You're experiencing unpredictable crashes on your low-mid range PC, but it's hard to pinpoint the issue.
You're experiencing unpredictable crashes on your low-mid range PC, but it's hard to pinpoint the issue.
Over the past week, I've experienced recurring, though unpredictable, shutdowns on my computer. Here are the details:
My system runs Debian GNU/Linux trixie/sid x86_64, with a price tag of $$. Host details: MS-7C95 1.0, kernel version 6.9.10-amd64. Uptime is about 50 minutes. Packages installed total 3667 (dpkg). Shell version is bash 5.2.21; resolution is 1920x1080.
I'm using Plasma 5.27.11, and the environment supports WM and GTK2/3 themes. I'm on a Konsole terminal. CPU specs: AMD Ryzen 5 5600X (12 cores) at 3.700GHz.
I tried connecting a breadboard and Arduino to lower the GPU voltage from 12V to +5V, allowing me to connect directly to my CPU via serial for logging. I captured some crash images and recovered from a few incidents. Most crashes occurred during Minecraft gameplay with shaders, but I also faced issues with Firefox (a tab crash, page fault killing a Minecraft instance), and CPU errors linked to cache invalidation at the kernel level.
Initially, I suspected the GPU was faulty, but the pattern of errors suggests RAM might be the culprit. When GPU crashes happen, I can use systemctl isolate rescue.target to restart without a full reboot, letting me resume from the last safe state. I ran Memtest86+ and it passed, though it might need more time. After replacing RAM and GPU, the gold pins looked undamaged.
Any suggestions on what might be causing these problems? I suspect RAM could be the root cause, given the shared memory issues.
Start from a portable OS like Ubuntu to check for crashes, which would indicate hardware issues.
I’d prefer to avoid these issues, as the problems seem rooted in hardware rather than software. The cache invalidation and Linux kernel have already flagged similar issues, so running a live system wouldn’t help. I’m checking how much I pushed the GPU voltage down, since overheating can spike to 90°C, and wondering if my aggressive undervolting might have led to power instability.