Random pc restarts when playing games

[iazertyuiopi]

Hi guys,

I have recently completed my new PC build - parts here : https://pcpartpicker.com/user/AZERTYUIOP/saved/TzzvZL

However I am encountering random crashes when playing - it seems that the more demanding the game, the faster it crashes (hard crash - pc reboot - no blue screens or anything).

On rocket league/ZZZ I can play for a few hours, but on Expedition 33 it's borderline unplayable. PC did crash once with only youtube/discord I think.

Symptoms :

• sometimes it reboots silently and instantly, sometimes I can hear last sound repeated for a few secs before reboot
• at one occasion I had a black screen but could still hear discord for a few seconds before reboot

Steps taken so far

• update BIOS to latest version
• reinstall windows 10 from scratch
• change PCI cables for graphic card
• ran OCCT tests, I do get errors sometimes quickly during the 'power' test

  https://imgur.com/a/sCwPOG1

  - corresponding hwinfo logs are here : https://files.catbox.moe/hmg893.CSV - power test can also trigger the reboot if ran for one hour but not 100%

It's very frustrating. I could try to buy random parts to replace them but I would like to be able to pinpoint the issue first, so would really appreciate if someone could guide me in the right direction to find if it's a hardware issue and which part is faulty. If needed I could try to reproduce with hwinfo logging enabled. Right now I am trying to assess if behavior is impacted by hour of day.

thanks !

 

[eidairaman1]

Ok so it reboots, i suggest you disable automatic restarts when windows crashes and get bluescreenview

Also unplug the argb/drgb connector from the motherboard for that 3pack fanteks fan

 

[iazertyuiopi]

Thanks for the reply, appreciate it.

I will try the suggestions and report back on Wednesday

 

[Bill_Bright]

the more demanding the game, the faster it crashes

That is a classic symptom of excessive heat in one or more sensitive component. That case supports a wide variety of fans and cooling options in a wide variety of configuration options. You want a good "flow" of cool air through the case. Sadly, especially with many fans in use, they may be configured in a way that causes excessive turbulence that disrupts that "flow", creates pockets of stagnant (and heated) air, and is counterproductive to the most efficient "flow". So make sure you are providing a good flow through the case. Consider opening the side panel and blasting a desk fan in there (for testing purposes) to see what happens. Remember heat sensitive devices include the VRMs, RAM modules, chipset modules and more - that is, not just the CPU or graphics.

This is also a classic symptom of power issues. Even the best models from the best makers can have a defect. You might want to consider temporarily swapping in another capable, known good supply to see what happens.

Also, every home and every computer user should have access to a AC Outlet Tester to ensure the wall outlet is properly wired and grounded to Earth ground. I recommend one with a GFCI (ground fault circuit interrupt) indicator as it can be used to test bathroom and kitchen outlets (outlets near water) too. These testers can be found for your type and voltage outlet, foreign or domestic, (like this one for the UK, or this one for German outlets) at most home improvement stores, or even the electrical department at Wal-Mart. Use it to test all the outlets in the home and if a fault is shown, have it fixed by a qualified electrician.

Try running with 1 stick of RAM. If no problems, swap in 2nd stick.

If none of the above works, I would remove everything out of the case and assemble the computer on a large, unfinished, wooden bread/cutting board and see if it remains stable there.

While everything is out, triple check to make sure there were no foreign objects under the motherboard and that there is a case standoff ONLY where there is a corresponding motherboard mounting hole.

When reassembling inside the case, double and triple check to make sure every power and data cable are securely connected and that each RAM stick and expansion card is fully and securely seated in its socket/slot.

 

[Hyderz]

New build! Could be plastic film between thermal paste, cpu and cpu cooler… overheating causing system to restart

 

[blahoink01]

I see the OCCT test. I'd get to basics and run memtest outright. Every adjustment I make in my BIOS, every new part, or re-paste gets a memtest run for several hours. If I get one error, I know it was done incorrectly. That won't capture everything. If it comes back clear, then I run Prime95 for a few hours and usually call it. If the reboots come back, I run Prime95 for a full day to see if it passes.

