Processors

[calee]

Specs:

CPU: Ryzen 5 2600x (stock, no overclocking)

Motherboard: MSI B450M MORTAR (bios version 7B89v1D)

RAM: Gloway DDR4 2666 8GBx2 TYA4U2666D19082C

Graphics card: Maxson GTX1650

SSD: Intel 760p 512GB

PSU: SUPER FLOWER GX450 450W

OS: Windows 10 Pro version 1909 build 18363.836

I build this PC at May 2019 and it works normal. Since May 13, when I was playing CS:GO the computer frequently freezing crash or blue screen. The computer needs hard reboot after a freezing crash. I tried to verify integrity of game file and found it not the problem. Then I found a method that steadily repeat the crash while I was checking CPU thermal: use aida64 or prime95 stress test CPU for 3 minutes or more then stop the stress test, the computer will freeze in 1 minute. The computer works normal while stress test.

Here's what I did for troubleshooting:

1. The very first BSOD code is tcpip.sys DRIVER_IRQL_NOT_LESS_OR_EQUAL. So I reinstalled the network driver (both ethernet and wireless module).
2. Then I got BSOD code PAGE_FAULT_IN_NONPAGED_AREA. I tried

DISM.exe /Online /Cleanup-image /Restorehealth

sfc /scannow

by Microsoft Community’s suggestion. I also tried Memtest86 to check RAM and there’s no errors.

3. Then I got UNEXPECTED_KERNEL_MODE_TRAP. Follow Microsoft Community’s suggestion I reinstalled system(Keep Everything), and it doesn’t work.
4. I tried update GPU driver, chipset driver and motherboard BIOS to latest version.
5. I tried clean install a new Windows10 on a different SSD and it didn’t solve. (seems not a software problem)
6. I tried remove Heatsink and CPU, clean and reapply thermal compound. (not a thermal problem. CPU temperature is around 60 degree under stress test)

I'm so annoying about this problem. Hope you guys have suggestions.

[curse127]

Most likely, the instability of the processor memory subsystem.
Do you use memory at 2666 MHz? Then there should be no problems with timings(all automatic?). You can fix the voltage at 1.35V and frequency 2933. Look at the Soc/NB parameter, manually set it to 1.05 or 1.1v. Set cldo_vddp parameter to 0.900v.

The problem may be due to insufficient processor voltages during transitions from a loaded state to a resting state.
All settings are in automatic mode? Look at the voltage the CPU puts on the metric board. Do you use Offset or LLC?
Try to put 2 or 3 level of LLC manually.

[calee]

I run all the settings in auto mode. The memory is at default (2666MHz 19-19-19-43).

I figure out that my channel B has a problem. One stick on channel B will cause the exact problem. Now I'm running both of the RAM on channel A and everything is OK. I'm wondering whether it's CPU memory controller malfunction or motherboard malfunction.

[mccharle]

hello,

you have tried on another SSD, but the first one did it stay connected ?

AMD chipset drivers 2.04 installed ?

SSD intel 760P have the latest firmware 3.0.9 ?

Latest drivers for NVME installed "intel NVMe driver 4.4.0.1003" ?

Tried another PSU if all above don't work.

With windows 10 updating continuously, sometimes your system can crash with incompatibility between new update and old drivers installed or with only default drivers.

[ianl1427]

I have a similar problem. Laptop configuration is Ryzen 5 4600H 16Gb with GTX1650 4Gb. Laptop is less than a month old. Worked for less than 2 weeks, Passmark and Assetto Corsa was able to run fine until a bunch of windows update caused the GTX to crash upon any launch of 3D. Checked nVidia forum and many of the similar problems reported were also running AMD. Intel users hardly have any problems it seems....

One of the users reported that turning on Low Latency Mode in nVidia Conrtrol Panel helps. I will try this also and report back.