I've got an anomalous GPU issue I'm curious if can be resolved

I'm prepared to go into as great of detail as is necessary, but I'll just summarize for right now to gauge if it's worth the endeavor so I'll try to be concise.

I'm on windows 11, which from as far as I can tell (hard to pinpoint the issue due to its oddity) was when this issue began, back when 11 first became available outside of beta, when I switched from 10 to 11. I've got an Asus 8GB RTX 2060 Super, an i9-9900k, 32GB gskill trident 3200Mhz RAM (8gb 4x), in an MSI Z390-A Pro.

My issue is that seemingly unprovoked, my GPU has manifested this behavior that, when something occurs that would normally crash the video driver (mostly just by asking way too much from the card, 4k ray tracing, whatever, coupled with a perhaps overzealous at times core clock adjustment from afterburner(none of which is new as far as my use of the card and the before and after of this issue is concerned)) my audio will persist streaming content (from a background youtube video) but my screen is frozen, I can move the mouse, but other than that, the system is entirely unresponsive to anything other than a hard power reset, when before it would just crash the video driver, close the offending program, and restart video all on its own without any major loss of anything other than whatever I might have been doing in game.

Long story short, instead of crashing under an unstable load, my GPU now seems to be causing a pseudo-freeze rather than killing the process that instructed the load.

My event viewer seems to be uncooperative in this matter due to the only error I can see (with my limited expertise with the tool) is a critical that occurs upon restarting stating along the lines of an improper or unexpected shutdown or loss of power occurred (I forget the exact language, but that's the gist)

Any insight would be greatly appreciated, even if it may not lead to resolution, and I'm an open book in terms of any sort of logs or questions or checks anyone might want to see or have me try. Thanks in advance!

Is Windows up-to-date including all drivers? Especially the graphics drivers. Also are the graphics drivers not beta test versions but 'stable' versions?

Yeah, everything's up to code. I should have prefaced with the disclaimer that I'm a savvy user, been trial and erroring my way to a thorough understanding of tech for going on probably 20+ years lol.

I'm the quintessential "one everyone calls for tech support" here in my small corner of the world

If you are a 'perfomance' user then could it be heat problem? Maybe look inside and remove dust that may be lying on the CPU/GPU/Memory/SSDs?

Just a thought but I am out of ideas now so I hope someone else can help you. There are some first class experts on this forum.

Again, everything is in great condition. I regularly clean my components, as well as maintain fresh applications of thermal pads on the GPU's memory modules and mosfets as well as thermal compound on the die. My cpu uses a 360mm CLC, and my graphics card can breathe very well the constant fresh air provided by the positive air pressure environment produced by my configuration.

Hello Defuse, I could be wrong but after taking a look at the specs of your system, if the problem is not the GPU itself, i have the feeling that your issue could be related to your PSU, I think that 650W may not be enough to power your hardware properly.

You use MSI Afterburner. Have you tried MSI Kombustor to stress test the graphics card too?

It displays the GPU temperature. More detailed monitoring is available from GPU-Z.

Hello Again Defuse, here is a great link to calculate the required power supply wattage.

defuse said:

I'm prepared to go into as great of detail as is necessary, but I'll just summarize for right now to gauge if it's worth the endeavor so I'll try to be concise.

I'm on windows 11, which from as far as I can tell (hard to pinpoint the issue due to its oddity) was when this issue began, back when 11 first became available outside of beta, when I switched from 10 to 11. I've got an Asus 8GB RTX 2060 Super, an i9-9900k, 32GB gskill trident 3200Mhz RAM (8gb 4x), in an MSI Z390-A Pro.

My issue is that seemingly unprovoked, my GPU has manifested this behavior that, when something occurs that would normally crash the video driver (mostly just by asking way too much from the card, 4k ray tracing, whatever, coupled with a perhaps overzealous at times core clock adjustment from afterburner(none of which is new as far as my use of the card and the before and after of this issue is concerned)) my audio will persist streaming content (from a background youtube video) but my screen is frozen, I can move the mouse, but other than that, the system is entirely unresponsive to anything other than a hard power reset, when before it would just crash the video driver, close the offending program, and restart video all on its own without any major loss of anything other than whatever I might have been doing in game.

Long story short, instead of crashing under an unstable load, my GPU now seems to be causing a pseudo-freeze rather than killing the process that instructed the load.

