issue got worse and I searched the entire internet and tried all the fixes. The only thing that worked was to reinstall the Nvidia sound drivers along with the graphics driver because I can't seem to find how to do it separately. This completely fixes the issue but it comes back after a couple days. And now I have had BSOD and an issue where half my screen was on the wrong side and the whole screen was flickering. Restarting and fix that. I used DDU more than once now to completely clean out the drivers and then reinstall them fresh. The screen flickering issue and sound issue return again. I have used AMD's software to do a mild overclock And also MSI afterburner to do a mild overclock on the GPU. I have a custom loop and everything runs very cool. I obviously can't keep reinstalling my drivers every other day so I'm wondering should I completely reinstall windows? Any ideas what to try? I unfortunately can't read what it says when I get the BSOD because it restarts so fast I don't even get a chance to see it. Any help would be appreciated.

 

[hang-the-9]

Sounds more like the video card is failing than anything with the drivers. I'd reset everything to default settings and see how it runs. Did you check temps to make sure nothing was overheating.

 

[Joel_228]

Sounds more like the video card is failing than anything with the drivers. I'd reset everything to default settings and see how it runs. Did you check temps to make sure nothing was overheating.

Thanks for the reply. Temps stay low. CPU is 60C while gaming. GPU gets to around 40C. 42-44C under stress. Last I fiddled with it was last night. I'll make sure everything is back to default and see if that fixes anything. It's just so weird that reinstalling the drivers fixes the issue temporarily.

 

[geofelt]

Some thoughts:

I would not bother with helper apps such as DDU. You can get unwanted software installed.
The Nvidia driver install process will do a clean install for you.
Select a custom install option.
That lets you not bother with graphics card sound drivers.
They are not needed unless you are planning on sound from you monitor which is usually crummy.
You can probably omit physx and 3d also.
Do you really need the nvidia experience app?

Do you really need to OC a 3090 gpu?
Graphics card vendors bin their chips and use the better ones in factory overclocked versions
that they can sell for more. You may be able to OC more, but then again, perhaps not.
If your 3090 was a top bin, it would come already factory overclocked.
Why bother?

Remove the oc on the 5800 until you get stability sorted out.

Reinstall the driver software for your sound adapter.
Download directly from either the motherboard or sound adapter source.

 

[Joel_228]

Some thoughts:

I would not bother with helper apps such as DDU. You can get unwanted software installed.
The Nvidia driver install process will do a clean install for you.
Select a custom install option.
That lets you not bother with graphics card sound drivers.
They are not needed unless you are planning on sound from you monitor which is usually crummy.
You can probably omit physx and 3d also.
Do you really need the nvidia experience app?

Do you really need to OC a 3090 gpu?
Graphics card vendors bin their chips and use the better ones in factory overclocked versions
that they can sell for more. You may be able to OC more, but then again, perhaps not.
If your 3090 was a top bin, it would come already factory overclocked.
Why bother?

Remove the oc on the 5800 until you get stability sorted out.

Reinstall the driver software for your sound adapter.
Download directly from either the motherboard or sound adapter source.

I tried DDU because I'm having a problem. I always do a clean install.
I use the Nvidia sound through the GPU to my 65 inch OLED with ARC passthrough to a very high end surround system.
I'm not sure why you mentioned GeForce experience. I dont have it.
I put a very mild OC on my 3090 because I can. It's been very stable until recent. And it runs very cold. Maybe it caused a problem? It would surprise me at +85 and +200.
I will remove the AMD Ryzen Master OC later and see if that helps. But again it's been very stable and runs cool. Thanks for your help. I'll report back if anything changes tonight.

 

[Karadjgne]

DDU should be run through windows safe mode since you cannot delete drivers that are actively in use.

PhysX, anything 3d are a vital part of the cards usage, they are highly integrated with DLSS and Raytracing as well as other particle aspects.

 

[geofelt]

Out of curiosity, is your TV one of the LG OLED CX units?
I have the 48" version and all works well with one possible issue.
When the pc puts it to sleep, it actually powers off and will not wake unless it is powered on first.

Your motherboard has integrated sound.
What happens if you use that sound adapter?
I guess it could be awkward if you alternate between pc usage and TV watching.

 

[Joel_228]

Out of curiosity, is your TV one of the LG OLED CX units?
I have the 48" version and all works well with one possible issue.
When the pc puts it to sleep, it actually powers off and will not wake unless it is powered on first.

