When running DPA-1 with 'dp train train-2.json', it gets stuck at the 'OMP' step.

[wangyi01]

Summary

When using the 4090 card for operation, it will get stuck at:
OMP: Info #254: IMP AFFINTTY: pid 27858 tid 27975 thread 11 bound to OS proc set 5
OMP: Info #254: KMP AFFINITY: pid 27858 tid 27976 thread 19 bound to OS proc set 0

DeePMD-kit Version

deepmd-kit:2.2.1-cuda11.6

Backend and its version

It didn't run until printing the DeePMD-kit logo.

Details

There is no further information afterwards, and when checking with 'nvidia-smi', it is found that it did not start running.

[njzjz]

Will it still happen when you limit OMP threads?