Issue with PC freezing under heavy usage, causing restarts and occasional crashes
Issue with PC freezing under heavy usage, causing restarts and occasional crashes
My computer tends to freeze unexpectedly during intense tasks such as gaming or training neural networks. Up to now, I've experienced only one complete crash and a few restarts, along with many random interruptions. These issues haven't occurred while the system was idle or under minimal load. Even adjusting game settings to low levels didn’t prevent the problems.
The freezes typically start after returning from a month-long break. Prior to that, my PC operated smoothly, handling demanding applications effortlessly—even in summer. However, after months of inactivity (not powering it on or off), these instability episodes appear.
During a freeze, the monitors often go black and display "no signal found." After a short pause, they come back on, showing a frozen snapshot of what was being viewed before the loss of connection. This snapshot usually appears blurrier, with faint rainbow patterns and reduced clarity overall. Despite this, audio remains audible. I’ve managed to stay connected during calls, talking clearly with friends.
To resolve these freezes, I must perform a hard reset of my PC. So far, no error logs appear in the event viewer except for one message indicating an unexpected system restart. When the system restarts, it behaves similarly—random shutdowns instead of freezing.
When a crash occurs, the screen becomes pixelated and displays strange rainbow effects around the edges. Audio also stutters. Despite these issues, I’ve tried several fixes without success:
- Re-seating the GPU and RAM
- Resetting the CPU
- Monitoring temperatures with gpu-z (temperatures were high, sometimes reaching 100°C)
- Performing a Windows RAM test (no issues found)
- Cleaning the PC
- Checking system specs and components
I’m unsure what might be causing this behavior. I lack a spare PSU and don’t know how to systematically test each part. Any guidance on troubleshooting would be greatly appreciated.
Welcome to the forums, newcomer!
Here are your system details in the requested format:
CPU: intel i9-9900K
CPU cooler: not specified
Motherboard: Asus Rog Strix z370-H Gaming
PSU: Corsair HX1000
GPU: RTX 3070-TI
Monitor: not specified
Please include the age of your PSU along with its make and model. BIOS version for your motherboard at this time.
What driver version are you using for your GPU?
Monitor temperatures with gpu-z (before and after CPU repair) – both cases showed temperatures reaching around 95 degrees, sometimes up to 100+ for short periods. I should note that my PC managed these temps well over the past few years (at least I believe so).
Abrupt reboots during heavy use will likely occur due to either:
1) an overheating issue
2) your PSU can’t supply enough power under load.
Also, could you clarify what timeframe you mean by “the past few years”?
Remove your GPU, link your monitor to MoBo, and try your best to trigger BSoD or freeze.
Currently, you can't run GPU-heavy applications, but you can still test your CPU performance using tools like AIDA64 or Prime95.
If possible, perform tests with a known working GPU, as the symptoms point toward a GPU problem (likely due to the artifacts observed).
Please note the age of your PSU and whether it was new or used/refurbished.
Provide complete system specifications if you can.
You’re missing details such as RAM, CPU cooler, and storage capacity. Full system specs are essential. For example, if you have an RTX 3070 Ti, confirm whether it’s an F or AIB model and the exact version.
As your PC gets older, components degrade. Performance drops and temperatures rise compared to when everything was new.
Use HWinfo64 to monitor temperatures, voltages, and frequencies accurately.
Link: https://www.hwinfo.com/download/
To stress-test the GPU, run Unigine Superposition in one of the presets (e.g., 1080p Extreme).
Link: https://benchmark.unigine.com/superposition
Keep an eye on your monitor or use HWinfo64 logging to observe changes in temps and other metrics during the test.
If the benchmark finishes without issues, you can share your results to compare your GPU with others of the same model.
For RAM testing, download memtest86 and run it (link: https://www.memtest86.com/).
Installation guide: https://www.memtest86.com/tech_creating-window.html
A minimum of one full pass (15 tests) is required. Two full passes are recommended, while four full passes are acceptable.
Because the process can take time, it’s best to run it overnight.
One 8GB RAM typically takes about an hour per full pass; four full passes would take roughly 4 hours.
Two 8GB RAMs take around 2.5 hours per pass, totaling about 10 hours for four passes.
Two 16GB RAMs require about 5 hours per pass, which is approximately 20 hours for four full passes.
Four 32GB RAMs would need 32 full passes (test #7), which most people won’t attempt due to the long duration, especially with ample RAM available.
For thorough RAM testing, 32 full passes are necessary.
Further reading: https://www.memtest86.com/tech_individua...descr.html
However, it’s unlikely anyone will run 32 full passes because it would take too long and generate excessive load.
You can check your exact specifications by looking at the motherboard manual or checking the system information on your computer. If you're unsure, you might need to consult online resources or contact the manufacturer for assistance.
Have them memorized?
🤔
E.g i have 3x desktop PCs and when i bought individual items, i put it all down in the list. Then again, i have all components memorized.
But i digress.
Is your PC prebuilt or DIY?
HWinfo64 Full mode tells what you have.
CPU cooler wise, well, i guess you have PC case with transparent side panel? If not, open up the PC and take a pic of the innards. Afterwards, upload pics to the net, e.g
www.imgur.com
and share them here.
From the pics, we can also see and identify GPU better. Especially when you take a pic where GPU fans are seen.
Same with PC case and fans. If you take pics of the fans. Now, most fans have label at the back, on the hub. But to see that, one needs to unscrew the fan, since back end of the fan is usually blocked by the PC case. Here, i don't think fan exact model is needed. Knowing fan size (120/140mm), their location and orientation will do. This would tell us if you have proper airflow or do you have fans blowing into each other.
E.g here's the pic of my main build, with side panel off. Easy to identify components in it, including airflow path:
Spoiler:
Click here to view
Skylake build. Full specs with more pics in my sig.
Now, PSU age isn't listed anywhere, and for that, you need to tell us when you bought it and was it brand new or used/refurbished. Since else-ways, PSU age is considered maximum it can be, which is 8 years old (HX-series came out in 2018). Also, do you have regular HX-series or HXi-series? To confirm that, you need to look at the PSU's label.
Oh no, new to the forums and unsure about what to do, thank you for the warning! Do you have any suggestions or tools you could use to discover more detailed information such as the exact GPU model? I'm not sure where to start or how. I'll update the main post with the revised specifications.
I haven't paid much attention to it before, but I've been seeing these high temperatures since I got RDR2 for PC on release.
I also updated my post with more detailed specifications and some images to cover some details or model numbers I'm not sure about anymore.
Chassis details were unclear, sorry. Capture images and share them on a platform like Imgur, then forward the link here. If details are scarce, our help options will be limited. I noticed links to Imgur in your first message. It seems you have a Corsair Crystal 570x chassis. The CPU cooler appears reversed, facing backward—should be expelling air from the back of the case, near the left side of the photo. It’s also a Cryorig H7 Quad Lumi. I own one on my shelf, but I’m concerned about its ability to cool an i9-9900K. If you were me, I’d disassemble the whole setup and thoroughly clean it before reassembling it breadboard-style to see if the problem remains. You mentioned having an HX1000 power supply; what color are the stickers on its side?