System reboots under load

[TheBeanMan]

Hi all,

In case you haven't read the title, I've recently been getting random reboots under load. It seems to happen more quickly in CPU heavy software, but it will just happen when the PC is under any kind of load.

The first thing I thought of was overheating, but it doesn't even go above 80°C before rebooting. I thought maybe it was a problem with the CPU but I've tried using a different CPU and the same problem occurs. There's no overclock or anything on the CPU.

I've done memtest86 and it has passed so I don't think it's a memory issue.

I've stress tested the GPU and had no crashes.

My only other thoughts is that it's either an issue with the power supply or the motherboard, but I don't have any of those spare and I don't won't to buy a new one just to find out it's not the problem.

I have reinstalled all of my drivers and updated my bios and chipset, which hasn't fixed the problem.

I have noticed that when it crashes I get an error in event viewer from WHEA-Logger that says "A fatal hardware error has occurred" with event id 18 (screenshot attached).

I've tried researching this but all I can seem to find is that it probably means a faulty CPU, but it's highly unlikely that both CPUs that I've tried would be faulty.

Does anyone have any ideas what the problem could be or any suggestions on what I can do to fix it?


Specs:
AMD Ryzen 7 5800X
EVGA NVIDIA RTX 3080 Ti
Corsair Vengeance DDR4 2x8GB RAM
MSI MAG X570S Tomahawk Max Motherboard
Corsair RM850e Gold 850W PSU
be quiet! Pure Loop 240mm AIO
Samsung 970 EVO 250GB M.2 SSD
Seagate Barracuda 4TB HDD

Windows 10 Pro 64-bit 21H2

[Cyberat]

What is your ram speed set at ?

[TheBeanMan]

4 hours ago, Cyberat said:

What is your ram speed set at ?

In the bios it seems to be set at 2133mhz, but I have 3600mhz ram. Should this be higher? It was just the default that the bios set.

[vertigo220]

Doesn't seem like PSU since you're actually getting an error instead of the PC just dying. Not a perfect test, but your GPU draws over twice the power of your CPU, so you can run a GPU-only stress test and a CPU-only stress test, and if it only crashes on the CPU one, very unlikely to be power related. Could be a bad motherboard which is looking like a CPU issue since the motherboard is the main interface for the CPU. Troubleshooting to determine that is much harder, though, and generally requires a swap to see. It could be RAM as well, but that can be tested on its own with memtest or even something like Prime95 blended and non-blended to see if it happens when stressing the RAM more but not when the RAM isn't stressed as much.

[TheBeanMan]

1 hour ago, TheBeanMan said:

In the bios it seems to be set at 2133mhz, but I have 3600mhz ram. Should this be higher? It was just the default that the bios set.

Setting my ram to 3600mhz seems to have fixed the original problem, but now I have another problem. My PC is incredibly slow and sluggish now. In games where I'd get 60+ FPS I now only get 20 if I'm lucky. Even opening the browser to write this was very slow.

[TheBeanMan]

1 hour ago, vertigo220 said:

Doesn't seem like PSU since you're actually getting an error instead of the PC just dying. Not a perfect test, but your GPU draws over twice the power of your CPU, so you can run a GPU-only stress test and a CPU-only stress test, and if it only crashes on the CPU one, very unlikely to be power related. Could be a bad motherboard which is looking like a CPU issue since the motherboard is the main interface for the CPU. Troubleshooting to determine that is much harder, though, and generally requires a swap to see. It could be RAM as well, but that can be tested on its own with memtest or even something like Prime95 blended and non-blended to see if it happens when stressing the RAM more but not when the RAM isn't stressed as much.

I've already tested GPU, CPU and RAM and I only get crashes on the CPU stress test. Does that mean its a motherboard problem? Is there anyway to check this without swapping it? I'd have to buy a new motherboard to see if swapping it fixes the issue but I don't really want to spend the money if it doesn't fix it.

[vertigo220]

Hard to say, but probably motherboard. I don't know of any way to test it without swapping it, unfortunately. Depending on the return policy of where you buy the replacement, if that doesn't fix it you may be able to return it. Does the motherboard have any sort of onboard diagnostics? Like a small LED light or screen that shows codes.

[TheBeanMan]

17 minutes ago, vertigo220 said:

Hard to say, but probably motherboard. I don't know of any way to test it without swapping it, unfortunately. Depending on the return policy of where you buy the replacement, if that doesn't fix it you may be able to return it. Does the motherboard have any sort of onboard diagnostics? Like a small LED light or screen that shows codes.

