Strange system's hard disk problem with infinite while loop in CUDA kernel

Dear all,

I have the following unusual and strange problem:

I am running a kernel of matrix multiplication. The only deference is that I have force the kernel’s operations to run forever
global void kernel


in order to keep the GPU forever busy.

The prolblem is that when I run that kernel the hard disk of system gets unmounted(!)
I know that writting an infinite while loop in the CUDA kernel might not be a good idea about GPU (in terms that the CUDA driver might kill this process) but why my hard disk keeps unmounting? How the infinite GPU execution affect the hard disk? I read and write nothing from the hard disk.