rml_layout
Usage
Physical memory layout of the tensor data.
API
Main
Possible Values
Name |
Value |
Brief Explanation |
|---|---|---|
RML_LAYOUT_UNSPECIFIED |
0 |
Unspecified layout. |
RML_LAYOUT_SCALAR |
200 |
Tensor layout for a scalar value. |
RML_LAYOUT_C |
210 |
Tensor layout for a one dimensional tensor. |
RML_LAYOUT_HW |
220 |
Tensor layout with the folowing dimensions: height, width. |
RML_LAYOUT_NC |
221 |
Tensor layout for a two dimensional tensor with data stored in the row-major order, where C - number of elements in a column, N - number of elements in a row. |
RML_LAYOUT_CHW |
230 |
Tensor layout for a single image in planar format with the following dimensions: number of channels, height, width. |
RML_LAYOUT_HWC |
231 |
Tensor layout for a single image in interleaved format with the following dimensions: height, width, number of channels. |
RML_LAYOUT_NCHW |
240 |
Tensor layout with the following dimensions: number of images (batch size) , number of channels, height and width. |
RML_LAYOUT_NHWC |
241 |
Tensor layout with the following dimensions: number of images (batch size), height, width, number of channels. |
RML_LAYOUT_OIHW |
242 |
Tensor layout with the following dimensions: number of output channels, number of input channels, height, width. |
RML_LAYOUT_HWIO |
243 |
Tensor layout with the following dimensions: height, width, number of input channels, number of output channels. |