Help me diagnose these random reboots please :(

Haggisman

Haggisman

Haggisman

Soldato
Joined
6 Oct 2004
Posts
18,729
Location
Birmingham
I've been having this issue for a while, but it seems to have become a lot more often since I changed GPU a few weeks ago.

When I'm gaming, sometimes my PC will just restart, no warning, no errors, it will just reboot. Nothing in the event viewer and other than this issue, everything is working fine.

It seems the more demanding the game the more likely it is to happen.

I've been playing through Horizon Zero Dawn recently, sometimes directly on the PC, sometimes streaming to my Nvidia Shield using the Steam Link app. It happens quite consistently when I'm streaming; after 10-15 minutes, it will reboot, but after that I can play for hours with no issues. As far as I can remember, it's not occurred while playing directly on the PC.

I finished HZD a few days ago, so moving on to Phantom Liberty (directly on the PC this time), and it did exactly the same thing - silent reboot when I got into the game, just after creating a new character. Fired it up again after, and it ran for an hour fine (then I had to go to bed).

I've also had the same issue in a couple of other games; Deep Rock Galactic and Risk of Rain 2 have both done it when quitting a lobby to return to the menu.

My rig is as follows:

5800X3D
Alpenföhn Matterhorn cooler
B550 Tomahawk
Patriot Viper Steel 16GB (2x8GB) DDR4 3600MHz
Sapphire RX 7800XT MBA (this was previously a 6800)
Corsair RMX 750w PSU
Drives:
  • Teamgroup MP34 NVME (256GB) (boot)
  • T-Force Vulcan SSD (1TB)
  • Samsung 860 EVO SSD (1TB)
  • WD Black SN850X NVME (2TB)
Full disclosure, I haven't reinstalled Windows in probably 10 years, so this install has been through several hardware changes, including a change from intel to AMD and a migration from one drive to another, so that probably isn't helping the stability of the system, but I can't bring myself to go through the effort unless I really need to :p

My initial thought was that it's power spikes from the GPU causing it - e.g. in some of the examples above, going back to menus causing a big spike in FPS and a corresponding power draw, especially given the increase in occurrences since I changed GPU. I believe 750w is the minimum recommended for the 7800XT? The PSU is probably the oldest part of the system, just over 4 years old, which isn't particularly ancient, but if it's running at the top of its range then maybe pushing it a bit too hard?

I'm pretty sure it's not an overheating problem, the CPU never goes above the low 80s, even under heavy load, and the GPU maxes out around 90c on the hotspot/junction temperature.

Just wondering where is the best place to start troubleshooting?

Thanks!
 
I would also not rule out the CPU, not usually a problem but lately at work we have had 3 Ryzen CPUs that we have had to RMA.

All of which were exhibiting random crashes, each machine was fixed with replacement CPU. (We tried testing everything else prior to finding it was possibly the CPU).

These were:

2x 3600
1x 5800
 
Webzta said:
Maybe start with a PSU from the rainforest, as it's easy to return if that was not the issue.
Click to expand...

paradigm said:
My gut feeling also says voltages.

Grab HWMonitor or similar and see what your voltage rails are like under load.
Click to expand...

Does it sound like the PSU could be the culprit then?

I've downloaded HWMonitor, but it doesn't look like there's any way to save the results on the go or monitor remotely, so if the PC does reboot due to a voltage drop then it's going to lose the results?

Webzta said:
I would also not rule out the CPU, not usually a problem but lately at work we have had 3 Ryzen CPUs that we have had to RMA.

All of which were exhibiting random crashes, each machine was fixed with replacement CPU. (We tried testing everything else prior to finding it was possibly the CPU).

These were:

2x 3600
1x 5800
Click to expand...

That would be annoying - I can temporarily steal my son's 2700 to test, but I don't think mine will work in his board :(
 
Last edited:
AMD CPUs do that when unstable. You get a lot of random reboots when fine tuning Curve Optimizer

If yours is stock, guess it is running close to edge of voltage stability

For some people tweaking Load Line Calibration helps. My gut says decrease or disable it entirely, but maybe increasing would help some cases
 
Haggisman said:
Does it sound like the PSU could be the culprit then?