 

[iazertyuiopi]

That is a classic symptom of excessive heat in one or more sensitive component. That case supports a wide variety of fans and cooling options in a wide variety of configuration options. You want a good "flow" of cool air through the case. Sadly, especially with many fans in use, they may be configured in a way that causes excessive turbulence that disrupts that "flow", creates pockets of stagnant (and heated) air, and is counterproductive to the most efficient "flow". So make sure you are providing a good flow through the case. Consider opening the side panel and blasting a desk fan in there (for testing purposes) to see what happens. Remember heat sensitive devices include the VRMs, RAM modules, chipset modules and more - that is, not just the CPU or graphics.

This is also a classic symptom of power issues. Even the best models from the best makers can have a defect. You might want to consider temporarily swapping in another capable, known good supply to see what happens.

Also, every home and every computer user should have access to a AC Outlet Tester to ensure the wall outlet is properly wired and grounded to Earth ground. I recommend one with a GFCI (ground fault circuit interrupt) indicator as it can be used to test bathroom and kitchen outlets (outlets near water) too. These testers can be found for your type and voltage outlet, foreign or domestic, (like this one for the UK, or this one for German outlets) at most home improvement stores, or even the electrical department at Wal-Mart. Use it to test all the outlets in the home and if a fault is shown, have it fixed by a qualified electrician.

Try running with 1 stick of RAM. If no problems, swap in 2nd stick.

If none of the above works, I would remove everything out of the case and assemble the computer on a large, unfinished, wooden bread/cutting board and see if it remains stable there.

While everything is out, triple check to make sure there were no foreign objects under the motherboard and that there is a case standoff ONLY where there is a corresponding motherboard mounting hole.

When reassembling inside the case, double and triple check to make sure every power and data cable are securely connected and that each RAM stick and expansion card is fully and securely seated in its socket/slot.

Thanks sir. I don't believe it's a heat issue as I've got hwinfo64 logs (polling sensors every 100ms) of my last game session leading to a crash and I can't see anything.

All suggestions above are solid though, will definitely test and revert back.

New build! Could be plastic film between thermal paste, cpu and cpu cooler… overheating causing system to restart

Haha I actually did this on first launch, did spot the issue quickly though^^'

I see the OCCT test. I'd get to basics and run memtest outright. Every adjustment I make in my BIOS, every new part, or re-paste gets a memtest run for several hours. If I get one error, I know it was done incorrectly. That won't capture everything. If it comes back clear, then I run Prime95 for a few hours and usually call it. If the reboots come back, I run Prime95 for a full day to see if it passes.

Thanks !! Just to be sure, are we talking memtest86 or 64?

 

[blahoink01]

Thanks sir. I don't believe it's a heat issue as I've got hwinfo64 logs (polling sensors every 100ms) of my last game session leading to a crash and I can't see anything.

All suggestions above are solid though, will definitely test and revert back.


Haha I actually did this on first launch, did spot the issue quickly though^^'


Thanks !! Just to be sure, are we talking memtest86 or 64?

memtest org, I use the windows 64 bit installer to a usb and it takes care of the rest.

edit: Sorry, I should've posted TPU's link. https://www.techpowerup.com/memtest64/

 

[iazertyuiopi]

As a small update, unplugging the rgb cable and running with only one RAM stick didn't improve the situation - ran memtest64 for a few hours without issue. I will be doing the prime95 testing too overnight.

Next move will be to run with same PSU and another graphics card (but I only have access to a 3070 that draws less power than the 9070xt) and then same graphics card and another PSU (which I would need to acquire).
I have also ordered this thing : https://www.amazon.fr/dp/B0CQYQ4NC4?smid=A938M5W8T2EXA&ref_=chk_typ_imgToDp&th=1 that looked to be a AC outlet tester.

For completeness I also disabled 'automatic reboot on system failure' in windows and was able to get my longest gaming session in a while but I don't think that's related.

 

[Bill_Bright]

I use and recommend PassMark’s MemTest86. It is regularly updated with its most recent version being V 11.3 from 2/May/2025. Allow the diagnostics to run for several passes or even overnight. You should have no reported errors – not even one.

