Potential GPU problems

Once again i have recently been experiencing some issues possibly with my GPU or some other part. Despite my system has been completely stable the past 2 years, I've experienced two instances of a complete system freeze the past couple days. The first one was while gaming and what happened was the system instantly froze and the screen became a checkerboard of pink/green static squares. The second time however i was just watching a youtube video when it froze, but this time there were no unusual screen artifacts, just the screen as it was.

At first i thought after the second time that there's a possibility that i might have an issue with the RAM instead, but after running 5 instances of Memtest at 2.1gb ram each with each one hitting about 350% passes without any errors, i feel like i'm back to square 1 on the possibility of it being a GPU issue.

Is it possible that streaming videos can cause the gpu to lock up if hardware acceleration is enabled?

Or another question might be if a GPU driver can spontaneously go bad and need a reinstall.

Artifacting squares are bad in most cases. I've seen some caused by drivers and bad programming, but usually it's hardware related in some form. If you can screenshot the pattern - if any - on screen it's actually possible to find some clue. You want to monitor your GPU first for heat using Afterburner or similar software. Next is to do a stress test, benchmark or a "skybox" test while monitoring the card's usage. Consider updating the driver for it completely fresh using DDU.
How to use DDU

1) Yes, but usually you should see a timeout and recovery (tdr) before a total system crash.
2) It can happen during a failed install, corrupt Windows, infected system or if the driver is inherently bad, but they don't spontaneously go bad without a trigger point.

I'll definitely take a picture of it if it happens again and creates that pattern. So far it's only done that once but it's hard to say if it'll repeat. I am definitely hoping the new nvidia drivers fix it but i wouldn't bet on it. Right now i've been running OCCT error test and so far 1 hour in it's 0 errors and no crashes. I have no idea what to make of that.

The best i could describe is that you could literally not see any of the original screen in the checkerboard pattern. It literally covered everything in the pink/green squares.

Yes, try download the latest Nvidia driver. Disconnect from the internet to prevent Windows Update force-updating a driver for you, run DDU in Windows safe mode, click "uninstall and restart (recommended)". Install the driver once you're back in normal mode. This way we know the issue is not caused by a corrupt install. What's the card's usage and temperature when you run the stress test? If you have small icons on the pattern it may be a faulty VRAM module on the card.

Sure, but honestly the best way to stress test a card is to play games because they will utilize the card through different core and memory frequencies. It's easier to find faults other than heat when a card is going all over the place rather than a full load. Same goes for RAM. Benchmarks are also good alternatives. What PSU do you have by the way? And what game did you play when you had this?

I have a 500w FSP Mini ITX PSU. The game that i was playing the first time was Fallout 4. Ran the render test in gpu-z for 10-15m or so. Temperatures hit the mid 70's on the GPU at 100% usage and about 75% on the memory controller. You're definitely right about the game being the best benchmark for stability. I guess it's just the pessimist in me that cannot believe the driver updates may have fixed it. It just seems too easy i suppose.

Due to whatever bad luck i have, in the dozen or so big PC issue's i've had, all of them ended up requiring a hardware replacement.

*Corrected 500w not 400w*

What driver did you run before it? But 400w psu, is it the 80+ gold rated one? It's a decent unit, but it's important to keep in mind that the capacitors wears over time, and you're at the minimum of overall amperage of its rail and wattage. It's possible that there's a power delivery or stability issue to your 1070 from the +12v rail causing it to crash? It's just a theory though, I'm not good at this particular topic, but it's something to keep in mind.

Ah, ok. I'm not familiar with OCCT, but does it mention what error it reports? Maybe updating the driver fixed this? I'm not sure what else I can suggest other than changing hardware, or reverting drivers way back. You should try and ask on the nvidia support forum if no one else in here have suggestions for you. Since this points a lot towards the gpu being faulty I'm sure someone there can point you in the right directions.