get_device

paddle.Tensor. get_device ( self: Tensor ) → int

Return the device id where the Tensor is located.

Returns

The device id of the Tensor. Returns -1 for CPU tensors; for GPU tensors,

returns the CUDA device id (e.g., 0 for gpu:0).

Return type

int

Examples

>>> import paddle
>>> x = paddle.to_tensor([1, 2, 3], place=paddle.CPUPlace())
>>> x.get_device()
-1

>>> 
>>> y = paddle.to_tensor([1, 2, 3], place=paddle.CUDAPlace(0))
>>> y.get_device()
0