ryzen-based system rebooting during gaming (cache hierarchy)

[difelim644]

Hi, for a few months (~3) i've had a very stressful issue where my PC will just reboot during games. This can be anywhere from 1 second - 5 hrs into a game. The system is completely stable when not playing games and I usually leave it on overnight, not once has it crashed.
Synthetic stress tests cannot replicate this issue: I can run p95 for days on each setting, memtest86, occt, furmark.

SPECS: r7 3700x, rog strix 5700xt, prime x470, 32gb 2300 ram, cm1000x (1000w) psu, kingston a2000 nvme.

The system doesn't create logs all the time, when it does, only a log in event viewer is visible. this shows
"
A fatal hardware error has occurred.

Reported by component: Processor Core
Error Source: Machine Check Exception
Error Type: Cache Hierarchy Error
Processor APIC ID: 7
"


I have tried the following:
PSU: bought a cm650, issue persists.
RAM: played with timings, DOCP (even though none of the ram tests show any instability) to no effect.
CPU: disabled PBO with no effect. bought an aftermarket cooler with no effect.
GPU: updated vbios, tried 10+ diffent gpu driver versions with DDU in between. Tried quiet and normal bios.
HDD: bought a samsung 760evo as os drive, no effect.
mobo: flashed bios with different versions, no effect.

general: updated all drivers, played with power versions, tried windows 2004, 1803, 1903
temps are great: everything at max 70c, except for the GPU hotspot which gets up to 91c when stresstesting for 1h+
Assembled the system a couple times. MOBO is properly attached to case with all standoffs, nothing shorting, no dust.



Any advice would be appreciated. I suspect the CPU is the culprit.

 

[ntdouglas]

Its got to be the Radeon driver your using. I had the same problem. I was also having reboots when my system was just sitting at the desktop. It started happening when i kept updating every new driver. I cant remember which driver it is but go back say 8 previously released and start reading there descriptions or fixes and one them will say fixes in game crashes and windows rebooting issues. Its 20.something.something. I did a clean install of this one and the problem is gone. Also havent touched a new Amd driver since.

Ok just checked. Im running 20.1.2. Its been rock solid for months.

 

[Navipilot]

I have the same problem, you managed to solve the problem?

 

[RJARRRPCGP]

Looks far more serious this time. It doesn't look Radeon software-related at all. More likely a faulty CPU chip, unless there's a motherboard problem.

 

[exorch]

I had the same issue on one of my machines. AMD 3600, 32GB Samsung B-Die, RX 5700XT. All in a custom loop. drove me nuts to find the issue. always expected the gpu... exchanged PSU, mobo, etc.
Finally it was the SMT (AMD Version of HT). Disabled it in BIOS. the system was rock stable then.

If this fixes it for you, I would recommend to make the RMA via AMDs website after xmas/new year. for a dead 1600 it took 10 days until i received a new one. in europe they give you a DHL Express shipping sticker. great customer service. if you dont have a receipt they will take the SN and provide warranty time after manufacturing date (for 2nd hand buyers)

 

[Navipilot]

yesterday I tried to install the radeon 20.1.2 drivers, I reinstalled the chipset drivers for the msi b450 gaming plus max motherboard from the amd site, it seems I don't crash anymore, these days I will try more

My system specs:

CPU 3600 ryzen
msi b450 gaming plus max
5700 xt red devil
Crucial RGB 2x8GB 3200Mhz
1 SSD m.2 Adata 8200Pro 512 GB
1 SSD MX500 1 TB
1 SSD 860 EVO 1 TB
PSU 550W Seasonic Focus Gold
Fractal Mesify C
Artic 34

 

[HD64G]

Usually BIOS or UEFI update helps in those situations. Or resetting RAM settings (frequency-voltages). But another thing that might help is connect the GPU to the PSU with different cable for each 8-pin connector if you haven't already done so.

 

[Deleted member 193596]

same thing with my 5800X

 

[Navipilot]

these days I played about 5 hours of modern warfare and I didn't have any crashes. I would say that the video drivers from AMD are to blame. I use 20.1.2

I had the same issue on one of my machines. AMD 3600, 32GB Samsung B-Die, RX 5700XT. All in a custom loop. drove me nuts to find the issue. always expected the gpu... exchanged PSU, mobo, etc.
Finally it was the SMT (AMD Version of HT). Disabled it in BIOS. the system was rock stable then.

If this fixes it for you, I would recommend to make the RMA via AMDs website after xmas/new year. for a dead 1600 it took 10 days until i received a new one. in europe they give you a DHL Express shipping sticker. great customer service. if you dont have a receipt they will take the SN and provide warranty time after manufacturing date (for 2nd hand buyers)

try these drivers with SMT enabled. It would be a shame to keep that processor without SMT.

 

[Navipilot]

with those drivers 20.1.2 I had some textured artifacts and a bit of coilwine on the GPU.
Now I use these 20.4.2 drivers, I played a few hours COD Warzone and no crash, ok textures and without coilwine. So for now I will stick with these drivers.

 

[JelliedHealer]

Had the same problem, got in contact with Microsoft and turned off fast restart, worked like a charm

 

[TnF]

Same issue here, where is the OP to tell us if he found a fix?!