NVIDIA Holoscan SDK v0.5.1
Holoscan v0.5.1

Function holoscan::calc_strides

void holoscan::calc_strides(const DLTensor &tensor, std::vector<int64_t> &strides, bool to_num_elements = false)

Fill strides from the given DLTensor object.

The following fields are used to fill strides:

  • ndim

  • shape

  • dtype

If tensor’s strides is nullptr, strides argument is filled with the calculated strides of the given DLTensor object. Otherwise, strides argument is filled with the given DLTensor object’s strides. strides vector would be resized to the size of ndim field of the given DLTensor object.

Parameters
  • tensor – DLTensor object that holds information to fill strides.

  • strides[out] Strides to fill.

  • to_num_elments – If true, the strides in strides argument are in number of elements, not bytes (default: false).

© Copyright 2022-2023, NVIDIA. Last updated on Jul 28, 2023.