diff options
author | Adnan AlSinan <adnan.alsinan@arm.com> | 2021-07-05 13:12:52 +0100 |
---|---|---|
committer | Georgios Pinitas <georgios.pinitas@arm.com> | 2021-07-25 13:04:23 +0000 |
commit | 7075fe2c5ee6f7cfe7cfd9454d905235e70b9ac4 (patch) | |
tree | b65671bdf37eb1ef8cc30ef64ab572da795546fa /src/core/CL/cl_kernels/copy_tensor.cl | |
parent | 22f5ed51f1b01f7cf6993a556a0b763e437926fc (diff) | |
download | ComputeLibrary-7075fe2c5ee6f7cfe7cfd9454d905235e70b9ac4.tar.gz |
Reorganize the kernels into nhwc, nchw and common folders
The Following kernels have been split into nchw/nhwc kernels files:
- batchnormalization_layer
- batch_to_space
- channel_shuffle
- depth_to_space
- dequantization_layer
- im2col
- normalization_layer
- normalize_planar_yuv_layer
- normalize_planar_yuv_layer_quantized
- pooling_layer
- pooling_layer_quantized
- remap
- reorg_layer
- scale
- scale_quantized
- space_to_batch
- space_to_depth
- upsample_layer
- winograd_filter_transform
- winograd_input_transform
- winograd_output_transform
The following kernels have been moved to nchw folder:
- direct_convolution1x1
- direct_convolution3x3
- direct_convolution5x5
- direct_convolution_quantized
- prior_box_layer
The following kernels have been moved to nhwc folder:
- direct_convolution
- dwc_native_fp_nhwc
- dwc_native_quantized_nhwc
The following kernels have been removed:
- sobel_filter
While the rest kerenls have been moved to the common folder.
Partially resolves COMPMID-4453
Signed-off-by: Adnan AlSinan <adnan.alsinan@arm.com>
Change-Id: Ic327ac935687ec351c610c65a3c6357f364a5a58
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5919
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/core/CL/cl_kernels/copy_tensor.cl')
-rw-r--r-- | src/core/CL/cl_kernels/copy_tensor.cl | 72 |
1 files changed, 0 insertions, 72 deletions
diff --git a/src/core/CL/cl_kernels/copy_tensor.cl b/src/core/CL/cl_kernels/copy_tensor.cl deleted file mode 100644 index 9c90969827..0000000000 --- a/src/core/CL/cl_kernels/copy_tensor.cl +++ /dev/null @@ -1,72 +0,0 @@ -/* - * Copyright (c) 2018-2020 Arm Limited. - * - * SPDX-License-Identifier: MIT - * - * Permission is hereby granted, free of charge, to any person obtaining a copy - * of this software and associated documentation files (the "Software"), to - * deal in the Software without restriction, including without limitation the - * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or - * sell copies of the Software, and to permit persons to whom the Software is - * furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in all - * copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE - * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER - * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, - * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ -#include "helpers.h" - -#if defined(DATA_TYPE) && defined(VEC_SIZE) && defined(VEC_SIZE_LEFTOVER) -/** Performs a copy of input tensor to the output tensor. - * - * @note The following variables must be passed at compile time: - * -# -DDATA_TYPE : Input and output datatypes. - * -# -DVEC_SIZE : The number of elements processed in X dimension - * -# -DVEC_SIZE_LEFTOVER: Leftover size in the X dimension; x_dimension % VEC_SIZE - * - * @param[in] in_ptr Pointer to the source tensor. Supported data types: All - * @param[in] in_stride_x Stride of the source tensor in X dimension (in bytes) - * @param[in] in_step_x input_stride_x * number of elements along X processed per workitem(in bytes) - * @param[in] in_stride_y Stride of the source tensor in Y dimension (in bytes) - * @param[in] in_step_y input_stride_y * number of elements along Y processed per workitem(in bytes) - * @param[in] in_stride_z Stride of the source tensor in Z dimension (in bytes) - * @param[in] in_step_z input_stride_z * number of elements along Z processed per workitem(in bytes) - * @param[in] in_offset_first_element_in_bytes The offset of the first element in the source tensor - * @param[out] out_ptr Pointer to the destination tensor. Supported data types: same as @p in_ptr - * @param[in] out_stride_x Stride of the destination tensor in X dimension (in bytes) - * @param[in] out_step_x output_stride_x * number of elements along X processed per workitem(in bytes) - * @param[in] out_stride_y Stride of the destination tensor in Y dimension (in bytes) - * @param[in] out_step_y output_stride_y * number of elements along Y processed per workitem(in bytes) - * @param[in] out_stride_z Stride of the source tensor in Z dimension (in bytes) - * @param[in] out_step_z output_stride_z * number of elements along Z processed per workitem(in bytes) - * @param[in] out_offset_first_element_in_bytes The offset of the first element in the destination tensor - */ -__kernel void copy_tensor( - TENSOR3D_DECLARATION(in), - TENSOR3D_DECLARATION(out)) -{ - Tensor3D in = CONVERT_TO_TENSOR3D_STRUCT(in); - Tensor3D out = CONVERT_TO_TENSOR3D_STRUCT(out); - - // Boundary-aware access: - // If the there's left-over in width (VEC_SIZE_LEFTOVER > 0): - // Shift all accesses other than the first to avoid accessing out of bounds - const int shift = max((int)(get_global_id(0) * VEC_SIZE) - (int)VEC_SIZE_LEFTOVER, 0) % VEC_SIZE; - in.ptr -= shift * in.stride_x; - out.ptr -= shift * out.stride_x; - - // Load data - VEC_DATA_TYPE(DATA_TYPE, VEC_SIZE) - data0 = VLOAD(VEC_SIZE)(0, (__global DATA_TYPE *)in.ptr); - - // Boundary-aware store - STORE_VECTOR_SELECT(data, DATA_TYPE, (__global DATA_TYPE *)out.ptr, VEC_SIZE, VEC_SIZE_LEFTOVER, VEC_SIZE_LEFTOVER != 0 && get_global_id(0) == 0); -} -#endif // defined(DATA_TYPE) && defined(VEC_SIZE) && defined(VEC_SIZE_LEFTOVER)
\ No newline at end of file |