NVIDIA GeForce GTX 1060 3GB crashing and unstable in gaming

[runarhalldor]

My pretty old (prebuilt) pc has been crashing for the last couple of years under heavy load (gaming and in furmark). The screen goes black, then comes back but the game is frozen until i force quit it in task manager or it closes automatically later. Usually hearing a windows sound like as if i unplugged something.

 

Checking the system logs reveals nvlddmkm errors, particularly "The description for Event ID 0 from source nvlddmkm cannot be found" and  "Display driver nvlddmkm stopped responding and has successfully recovered."

Ive attached screenshots of those logs.

 

 

 

 

I have tried DDU to fresh install drivers, and ive even replaced the SSD (with no data transfer) for unrelated reasons. 

Ive also ran some sort of RAM test via cmd but i cant remember what kind of test

The only thing that helps is using MSI Afterburner to lower the core clock and memory clock by about ~150mhz each. But this is not foolproof and doesnt work in some cases like Overwatch 2 and will still crash occasionally in games that work.

 

What are some steps i can take to fix this issue or at least kick the can down the road by about 3 months when i hope to be able to buy a completely new build.

 

Specs:

• Intel(R) Core(TM) i5-7500 CPU @ 3.40GHz
• VENGEANCE® LPX 16GB (2 x 8GB) DDR4 DRAM 2133MHz C13 Memory Kit
• NVIDIA GeForce GTX 1060 3GB
• Samsung SSD 870 EVO 1TB
• Unknown Cooler master PSU ( i want to say its 650watts but im unsure)
• Windows 10 Home 21H2(version)  64bit

EDIT: the motherboard is a MSI B250M MORTAR (MS-7A69)

[porina]

41 minutes ago, runarhalldor said:

The only thing that helps is using MSI Afterburner to lower the core clock and memory clock by about ~150mhz each. But this is not foolproof and doesnt work in some cases like Overwatch 2 and will still crash occasionally in games that work.

Could be either PSU is unstable or the GPU is on the way out. Don't think there is much else you can do without replacing either to diagnose further.

[runarhalldor]

1 minute ago, porina said:

Could be either PSU is unstable or the GPU is on the way out. Don't think there is much else you can do without replacing either to diagnose further.

Its been like this for about 1.5-2 years though. You think it could be failing for that long?

[porina]

1 minute ago, runarhalldor said:

Its been like this for about 1.5-2 years though. You think it could be failing for that long?

Maybe it isn't getting worse, but it was marginal to start with. Still would require component swapping to diagnose.

 

You got a workaround of underclocking. Actually, you could play a bit more with that. Instead of adjusting the clocks directly, can you lower the power limit for the card? Does that help?

[runarhalldor]

Just now, porina said:

Maybe it isn't getting worse, but it was marginal to start with. Still would require component swapping to diagnose.

 

You got a workaround of underclocking. Actually, you could play a bit more with that. Instead of adjusting the clocks directly, can you lower the power limit for the card? Does that help?

Im a COMPLETE noob when it comes to overclocking and etc, i pretty much only know how to XMP (cause its one button) . How would i go about doing that?

[porina]

7 minutes ago, runarhalldor said:

Im a COMPLETE noob when it comes to overclocking and etc, i pretty much only know how to XMP (cause its one button) . How would i go about doing that?

In MSI afterburner there are usually 4 sliders:

1. GPU core clock
2. VRAM clock
3. Power limit
4. Fan speed

Sounds like you tried reducing 1 and 2. I'm suggesting instead to try lowering 3. This will lower clocks in a different way which may or may not work better. Of course you'll also lose some performance this way but I'd take it stability would be more welcome.

 

Note MSI Afterburner may have a checkbox set that links 1 and 3 together, so that would need to be unchecked if you want to adjust 3 alone.

 

[runarhalldor]

1 minute ago, porina said:

In MSI afterburner there are usually 4 sliders:

1. GPU core clock
2. VRAM clock
3. Power limit
4. Fan speed

Sounds like you tried reducing 1 and 2. I'm suggesting instead to try lowering 3. This will lower clocks in a different way which may or may not work better. Of course you'll also lose some performance this way but I'd take it stability would be more welcome.

 

Note MSI Afterburner may have a checkbox set that links 1 and 3 together, so that would need to be unchecked if you want to adjust 3 alone.

 

You're referring to this bar?

 

Do you have any recommended value to set it to? or just incrementally lower it until i see a difference?

[porina]

1 minute ago, runarhalldor said:

You're referring to this bar?

 

Do you have any recommended value to set it to? or just incrementally lower it until i see a difference?

That's the one. And my memory was off slightly, the tick box in your screen is the chain icon which is "ON". It actually links power limit and temperature limit, not what I said earlier. Forgot about the temp limit. I'd suggest seeing how low it lets you go and trying that. If that doesn't help, it isn't worth trying other settings. If it does help, you can try increasing it to find a balance between stability and performance. How low it lets you go depends on the GPU and manufacturer. I think I've seen minimums of around 40% and as high as 70%.

 

To recap, reset core clock and memory clock to normal. Unlink the power and thermal limit if it is linked (only one should move). Lower power limit to minimum and see if it is stable. If so, try finding how high you can go and still be stable.

[runarhalldor]

2 minutes ago, porina said:

That's the one. And my memory was off slightly, the tick box in your screen is the chain icon which is "ON". It actually links power limit and temperature limit, not what I said earlier. Forgot about the temp limit. I'd suggest seeing how low it lets you go and trying that. If that doesn't help, it isn't worth trying other settings. If it does help, you can try increasing it to find a balance between stability and performance. How low it lets you go depends on the GPU and manufacturer. I think I've seen minimums of around 40% and as high as 70%.

 

To recap, reset core clock and memory clock to normal. Unlink the power and thermal limit if it is linked (only one should move). Lower power limit to minimum and see if it is stable. If so, try finding how high you can go and still be stable.

I'll do just that. Thank you for the help! is this where i mark the thread solved or do i wait until i see actual results?

[porina]

7 minutes ago, runarhalldor said:

I'll do just that. Thank you for the help! is this where i mark the thread solved or do i wait until i see actual results?

I'll leave it to your judgement but probably want to confirm it works first. Otherwise I'm just guessing