Question System unstable, but ok when gaming ?

[natcha12]

Have a bit of a weird one guys,

It started when I was still on Windows 10 last year with an i7-6700 CPU, with random restarts at intervals of between 5-30 min after booting up.

I thought what the heck, I'll upgrade to Windows 11 to see if it helps because of the upcoming security deadline anyway but it didn't help at all.

Then I tried a clean install, nada. Upgraded to a Ryzen 5800X with same ram sticks, nothing.
Two more clean installs later, turned on dummy power loading in BIOS as a shot in the dark too, still nothing.
I tried out Ryzen Master at some point but I wasn't making much use of the gains so reset everything.
The current OS install has not touched RM and has had CMOS cleared beforehand too.

All throughout this though, it absolutely never happened while gaming. I'm talking pc game pass, steam, ubisoft. Anything remotely taxing on my system seems to stabilise it...

I'm currently setting up a memory scan. I'm at a loss what to do next. I've left Space Engineers open in the background while my PC has been re-downloading my game library for 3 days and it's totally normal.

Please help me check my emails and the news in peace!

 

[Lutfij]

When posting a thread of troubleshooting nature, it's customary to include your full system's specs. Please list the specs to your build like so:
CPU:
CPU cooler:
Motherboard:
Ram:
SSD/HDD:
GPU:
PSU:
Chassis:
OS:
Monitor:
include the age of the PSU apart from it's make and model. BIOS version for your motherboard at this moment of time.

Considering you performed an upgrade, we'll be seeing two sets of specs, one for before and one for after.

Where did you source the installer for the OS?

I wasn't making much use of the gains and reset everything
You're advised to reinstall the OS but in offline mode, manually installing all necessary drivers in an elevated command while in offline mode.

 

[natcha12]

Where did you source the installer for the OS?

I wasn't making much use of the gains and reset everything
You're advised to reinstall the OS but in offline mode, manually installing all necessary drivers in an elevated command while in offline mode.

I got the official win11 64 from the microsoft site. Not a recovery version, and not linked/customised for the current PC. I wanted nothing to do with my old setup software wise.

You have taken my sentence out of it's context of Ryzenmaster... I reset settings in the program itself, uninstalled it then cleared CMOS.

As requested
CPU: 5800x stock now (second hand), from i7 6700
CPU cooler: Noctua NH D15 dual 140mm, old cooler was some basic intel I don't have anymore
Motherboard: x570 Aorus, from B250N phoenix (both second hand)
Ram: Crucial 2x 8Gb
SSD/HDD: OS drive, WD blue SATA 500Gb
GPU: Zotac twin edge 4070
PSU: Either aerocool 750 or higher/corsair cx 750w. up to 8 years old
Chassis: Phanteks enthoo pro 2 server edition, 6x 140mm fan
OS: win 11 pro
Monitor: Samsung Neo QN90A tv

I have doubts it would be a driver issue. I can't easily and wouldn't want to try and repeat to confirm it, but on one of the total resets I do believe it restarted partway through Windows installation and subsequently failed. Wouldn't this mean it happens before any drivers were even installed?

Memtest86 came back clear after a default run earlier.
chkdsk on OS drive and sfc scans have found no errors, multiple runs.

 

[natcha12]

I'm testing with another PSU this week.
Things seem to point to the low draw mode being unstable with the one I have instead of any software issue. Windows power profile on high performance has no effect.
It's gotten worse, older games don't seem to be enough to stop the restarts and it's happening sooner than before, like 5-10min instead of 20+. I don't want to run the furmark bench test just in order to use my pc haha

 

[Misgar]

Ram: Crucial 2x 8Gb

Have you tried running your RAM at stock JEDEC speed (usually 2133 or 2400MT/s)?

If you're running a high XMP overclock at more than the rated 3200MT/s for the 5800X, this could cause instability.
https://www.techpowerup.com/cpu-specs/ryzen-7-5800x.c2362

A single pass on MemTest86 is a good sign, but an extended test overnight with multiple passes would be better.

If you are running a high XMP setting, try reducing the speed to 3000MT/s, run some games, run MemTest86.

If stable, increase to 3200MT/s and repeat tests.

Even one error in MemTest86, reduce XMP by 200MT/s and test again. Repeat speed reduction until zero errors.

SSD/HDD: OS drive, WD blue SATA 500Gb

If none of the above work, try changing the boot drive to a different SSD. WD Blues are usually reliable, but another SATA SSD would rule out a problem with the WD.

Best bet for SATA would be a Samsung 870 EVO or similar drive from Crucial with DRAM cache.
https://www.amazon.com/Samsung-Factor-Intelligent-Magician-Software/dp/B08PC43D78

If you don't feel like spending too much, I've used numerous cheap Patriot SATA SSDs as Windows boot drives in unimportant old systems.

You could probably pick up a 120GB drive for around $13 and experiment. OK for a very basic Windows test installation and cheap. I don't use even cheaper brands, but Pny and Integral are OK. Avoid Lexar.
https://www.amazon.com/Patriot-Memory-P220-Internal-Solid/dp/B0BS9W3T48/ref=sr_1_5

Motherboard: x570 Aorus, from B250N phoenix (both second hand)

I've had pretty good luck with second hand motherboards, CPUs, GPUs, NICs, RAM, etc. on eBay, but I have had the occasionlal dud. One old mobo I bought for a friend failed after 6 months, but it was AMD AM3 running a Phenom II X4 955, so it was quite old.

The problem with any second hand part is you don't know if the previous owner observed full ESD handling procedures, or zapped it with static electricity and caused damage.