Your motherboard has integrated sound.
What happens if you use that sound adapter?
I guess it could be awkward if you alternate between pc usage and TV watching.

I have the LG C9 65". I don't have my PC set up to shut down the monitor. I haven't actually played with how that would work. But the TV goes in sleep mode where it shows a firework type display in different areas on the screen. That's a good question about using the onboard sound. I would probably be fine with that. But having everything run through the ARC to the sound bar is kinda nice. The issue now though is the BSOD that came out of nowhere.

 

[Joel_228]

Update: Running Ram, GPU, and CPU at stock default speeds. I have MSI Kombuster running. Fifteen minutes in...

 

[Karadjgne]

What was the bsod? If you look in Windows event viewer, it'll often have a reason/culprit, so can actually point in a certain direction for a fix.

Msi Kombuster is pretty decent imho, but I don't just run the fuzzy donut either. Run the full gamut of tests because they'll use different sections of the card more specifically. That helps identify if there's an instability in power delivery vs vram in the OC etc and can save hours of guessing games, windows reinstall, driver redo's etc.

OC is a gamble, never a guarantee. Even with the top line models, there are silicon limits depending on batch. Your gpu might have binned successfully for the cards station, but can easily be unstable at 1Hz more. You'd think a +200 would be ok, but could simply be that your card cannot accept or becomes unstable with that amount.

TechPowerUp did a review on my old 660ti. The Samsung ram was rated at 8k. They successfully got to 7960 I believe. My card got to 7822 and not a hair over. It also topped out at 14% power limit, even though it could be pushed to 25% on the slider. Too much power made the LLC erratic. But all of that was above stock clocks, so it's actual limits are an unknown other than what I found on my card, which was identical model, but a different batch of silicon.

 

[Joel_228]

What was the bsod? If you look in Windows event viewer, it'll often have a reason/culprit, so can actually point in a certain direction for a fix.

Msi Kombuster is pretty decent imho, but I don't just run the fuzzy donut either. Run the full gamut of tests because they'll use different sections of the card more specifically. That helps identify if there's an instability in power delivery vs vram in the OC etc and can save hours of guessing games, windows reinstall, driver redo's etc.

Thanks for the tip. I am not sure what I am look at here so I took a screenshot.

https://imgur.com/dhfX1Sq

View: https://imgur.com/dhfX1Sq

Not sure why I cant add image. looking that up now.

 

[Joel_228]

The one marked "Citical" I believe was when my pc last locked up.

 

[Karadjgne]

Kernal Power event 41 (63). That's a doozy. It basically states the cpu didn't receive the power it expected, so shut down. There's multiple reasons this can happen, many related to sleep modes. Could be as simple as setting hdd to sleep before the cpu, fast startup setting on nvme/ssd boot drive. Or could be a system driver conflict/corruption.

I'd definitely go through the power/advanced power settings and make sure everything is straight. Sleep is fine, but it's a pc so totally disable hybrid sleep/hibernation as those are battery saving stuff for laptops.

I'd also disable fast startup, both in any motherboard software, bios and possibly power settings. You aren't reliant on a hdd for boot, and fast startup is known to possibly be buggy with ssd's.

From a CMD with admin, type SFC /scannow (enter) and see if that grabs any errent system file conflicts/bugs.

Can also type DISM /Online /Cleanup-Image /Restorehealth (enter) and see if that picks up any other windows conflicts.

Personally, I never use sleep, and my pc runs 24/7. It does just fine in balanced power mode (drives set to sleep never) which keeps it at idle when not in use, but monitors shut down after 5 minutes.

Clarification: the gpu stops sending a signal after 5 minutes. A monitor that doesn't receive a signal will shutdown on its own. So when hooked to a TV, what you'll get is the standard blurb about no signal, prompting the TV to go into its Standby mode. Some Tv's that means a blank screen, off screen, Screensaver or combination. My TV goes to a rotating backdrop of trees for 2 hours, after which it shuts off the display until toggled or remote activated.

 

[Fiorezy]

Running cool after an overclock doesn't mean it is safe because there is a chance that you pushed unsafe voltage for the cpu/gpu which could cause a potential damage or degradation.

Bottom line, you have a beast, oc is really uneeded and it is more of a hustle than anything else. I mean frames on a game that is already running on high fps? or 2 seconds less to render a video? It is not worth it mate.

My advice is to have everything at stock settings except XMP, use it for a few hours and see if the issue occures again, if it happens again, I think you need to perform a clean reinstall of Windows.