Note, however, while software-based RAM diagnostic tools are good, none are 100% conclusive. If they report any errors, even one, the RAM is bad. However, while rare, on occasion they will report no problems, yet the RAM still fails in use, and/or when paired with other RAM. So, swapping in all new RAM is often the best test.

To conclusively test your RAM, you need to use sophisticated and very expensive test equipment, like this $2,495 Memory Tester (and that's for the cheap model)! So it is usually easier (and cheaper!) to swap in known good RAM and see what happens.

Yes, that outlook checker will work.

 

[Waldorf]

@iazertyuiopi
unless you are running stock settings for ram, and expect it to be defective, memtest wont do much.
test with TM5 or HCI, both will find ANY error, no matter if from tuning or defective hw.
love how ppl keep recommending something that will only catch problems with defective sticks, vs recommending those that do all.

outside that, is suspect the psu, even if its just the combination of parts, had more than one build where it just wouldnt run properly, swapping the psu with a friend fixed it,
while his system ran fine with the same psu not working on mine.

 

[GodisanAtheist]

Do you have your computer plugged into some ancient or cheap surge protection strip? Try plugging it directly into a wall outlet.

My main PC in my signature randomly started experiencing hard shutdowns (computer literally just turns off no warning) a few weeks ago when playing Shadow of War (which runs at like 200+ FPS). It was really perplexing, because the system was rock solid outside of that scenario.

Realized that I had the computer plugged into some 10+ year old powerstrip that likely isn't rated for the kind of wattage it was pulling in this specific scenario, plugged the PC directly into the wall outlet and the random crashing completely stopped. Was able to play the last 2-3 hours of the DLC for the game completely stable and shutdown free despite the computer shutting down closer to 30-45 minutes previously.

Worth a shot.

 

[Waldorf]

True, forgot about that since all stuff is on UPS with AVR, even screen

 

[Carlyle2020hs]

That CSV file ... is not fit for this purpose, since i´ve never seen anybody take the 30 minutes of time needed to make sense of one.

Kindly provide:

- a full HWinfo screenshot . - yeah i know its only good for idle values. But at least you get people to look at it. Which nobody will with the CSV file, i assume.

- a windows event viewer screenshot around a crash

fyi
It is known that those cards have hard-to-see foam in the fins that many forget to remove.

Suggestion:
Nobody stops you from downclocking the GPU to 25% to see if its a wonky psu.

Good luck!

 

[Bill_Bright]

love how ppl keep recommending something that will only catch problems with defective sticks, vs recommending those that do all.

Contrary to what you believe, Waldorf, neither of those are infallible either. No software based program is.

You might want to read this: https://www.tomshardware.com/how-to/how-to-test-ram as it talks about HCI, TM5 and Passmarks.

 

[iazertyuiopi]

That CSV file ... is not fit for this purpose, since i´ve never seen anybody take the 30 minutes of time needed to make sense of one.

Kindly provide:

- a full HWinfo screenshot . - yeah i know its only good for idle values. But at least you get people to look at it. Which nobody will with the CSV file, i assume.

- a windows event viewer screenshot around a crash

fyi
It is known that those cards have hard-to-see foam in the fins that many forget to remove.

Suggestion:
Nobody stops you from downclocking the GPU to 25% to see if its a wonky psu.

Good luck!

I don't see anything in event viewer

Regarding the csv file I found this useful tool https://www.hwinfo.com/forum/threads/logviewer-for-hwinfo-is-available.802/ that lets me see lots of graphs.
Not sure what you mean by hwinfo screenshot? My temps are fine (including under cinebench/furmark/etc.)
I will try running with GPU power limit lowered to the maximum amd software lets me set :

Do you have your computer plugged into some ancient or cheap surge protection strip? Try plugging it directly into a wall outlet.

My main PC in my signature randomly started experiencing hard shutdowns (computer literally just turns off no warning) a few weeks ago when playing Shadow of War (which runs at like 200+ FPS). It was really perplexing, because the system was rock solid outside of that scenario.

