System stability problems?
System stability problems?
Thanks for everyone's suggestions so far. I have somewhat of a weird update: whenever I set the
intake
fans (not the AIO cooler or GPU fans) to max speed, the crashing is less frequent. The current fan setup is a 360mm AIO liquid cooler on the top for exhaust and three 120mm fans on the front for intake. The case is an NZXT H6 Flow.
I've crashed ~2 times in the last week, once during idle and another 10 minutes after that. It's still very difficult to replicate the crashing since there is no bsod or error logs (other than Kernel 41 in EventViewer). These are the temps at idle with some programs open:
Component
Temp (°C)
CPU
37
GPU
50
RAM DIMM1
37
RAM DIMM3
35
SSD M.2
48
SSD 1
25
SSD 2
27
Motherboard
24
I aggressively adjusted all fan curves to ramp up early in addition to ordering some Noctua fans to replace the front intake and another 120mm in the back for exhaust to see if this fixes the issue. I doubt the issue now is something as simple as overheating (especially since I crash mainly while idle and not during any stress tests), but if this fixes the issue I don't mind the constant fan noise. I'd rather take stability than performance at this point 😅
So after setting up the new fans and adjusting them to 50% idle speed, the problem has fully resolved. I've been using my PC for weeks without any issues at idle, medium, high, or variable loads. My RAM is running at 6000 MT/s. What helped me was aggressive fan tuning (you can hear them when idle), turning off any CPU/RAM enhancements in BIOS (like PBO and BCO), updating or installing all drivers to the latest versions (even beta or unused ones like the Wi-Fi driver). I also performed a full Windows reinstall, updated AMD chipset drivers, and Nvidia drivers.
My CPU never exceeded 60°C with my AIO cooler, but I believe the problem lies more with RAM stability heat. I plan to test standard fan curves in the coming weeks to check for crashes at lower speeds. Also, I turned off memory context restore so it retrains RAM timings each time I restart, which takes about 45 seconds.
Given these cold conditions, I believe the voltage is probably too low. It makes no sense to maintain such artificially low temperatures while losing performance and stability.
I am certain it can be resolved using fan curves. Yesterday, when I ran the fans at their standard settings, the system crashed within an hour. The RAM voltages during EXPO are now VDDCR_SOC 1.235V (up from 1.03V previously), meaning I’m no longer operating at low voltage levels. I rarely run highly CPU-demanding tasks and typically stay under 60°C, yet it still crashed during benchmark tests reaching 90°C.
Final update, hopefully things are now stable. My PC runs smoothly at 5200 Mhz, unlike before when I experienced occasional crashes above that speed. I reviewed the RAM QVL list for my CPU on AMD's site and confirmed they support up to 5200Mhz. Now I can run the fans at a normal speed without more crashes. The DRAM light issue likely pointed me in the right direction—it was just a CPU/RAM problem, nothing else. Switching to a different motherboard manufacturer also helped, as it had failed twice on the MSI model.
My friend, I reminded you back in May to change the BIOS back to the defaults (which means taking out the memory overclock). If you had done that, all your PC issues would have been fixed in just about five minutes now. Talk about a lucky break!
Anyway, good luck with the fix.