Compute Library
18.05
|
#include "helpers.h"
Go to the source code of this file.
Functions | |
__kernel void | concatenate_width (__global uchar *src_ptr, uint src_stride_x, uint src_step_x, uint src_stride_y, uint src_step_y, uint src_stride_z, uint src_step_z, uint src_offset_first_element_in_bytes, __global uchar *dst_ptr, uint dst_stride_x, uint dst_step_x, uint dst_stride_y, uint dst_step_y, uint dst_stride_z, uint dst_step_z, uint dst_offset_first_element_in_bytes, int offset) |
This kernel concatenates the input tensor into the output tensor along the first dimension. More... | |
__kernel void | concatenate_depth (__global uchar *src_ptr, uint src_stride_x, uint src_step_x, uint src_stride_y, uint src_step_y, uint src_stride_z, uint src_step_z, uint src_offset_first_element_in_bytes, __global uchar *dst_ptr, uint dst_stride_x, uint dst_step_x, uint dst_stride_y, uint dst_step_y, uint dst_stride_z, uint dst_step_z, uint dst_offset_first_element_in_bytes, int3 offsets) |
This kernel concatenates the input tensor into the output tensor along the third dimension. More... | |
__kernel void concatenate_depth | ( | __global uchar * | src_ptr, |
uint | src_stride_x, | ||
uint | src_step_x, | ||
uint | src_stride_y, | ||
uint | src_step_y, | ||
uint | src_stride_z, | ||
uint | src_step_z, | ||
uint | src_offset_first_element_in_bytes, | ||
__global uchar * | dst_ptr, | ||
uint | dst_stride_x, | ||
uint | dst_step_x, | ||
uint | dst_stride_y, | ||
uint | dst_step_y, | ||
uint | dst_stride_z, | ||
uint | dst_step_z, | ||
uint | dst_offset_first_element_in_bytes, | ||
int3 | offsets | ||
) |
This kernel concatenates the input tensor into the output tensor along the third dimension.
[in] | src_ptr | Pointer to the source tensor. Supported data types: QS8, QS16, F16, F32 |
[in] | src_stride_x | Stride of the source tensor in X dimension (in bytes) |
[in] | src_step_x | src_stride_x * number of elements along X processed per workitem(in bytes) |
[in] | src_stride_y | Stride of the source tensor in Y dimension (in bytes) |
[in] | src_step_y | src_stride_y * number of elements along Y processed per workitem(in bytes) |
[in] | src_stride_z | Stride of the source tensor in Z dimension (in bytes) |
[in] | src_step_z | src_stride_z * number of elements along Z processed per workitem(in bytes) |
[in] | src_offset_first_element_in_bytes | The offset of the first element in the source tensor |
[out] | dst_ptr | Pointer to the destination tensor. Supported data types: same as src_ptr |
[in] | dst_stride_x | Stride of the destination tensor in X dimension (in bytes) |
[in] | dst_step_x | dst_stride_x * number of elements along X processed per workitem(in bytes) |
[in] | dst_stride_y | Stride of the destination tensor in Y dimension (in bytes) |
[in] | dst_step_y | dst_stride_y * number of elements along Y processed per workitem(in bytes) |
[in] | dst_stride_z | Stride of the source tensor in Z dimension (in bytes) |
[in] | dst_step_z | dst_stride_z * number of elements along Z processed per workitem(in bytes) |
[in] | dst_offset_first_element_in_bytes | The offset of the first element in the destination tensor |
[in] | offsets | The offsets to the first valid element of the output tensor in bytes |
Definition at line 81 of file concatenate.cl.
References CONVERT_TO_TENSOR3D_STRUCT, DATA_TYPE, arm_compute::test::validation::dst, Tensor3D::ptr, arm_compute::test::validation::src, tensor3D_offset(), VEC_DATA_TYPE, VLOAD, and VSTORE.
__kernel void concatenate_width | ( | __global uchar * | src_ptr, |
uint | src_stride_x, | ||
uint | src_step_x, | ||
uint | src_stride_y, | ||
uint | src_step_y, | ||
uint | src_stride_z, | ||
uint | src_step_z, | ||
uint | src_offset_first_element_in_bytes, | ||
__global uchar * | dst_ptr, | ||
uint | dst_stride_x, | ||
uint | dst_step_x, | ||
uint | dst_stride_y, | ||
uint | dst_step_y, | ||
uint | dst_stride_z, | ||
uint | dst_step_z, | ||
uint | dst_offset_first_element_in_bytes, | ||
int | offset | ||
) |
This kernel concatenates the input tensor into the output tensor along the first dimension.
[in] | src_ptr | Pointer to the source tensor. Supported data types: QS8, QASYMM8, QS16, F16, F32 |
[in] | src_stride_x | Stride of the source tensor in X dimension (in bytes) |
[in] | src_step_x | src_stride_x * number of elements along X processed per workitem(in bytes) |
[in] | src_stride_y | Stride of the source tensor in Y dimension (in bytes) |
[in] | src_step_y | src_stride_y * number of elements along Y processed per workitem(in bytes) |
[in] | src_stride_z | Stride of the source tensor in Z dimension (in bytes) |
[in] | src_step_z | src_stride_z * number of elements along Z processed per workitem(in bytes) |
[in] | src_offset_first_element_in_bytes | The offset of the first element in the source tensor |
[out] | dst_ptr | Pointer to the destination tensor. Supported data types: same as src_ptr |
[in] | dst_stride_x | Stride of the destination tensor in X dimension (in bytes) |
[in] | dst_step_x | dst_stride_x * number of elements along X processed per workitem(in bytes) |
[in] | dst_stride_y | Stride of the destination tensor in Y dimension (in bytes) |
[in] | dst_step_y | dst_stride_y * number of elements along Y processed per workitem(in bytes) |
[in] | dst_stride_z | Stride of the source tensor in Z dimension (in bytes) |
[in] | dst_step_z | dst_stride_z * number of elements along Z processed per workitem(in bytes) |
[in] | dst_offset_first_element_in_bytes | The offset of the first element in the destination tensor |
[in] | offset | The offset to the first valid element of the output tensor in bytes |
Definition at line 46 of file concatenate.cl.
References CONVERT_TO_TENSOR3D_STRUCT, DATA_TYPE, arm_compute::test::validation::dst, Tensor3D::ptr, arm_compute::test::validation::src, VEC_DATA_TYPE, VLOAD, and VSTORE.