My event viewer seems to be uncooperative in this matter due to the only error I can see (with my limited expertise with the tool) is a critical that occurs upon restarting stating along the lines of an improper or unexpected shutdown or loss of power occurred (I forget the exact language, but that's the gist)

Any insight would be greatly appreciated, even if it may not lead to resolution, and I'm an open book in terms of any sort of logs or questions or checks anyone might want to see or have me try. Thanks in advance!

Click to expand...

When this occurs, does the Video signal reset (Windows key + Control + Shift + B) work to reset and bring it back?

johnlgalt said:

When this occurs, does the Video signal reset (Windows key + Control + Shift + B) work to reset and bring it back?

Click to expand...

Hmm, I'll have to get back to you on that. I've never known about that shortcut. To be clear, I can move the mouse and see it moving, but aside from that and my background youtube video not stopping, it's unresponsive. I might be imagining this, but I think I can recall that the scroll/number/caps lock buttons don't change when pressed, but I might be conflating a different thing with this. I'll surely try that though if the opportunity presents itself

Gas said:

Hello Defuse, I could be wrong but after taking a look at the specs of your system, if the problem is not the GPU itself, i have the feeling that your issue could be related to your PSU, I think that 650W may not be enough to power your hardware properly.

Click to expand...

I went and checked out the calculator and I'm well within appropriate levels, I even went heavy on the details ensuring that I was comfortably within what the calculator suggested.

bobkn said:

You use MSI Afterburner. Have you tried MSI Kombustor to stress test the graphics card too?

It displays the GPU temperature. More detailed monitoring is available from GPU-Z.

Click to expand...

I think you've missed the point. Though I am aware of these tools.

A lot of people don't know about the Graphics reset, I only learned of it in the last 2 or 3 years myself lol.

As for Bobkn's post - I think he is suggesting that you examine your GPU performance more closely, and see if there are any possible runaway cores that are spiking heavily with usage when there is no actual use being performed, or something like that. Hence the suggestion for the stress testing of the GPU.

Another really stupid thing that was giving me almost similar issues was using HDMI 2.1 off my 3080 Ti for one of my monitors, the primary one that I game on. I swapped it to DP 1.4 (the secondary monitor was already on DP) and 90% of my issues with displays, and particularly your exact issue, have not occurred since. FWIW, I'm using a pair of Eve 4K Monitors, one matte and one glossy, and neither likes to be on HDMI at times.

defuse said:

I think you've missed the point. Though I am aware of these tools.

Click to expand...

My point was that stressing the GPU may offer a more systematic means of diagnosing the problem. If you can beat on the card without a crash, then it's probably a software issue.

johnlgalt said:

A lot of people don't know about the Graphics reset, I only learned of it in the last 2 or 3 years myself lol.

As for Bobkn's post - I think he is suggesting that you examine your GPU performance more closely, and see if there are any possible runaway cores that are spiking heavily with usage when there is no actual use being performed, or something like that. Hence the suggestion for the stress testing of the GPU.

Another really stupid thing that was giving me almost similar issues was using HDMI 2.1 off my 3080 Ti for one of my monitors, the primary one that I game on. I swapped it to DP 1.4 (the secondary monitor was already on DP) and 90% of my issues with displays, and particularly your exact issue, have not occurred since. FWIW, I'm using a pair of Eve 4K Monitors, one matte and one glossy, and neither likes to be on HDMI at times.

Click to expand...

My thing happened again earlier in the day, and while I can confirm that the number/caps/scroll lock keys are all still alive enough when it happens that they'll appear to still switch on and off when pressed (the light on the keyboard corresponding to them goes/stays on/off when I press them) and yet the reset shortcut keys as you've described was unfortunately uneventful (nothing happened, as per usual). Just wanted to add that clue to the thread.

On another note, sorry to Bobkn if I came off sounding like a dickhead, I hadn't meant anything by it if I sounded a bit terse.

And lastly, it has actually never occurred to me that just the hdmi could have somehow borked...I'd feel dumb if a replug would do the trick.
I've only ever used this pc (2060 super) hdmi into my Samsung UN55MU6500 (55" curved 4k tv) which is only 2.0, as is the gpu. Though the fans on the card relaxing when the freeze happens leads me to believe it won't be so trivial.

Anecdotally, it feels as if among the myriad services that windows has for every little thing, the one who kills the process or attempts to do its job to recover the shell, or perhaps even the one who's paying any attention to whether anything's even gone wrong at all is failing to take action (or be triggered, or successfully execute, whatever). I don't know, just spitballing in case it sparks the odd thought in someone.

edit: I suppose my next task will be to log some stress tests and see what turns up.