Bizarre Crash
Bizarre Crash
I am experiencing a very strange crashing problem.
My computer will abruptly shut down without any additional details in the event log beyond the standard 'unexpected shut down' for a power failure. After booting, sometimes it displays a message saying "hot component detected." I increased the fan speed and monitored temperatures using 'open hardware monitor.' The temperature readings stayed below 70°C after adjusting the fan, yet the issue persisted. I then switched to nuclear mode, opened the case, and placed a large room fan inside, but this did not resolve the problem. The logs show CPU, GPU, HDD, and SSD readings—no power or RAM sensors were active.
What becomes more unusual is that the problem only appears when running two ANN models simultaneously: yolov11 and SAM-2. It happens no matter whether I store data on an SSD or HDD.
I understand ANNs are demanding, so I tried FURMARK, but it worked perfectly without any issues! Unlike the ANN workload, FURMARK can raise GPU temperatures up to 95°C, which is quite surprising. CPU stress tests also showed no problems.
Specifications:
CPU: i5-6600
GPU: NVidia 2060 Super
RAM: 32GB "My Technology Geeks" (the name of the reman manufacturer, assumed to be a Chinese company)
SSD: 512GB 'already'. Definitely from China; temperatures consistently report around 40°C even when other components are nearer to 30.
HDD: 3TB Hitachi
PSU: Dell (365W), which logged a maximum power draw of 258w during ANN usage (CPU package, cores, graphics, DRAM, GPU included).
From what I can gather, this issue shouldn’t occur.
is this a 2x 16GB configuration, two separate 16GB units, four 8GB units, etc.?
Proceed with running several rounds of Memtest from a USB drive to check for issues.
Typically these problems stem from a power supply that struggles under the load.
Since you perform stress tests without similar responses, monitor power consumption during these tasks and observe if the multiple AAN instances contribute to increased power usage.