Torch.onnx.export can use multiprocessing.pool?

1

Hi, everyone
I have a question.
When I use multiprocessing.Pool to torch.onnx.export, it is slower than before.
Has anyone ever met this question? What’s the solution?

The code example is as follows:
for coding_value in list1:
pool.apply_async(process_one,args=(coding_value,))
pool.close()
pool.join()

the <process_one> includes
torch.onnx.export(xxx(coding=coding_value), (input), onnx_path, verbose=False, input_names=[‘Placeholder’], output_names=[‘output’], opset_version=11)