I've downloaded HWMonitor, but it doesn't look like there's any way to save the results on the go or monitor remotely, so if the PC does reboot due to a voltage drop then it's going to lose the results?



That would be annoying - I can temporarily steal my son's 2700 to test, but I don't think mine will work in his board :(
Click to expand...

I think it's going to be some trial and error, hopefully you get lucky and identify the component without having to try too many things.
 
I've got both HWInfo64 and WMonitor running (already had HWInfo64), and there's a bit of discrepancy between the 12v readings - HWInfo64 is showing 11.904v, HWMonitor shows 12.072. That's about 1.5% difference?
 
Haggisman said:
LLC *Whoosh* :cry:
Click to expand...
CPU Load Line Calibration is a single setting somewhere in CPU voltages area of bios
It is responsible for how agressively voltages get adjusted between idle and load

usually it has values like disabled/normal/standard, low, medium, high, extreme

depends on motherboard what setting is used by default

It is a single thing to experiment with (and quite safe to change), to see if maybe crashing goes away, then you know it is about CPU voltages.
If LLC is disabled, try Medium. If it was already at some value by default, try lowest or disabled/normal.
 
alec said:
CPU Load Line Calibration is a single setting somewhere in CPU voltages area of bios
It is responsible for how agressively voltages get adjusted between idle and load

usually it has values like disabled/normal/standard, low, medium, high, extreme

depends on motherboard what setting is used by default

It is a single thing to experiment with (and quite safe to change), to see if maybe crashing goes away, then you know it is about CPU voltages.
If LLC is disabled, try Medium. If it was already at some value by default, try lowest or disabled/normal.
Click to expand...

Thanks, I checked, and LLC was set to "Auto" - useful...

I've set it to "default" for now will see if that makes a difference, if not I'll try the middle setting, and then the lowest/highest and see if that helps
 
Bah, thought I'd gotten somewhere, but just had another reboot in CP2077 after playing for an hour or so with no issues :(

Changed LLC to "Mode 5" now, which is the middle setting.
 
Last edited:
I respect all the folks trying to bullseye your issue in here, not much changes over the years, but take it from me, I've been fixing people's pcs, for money, for 2 decades. First step is buy a hard drive, cheap if needed or an upgrade. Put clean windows on it (can keep current one intact) use clean windows for a few days, see what shakes loose. 90%+ reboots or crashes are installation or disk issues. Even if no obvious evidence. You MUST eliminate those first or you are looking for a needle in a stack of needles. If you get crashes on vanilla install with new storage, it's psu, then gpu,then ram then mobo then cpu in that order to test one by one.

I appreciate this isn't a convenient solution compared to randomly changing bios settings, but thats a bit of a misguided path that forums send you down, they have ever since ever certainly in all my years online. The fact is that of the several thousand repairs I've done over the years, across platforms and across configurations, its basically never an obscure bios setting. It's generally windows/disk or power/gpu (for this specific type of issue)

Doesn't mean it always is. But you are always best hedging bets, the headache is rarely cancer, the itch is rarely the clap, and the guy who hangs out with yer girl is rarely on the level

Hedge your bets
 
Last edited:
Cheers, that was a thought that crossed my mind, 500gb nvmes are cheap as chips these days and my 250gb boot drive could use a bit more space, it's just the effort and inconvenience I'm trying to avoid :p
 
When the pc reboots, is there any visual corruption just before it goes. Or is it a case of the screen goes black and the pc restarts?

I would say its either cpu or perhaps ram stability at play, i've had my fair share of problems in CP2077, Horizon Zero Dawn and Starfeild, when all of those games load up my cpu temp does spike up very high for around 30 seconds, sometimes i get a power off and to get my pc to power up i have to unplug the pc from the wall, drain power and then boot back up.

I have my 7950x at stock settings but am running ddr5 at 6200mhz, overclocking my ram has caused all kinds of weird stability problems and after many hours of testing timings and voltages i managed to crack it and all my games are working with no problems.

You could set your ram back to default speeds to check if game stability improves, if so you may need to test voltages and timings when you enable xmp/expo/docp, perhaps something isn't quite right, and when under heavy load it drops out, causing the reboots
 
You must log in or register to reply here.
Back
Top Bottom