I have been suffering from stability issues with my setup. I can usually run demanding stress tests for hours, including Karhu and VT3 while simulatenously stress testing the GPU. No errors. Then after reboot the overclock might become very unstable. At worst, I got error in Karhu after just one second.
This profile was stable for 4 hours in VT3 and 2 hours in Karhu. Just today I got a BSOD and ran Karhu just in case. Well, it is unstable again.
What I have tried:
- Updated bios
- Tried setting up manual ProcODT values
- Tried loosening timings
- Tried different VSOC values
- Tried removing CO
- Ensured that memory context restore is disabled
My RAM kit is a 3,5 years old 5600Mhz CL36 Hynix M-die kit. If I am able to run stress tests for hours but reboot and new memory training messes everything up, this shouldn't be IMC or RAM degradation, right? I also ran the same RAM kit at 6400Mhz CL30 for 3 years on Intel platform. Zero issues back then but being on AM5 has been a nightmare.