Why does torch.eigh function use multiple cpus?

Hello,

May I ask why torch.eigh uses multiple cpus? Some clues may exist in the torch function apply_lapack_eigh and the corresponding LAPACK function SSYEVD. I suspect that somewhere multithreading or multiprocessing was called, which is why multiple CPUs were utilized. However, after reading these codes, I still haven’t found the specific command that calls multithreading or multiprocessing.

Thanks in advance