When I receive a board, CPU or RAM wrapped in nasty static-inducing bubblewrap I groan inwardly, but if it cost only a few dollars, I accept it as "normal". I'm much happier it the parts arrive in ESD bags. It shows a professional approach on the part of the seller.

The labs where I've worked spent millions on anti-static everything - ESD floor tiles, benches, shoes, coats, wrist straps, etc.

If any of your second hand parts were damaged by static, it can take years for the fault to appear.

 

[natcha12]

Thanks for all the info.
Yes, I did fault finding with the ram using just a single pass. It was my first time using memtest so left everything on default.
I did check other forums and saw that after 3200 things get.. unruly. My ram is rated at 2133 stock (2400xmp) I think, I had an "unstable but passed the ryzenmaster bench" speed up to 29xx so backtracked to 2750/2800 After the first CMOS reset I just left at stock though.
I didn't think about ESD and OS drive issue... Maybe I'm just too trusting for better or worse. I don't have spare SSD and money a bit tight so I'll probably hook up a usb HDD as a boot if possible. But only after testing the PSU as I'm still convinced that's suspect number 1

 

[natcha12]

Unfortunately the other PSU (very dusty silverstone SFX) either has the exact issue 😆 or its something else entirely.

I also now know my current PSU is a Aerocool VP pro 850w so there's that.

I'm setting up memtest tonight. It can take all day tomorrow too for all I care if it can find something wrong!

I have 0 known good leftover RAM, so hopefully worst case scenario I just leave the bad stick out and carry on with 8gb fml

 

[natcha12]

So memtest running the max free 4 passes came up with nothing. I'm running it a second time now with sequential CPU mode as that sounds more concise.
I also had a thought, it can't really be ESD since it was happening with my old setup.
Hardware wise also, the only things that have been consistent with both setups are the drives themselves and the ram, which yeah I've just tested ok.

Failing a test install on a known good drive to discount those, I can only think it would be a cable issue maybe? I'm beginning to clutch at straws here though.

I'll keep going 🤪

 

[natcha12]

Have you tried running your RAM at stock JEDEC speed (usually 2133 or 2400MT/s)?

If you're running a high XMP overclock at more than the rated 3200MT/s for the 5800X, this could cause instability.
https://www.techpowerup.com/cpu-specs/ryzen-7-5800x.c2362

A single pass on MemTest86 is a good sign, but an extended test overnight with multiple passes would be better.

If you are running a high XMP setting, try reducing the speed to 3000MT/s, run some games, run MemTest86.

If stable, increase to 3200MT/s and repeat tests.

Even one error in MemTest86, reduce XMP by 200MT/s and test again. Repeat speed reduction until zero errors.


If none of the above work, try changing the boot drive to a different SSD. WD Blues are usually reliable, but another SATA SSD would rule out a problem with the WD.

Best bet for SATA would be a Samsung 870 EVO or similar drive from Crucial with DRAM cache.
https://www.amazon.com/Samsung-Factor-Intelligent-Magician-Software/dp/B08PC43D78

If you don't feel like spending too much, I've used numerous cheap Patriot SATA SSDs as Windows boot drives in unimportant old systems.

You could probably pick up a 120GB drive for around $13 and experiment. OK for a very basic Windows test installation and cheap. I don't use even cheaper brands, but Pny and Integral are OK. Avoid Lexar.
https://www.amazon.com/Patriot-Memory-P220-Internal-Solid/dp/B0BS9W3T48/ref=sr_1_5


I've had pretty good luck with second hand motherboards, CPUs, GPUs, NICs, RAM, etc. on eBay, but I have had the occasionlal dud. One old mobo I bought for a friend failed after 6 months, but it was AMD AM3 running a Phenom II X4 955, so it was quite old.

The problem with any second hand part is you don't know if the previous owner observed full ESD handling procedures, or zapped it with static electricity and caused damage.

When I receive a board, CPU or RAM wrapped in nasty static-inducing bubblewrap I groan inwardly, but if it cost only a few dollars, I accept it as "normal". I'm much happier it the parts arrive in ESD bags. It shows a professional approach on the part of the seller.

The labs where I've worked spent millions on anti-static everything - ESD floor tiles, benches, shoes, coats, wrist straps, etc.

If any of your second hand parts were damaged by static, it can take years for the fault to appear.

Okaaaaay. So lots of testing done the last few days.

I physically disconnected my WD blue drive from the board, along with all but 1 other SSD. Then did a (4th..?) completely fresh OS reinstall on the only drive connected. Cue multiple (4-6 separate occasions) restart crashes during the install. I push through and finally complete the install.
Now here's the thing, I then went and removed my second RAM stick from the A2 slot, and the pc was able to idle for over 24 hours absolutely fine while reinstalling my steam collection etc. Bearing in mind this is a stick that has passed 8 memtest passes without failure. Obviously this points to the board.
I then do some quick hardware swaps over the next hour with the RAM and am able to figure out that no matter which of my random RAM sticks I have lying around I have there, both the A2 and B1 slots (the centremost ones) cause the restarts. A1 and B2 seem to be the only viable options:

Unfortunately yes... this does mean I'm now using a 2 stick, 16gb single channel PC for gaming with probably a 10% performance hit. On the other hand, I now have the luxury of writing this massive post in one go without rewriting over and over again!

Thanks again for the drive suggestion that set me down the right path. I'll just put this down to random board slot failure or sneaky sellers

Off to reset my overclocks now, or atleast try to offset the disadvantage a bit.