Linux Mint used to freeze randomly even under light load, especially on an older kernel. After switching to a newer generic kernel, those random freezes stopped, but the whole system still locks up when I put it under extreme load, such as running two demanding games at once. The display becomes unresponsive and I have to force the computer off and reboot.
On Windows 10, the same kind of stress usually caused the individual game or program to stop responding while the operating system remained usable. I have tested the RAM and drives without finding errors, and I have reproduced similar behavior on another machine. RAM usage has stayed below about 21 GB out of 32 GB, and the system has roughly 2.1 GB of swap. The CPU temperature is normally below 61°C under comparable Windows workloads, although I have not yet confirmed the temperatures immediately before a Linux freeze.
Is this expected Linux behavior, or could it be specific to Mint, Ubuntu, Debian, the graphics stack, or the kernel? What should I check to determine whether the cause is hardware, thermals, power delivery, memory pressure, or software? My system has a Ryzen 5 3600X, Radeon RX 6700 XT, 32 GB of DDR4 memory, a 256 GB system SSD, and two 1 TB NVMe game drives.
4 Answers
Check memory pressure as well as total RAM usage. A memory leak or exhausted swap can make the desktop appear completely dead, although using only around 21 GB of 32 GB makes ordinary RAM exhaustion less likely. Watch RAM, swap, and GPU memory while testing, and inspect the previous boot's logs after a forced restart for kernel, GPU, or out-of-memory errors.
Since changing kernels reduced the random freezes, a kernel, graphics driver, or desktop-stack issue is still possible. Try a different supported kernel or graphics stack and compare results, but do not assume that Debian-based systems inherently kill the entire desktop under load. Normally a program should fail before the whole machine does; a total lockup suggests a driver, hardware, power, or thermal problem.
Thermals are an important clue. A CPU or GPU can freeze or reboot when it overheats, even if temperatures look fine during normal use. Monitor CPU and GPU temperatures, clocks, and fan speeds while reproducing the problem. Dust, poor contact, a failing fan, or old thermal paste can all contribute.
The CPU stayed below about 61°C under similar workloads in Windows, but I have not checked the Linux temperatures right before a freeze. I will monitor them during a stress test.
A complete lockup under heavy load usually points more toward hardware or power delivery than the distribution itself. I would check the power supply, motherboard, GPU power connections, and stability under CPU- and GPU-heavy tests. RAM testing is also worth repeating, preferably with a bootable memory tester, even if a previous test passed.
Windows has been stable on this machine and usually only terminates the game when pushed hard. RAM and drive tests were clean recently, and I have not noticed unusually high memory usage, so I am going to focus on temperatures, GPU logs, power delivery, and kernel messages next.