There are 4 LEDs (boot, cpu, vga & dram), that's all I've got in terms of onboard diagnostics.

[vertigo220]

So they don't flash in patterns to indicate specific issues? I assume none of them are lit or blinking at any point?

[TheBeanMan]

1 hour ago, vertigo220 said:

So they don't flash in patterns to indicate specific issues? I assume none of them are lit or blinking at any point?

When it boots the cpu is red for a while, then the vga is white for a while, and then the boot is yellow when it posts. The manual for the motherboard seems to just indicate that it should be on if there is an error so I think them flashing like this is part of the usual boot cycle.

[vertigo220]

Yeah, nothing like vague lights accompanied by vague instructions.

[TheBeanMan]

2 hours ago, TheBeanMan said:

Setting my ram to 3600mhz seems to have fixed the original problem, but now I have another problem. My PC is incredibly slow and sluggish now. In games where I'd get 60+ FPS I now only get 20 if I'm lucky. Even opening the browser to write this was very slow.

Scratch this. I fixed the slowness by resetting my bios settings (not sure how that fixed it but that's the joy of technology I guess), but now the PC still crashes no matter what the ram speed is set to. I'm pretty much back to where I started now.

[Cyberat]

Your CPU threshold for ram speed is 3200mhz, + 400 mhz will not gain you much more performance but INSTABILITY.

You want to optimize bios, windows & drivers, then your in-game FPS and focus on Stability & Compatibility.

Set Ram speed to 3200mhz, set CL/CAS timings to Auto.

Set PCIE to 3.0 if you have the option.

Shut off Green, Eco, Sleep aka C-States. Including the CPU, if it allows activate all cores manually, do not leave on Auto.

I have a much older chip, turning off HPET did great for reducing latency, you can test it out though. It works well on AMD CPUs.

Then - Windows turn off Sleep, Hibernate, Indexing, Search (service) & Pagefile.

HDD replace with SSD ASAP, the age of mechanical drives left the building more than 10yrs. ago. You can slap it into an enclosure and call it a backup drive, only on when you do your image backups or such. As for SSDs - 4TB Samsung 870 Evo ~$300.00. m.2 is not really faster than SATA 3 yet.

Drivers, depending on monitor capability but turn on Fast Sync instead of V-Sync, makes sure it's turned off in games. Turn on G-Sync if monitor has it.

Always optimize in-game settings, there are guides on the internet about it all, Ultra does not mean you'll like what you see, nor your system perform it best. Some "features" rob performance greatly and deliver very little. My pet peeves are motion blur, godrays, badly implemented depth of field (always blurry too close to view). I think human eyes create distance blur anyways, if the PC is capable of distant sharpness.

[TheBeanMan]

Update: My PC now doesn't boot at all and my motherboard shows a CPU error led. I highly doubt that the CPU is actually the problem however, as these problems I've been having would happen regardless of what CPU I tried. If I had to guess I'd say my motherboard is faulty. I guess I'm going to have to get a new motherboard and hope that fixes the problem.

[Mark Kaine]

4 hours ago, TheBeanMan said:

Update: My PC now doesn't boot at all and my motherboard shows a CPU error led. I highly doubt that the CPU is actually the problem however, as these problems I've been having would happen regardless of what CPU I tried. If I had to guess I'd say my motherboard is faulty. I guess I'm going to have to get a new motherboard and hope that fixes the problem.

just a question,  did you update the BIOS during this whole ordeal? 

 

also did the pc work ok at some point or did it always have these crashing issues? 

 

4 hours ago, TheBeanMan said:

had to guess I'd say my motherboard is faulty. I guess I'm going to have to get a new motherboard and hope that fixes the problem

its either motherboard or psu , yeah.

 

(and btw a bad psu can definitely produce some errors before crashing although not necessarily) 

[TheBeanMan]

On 12/27/2022 at 2:28 AM, Mark Kaine said:

just a question,  did you update the BIOS during this whole ordeal? 

The only time I updated the bios was when I was initially having problems. I didn't change anything when it stopped booting, it just seemed to happen on its own.

On 12/27/2022 at 2:28 AM, Mark Kaine said:

also did the pc work ok at some point or did it always have these crashing issues? 

It used to have crashing issues but very infrequently. Recently it just crashed so often that it was almost unusable for anything other than light workloads.

[TheBeanMan]

Update:

I have replaced the motherboard and it seems to have fixed all of the problems I was facing, looks like it was a faulty motherboard after all.

Thanks to all who posted suggestions and advice, it is greatly appreciated.