System stability problems?
System stability problems?
Hello,
I have been struggling with an issue for the past couple months that's left me stumped. My PC on Windows 11 will randomly crash with no BSOD at seemingly random times. I can be playing a game or running a stress test, leaving it on idle with no programs open, watching a YouTube video, checking emails, etc. and it will crash at a random time. I took it to a repair shop that did a stress test on all components and replaced the motherboard and RAM, but the issue still persists (pc was running fine for 2 weeks after receiving it back). I also just replaced the motherboard 2 days ago and it was running fine, but today I had the same issue again. I have also replaced the PSU and have a UPS so I don't think it's a power issue. My question is do I just have a CPU issue, and what other types of tests can I perform to see what component is faulty?
Crashing symptoms:
my pc completely freezes and is unresponsive from all inputs for 1-5 seconds before the screen turns off (all components like fans and rgb stay on) before attempting to restart. Sometimes it will crash <10 minutes into booting up and logging into windows, other times I can be using it for 2 hours before it crashes. The luckiest crashes are it just restarts the system automatically, but sometimes I have to reseat the CPU/RAM or do a bios flashback to get back into windows. The worst crashes don't even let me get into BIOS even after a flashback.
For some of the crashes, the DRAM light on the motherboard will come on after it crashes but that can usually be fixed if I hold down the power button and restart. In the event of the first two motherboards, eventually the DRAM and CPU lights come on so they had to be replaced. Checking the pins/connections there are no bent pins on the socket or any other visual issues on the RAM or CPU. EventViewer always shows a Kernel Power 41 error, but there's no other crash logs.
Specs:
CPU: AMD Ryzen 9 7900X 4.7 GHz 12-Core Processor
CPU Cooler: NZXT Kraken 360 78.02 CFM Liquid
GPU: Gigabyte GAMING OC Rev. 2.0 GeForce RTX 3060 12GB
Motherboard: MSI MAG B650 Tomahawk WIFI ATX --> Replaced with another MSI MAG B650 --> ASUS TUF GAMING B650-PLUS WIFI
RAM: G.SKILL Flare X5 DDR5 64GB 6000MT/s CL30 --> Corsair Vengeance RGB 32GB DDR5-6000 CL36
PSU: NZXT C850 Gold ATX3.1 --> CORSAIR RM1000x Fully Modular ATX 3.1
Storage: WD_BLACK SN7100 2 TB M.2-2280 PCIe 4.0 X4 NVME SSD
What I've tried so far:
CPU:
Reseating it with new thermal paste (does not fix issue but helps a bit when I can't even get into BIOS)
Monitoring temps while idle (crashes when it's <60°C, never goes above 90°C)
Running stress tests with OCCT (no errors or crashes)
GPU:
Monitoring temps while idle (crashes when it's <60°C, doesn't go above 83°C)
Updating to the latest drivers through the NVIDIA app, but also using DDU and installing the latest driver manually
Running stress tests with OCCT (no errors or crashes)
Motherboard:
Updating BIOS to newest versions with all 3 motherboards (did not solve issue but helped with stability a tiny bit)
Turning off PBO
Running stress tests with OCCT (no errors or crashes)
Disabling fast boot
Enabling/disabling any overclocking for CPU and RAM
RAM:
Reseating or running with 1 stick (does not fix issue but helps a bit when I can't even get into BIOS)
Running memtest + windows memory diagnostic for RAM (passed both, did not find any issues)
Misc.
Checking all the PSU cable connections and making sure all fans spin
Completely reinstalling windows 11 with all the default apps (crashed within an hour with just keyboard + mouse plugged in)
Removing all software and components so there's nothing other than the base OS running (still crashes randomly)
Running SFC and DISM commands (no issues found), in addition to windows update and windows troubleshooters
Checking the status of the M.2 SSD (healthy 100%)
Running antivirus on EVERYTHING (Malwarebytes), nothing found
Changing power settings to balanced, ultra performance
Switching monitor ports, running off just 1 monitor
When XMP/EXPO/DOCP is active at 6000MT/s, turn off the memory overclock and revert to the JEDEC standard of 4800MT/s. The BIOS may attempt memory training during startup and might not locate compatible timings at 6000MT/s.
Check if you're using two DIMMs or four DIMMs; two DIMMs tend to offer better stability. Slots A2/B2 are typically ideal for a two-DIMM configuration.
Perform an extended MemTest86 session (several hours) on each RAM speed setting to verify stability. Any failures suggest an issue. RAM should perform normally when tested at 4800MT/s, though errors may appear at 6000MT/s.
For more info: https://www.memtest86.com/
Have you thought about a software-related problem?
Examine the programs you have installed and delete any you don’t recognize, except those that are necessary.
The sole critical event in Event Viewer is Kernel-Power 41 (63) with these specifics.
My configuration shows 1.015V Min / 1.054V Max from HWMonitor. Have you configured a constant 1.250V in BIOS?
I am using two DIMMs on A2/B2 with XMP/EXPO/DOCP disabled (current speed 4800MT/s, likely JEDEC speeds). These are brand new RAM modules and passed memtest86, though I can retest anytime. Should I test both simultaneously or sequentially?
I recently reinstalled Windows with no software installed and it still crashed, so I suspect it’s not a software problem.
Also updated: performed another 30-minute CPU + RAM and GPU stress test without crashing, confirming stability. Additionally, I disabled the advanced settings in Display -> Graphics -> Advanced -> Hardware-accelerated GPU scheduling to check if that improves stability, as frequent alt-tabbing between windows sometimes triggers the crash.
I have configured my RAM using Kingston (Hynix.a) 6000MHz Cl32 at 6200MHz Cl30. This might explain the necessity for a higher VDDCR_SOC voltage, though it should ideally be between 1.1 to 1.2 volts, depending on the memory controller's quality. I manually adjusted it, but it still fluctuates by about 10% based on the load.
It's interesting to note that after posting this, I've only experienced a single crash. The issue wasn't severe and it quickly restored itself into Windows within about 45 seconds. The adjustments I made significantly improved stability:
- Turned off hardware-accelerated GPU scheduling
- Returned to a previous BIOS version from 2024 instead of the newer one from 2025
- Disabled PBO in BIOS settings
- Turned off CPU Boost Clock Overdrive in BIOS
- Used Snappy Driver Installer for audio and other drivers
- Added Armoury Crate (motherboard software) and refreshed its contents
I'm looking forward to identifying the main cause once I have more components and conduct further testing.
Just reduced a few aspects of the performance, have you considered my recommendation?
The safest method to handle instability is to reinstall the BIOS to factory settings and ensure it's updated to the newest release. Should this fail, resorting to a nuclear option and installing from a Windows USB drive becomes the subsequent choice. If all else fails, it's likely a hardware issue.