Realized that I had the computer plugged into some 10+ year old powerstrip that likely isn't rated for the kind of wattage it was pulling in this specific scenario, plugged the PC directly into the wall outlet and the random crashing completely stopped. Was able to play the last 2-3 hours of the DLC for the game completely stable and shutdown free despite the computer shutting down closer to 30-45 minutes previously.

Worth a shot.

I'm on a power strip, worth a shot indeed. Will try, didn't think of that.

 

[iazertyuiopi]

Small update :
- memtest86 is passing (left running during the day).
- setting power limit to -30% appears to have lengthened my gaming sessions (would need more tests to confirm) but crashes still occur on several games (rocket league and clair obscur at least)
- the outlet tester says 'missing ground' which of course is true, also most outlets with ground in my appartment seem to have neutral and live swapped (?). There's one green outlet in the kitchen but can't plug the pc on it ^^'
- I am using a new power strip now just in case it helps.

Next step will be to switch back to my old 3070 and see if crashes happen.

 

[Bill_Bright]

- the outlet tester says 'missing ground' which of course is true, also most outlets with ground in my appartment seem to have neutral and live swapped (?).

Not sure why you say "of course is true" for a missing ground. You should not have any missing grounds.

Sadly, it is not uncommon for the remaining pair to be swapped when only two wires are connected. I have seen where the ground wire is some times left off BECAUSE the ignorant/untrained/amateur person who did the wiring does not know what he or she is doing. Or worse, the owner is trying cut corners and costs since 3-wires cost more than 2. Connecting the ground wire when those are reversed often trips the circuit breaker so their "fix" is to leave them reversed. A proper certified electrician would never do that, nor would such wiring pass code inspection.

Note that is totally illegal in many jurisdictions, including all the United States for ALL buildings built or remodeled after 1975. This is because it can lead to someone being electrocuted. Not good. I don't know where you live but you should contact your landlord or apartment manager and inform them the wiring is wrong. If me, I would get with your neighbors, test their outlets, and if also wired wrong, get them to file a complaint too. The more complaints the more likely the landlord will do something about it or he/she risks bigger issues with the government. But again, that depends on where you live, your public safety laws, and how they are enforced.

Having said all that, two-wire outlets "should" not be causing your rebooting issues AS LONG AS those wires are securely connected all the way back to the service panel AND other devices on those circuits are not interjecting a bunch of noise and other anomalies on the line.

 

[iazertyuiopi]

I am reproducing errors with another GPU.

So it would be either motherboard or PSU as same GPU is running fine on my electric network in another machine.
I will stay with smaller GPU a bit until a crash happens, and then swap back and purchase a new PSU I guess

 

[Bill_Bright]

I definitely think you need to try another PSU, but it would be good if you could borrow/swap a known good PSU from another computer to try first, before spending money on another.

I would also consider temporarily moving the computer to another wall outlet, one you've verified is wired correctly.

 

[Shrek]

- the outlet tester says 'missing ground' which of course is true, also most outlets with ground in my appartment seem to have neutral and live swapped (?). There's one green outlet in the kitchen but can't plug the pc on it ^^'

You really want to get that fixed.

 

[iazertyuiopi]

Despite the errors, I didn't have a crash in 3 days with the 6800, and the 9070 xt reboots another PC (all different parts, 750W gold PSU) within 15min of power test
So either it's the graphic card, or it's drawing too much power for my electrical network

 

[Bill_Bright]

or it's drawing too much power for my electrical network

Not likely. If the circuit was overloaded, it would trip the breaker or blow a fuse, and everything on that circuit would go out. You then would have to manually reset the circuit or replace the fuse - not something people do in their sleep.

 

[iazertyuiopi]

Put back the 9070 XT -> two reboots in one day when the 6800 ran clair obscur on max settings for a whole day just fine, I will start the RMA process
Thanks again for all the help everyone !

 

[Waldorf]

@iazertyuiopi
get an electrician (at least i "told" you ) to install a AFCI, then you dont need to worry about ground missing.