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>

1 files changed, 154 insertions, 0 deletions

diff --git a/src/core/CL/cl_kernels/common/fft_digit_reverse.cl b/src/core/CL/cl_kernels/common/fft_digit_reverse.cl
new file mode 100644
index 0000000000..5f64d95bf9
--- /dev/null
+++ b/src/core/CL/cl_kernels/common/fft_digit_reverse.cl
@@ -0,0 +1,154 @@
+/*
+ * Copyright (c) 2019-2021 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(VEC_SIZE) && defined(DATA_TYPE)
+/** Computes the digit reverse stage on axis X
+ *
+ * @param[in]  src_ptr                           Pointer to the source tensor. Supported data types: F16/F32
+ * @param[in]  src_stride_x                      Stride of the source tensor in X dimension (in bytes)
+ * @param[in]  src_step_x                        src_stride_x * number of elements along X processed per workitem(in bytes)
+ * @param[in]  src_stride_y                      Stride of the source tensor in Y dimension (in bytes)
+ * @param[in]  src_step_y                        src_stride_y * number of elements along Y processed per workitem(in bytes)
+ * @param[in]  src_stride_z                      Stride of the source tensor in Z dimension (in bytes)
+ * @param[in]  src_step_z                        src_stride_z * number of elements along Z processed per workitem(in bytes)
+ * @param[in]  src_offset_first_element_in_bytes The offset of the first element in the source tensor
+ * @param[out] dst_ptr                           Pointer to the destination tensor. Supported data types: same as @p src_ptr
+ * @param[in]  dst_stride_x                      Stride of the destination tensor in X dimension (in bytes)
+ * @param[in]  dst_step_x                        dst_stride_x * number of elements along X processed per workitem(in bytes)
+ * @param[in]  dst_stride_y                      Stride of the destination tensor in Y dimension (in bytes)
+ * @param[in]  dst_step_y                        dst_stride_y * number of elements along Y processed per workitem(in bytes)
+ * @param[in]  dst_stride_z                      Stride of the source tensor in Z dimension (in bytes)
+ * @param[in]  dst_step_z                        dst_stride_z * number of elements along Z processed per workitem(in bytes)
+ * @param[in]  dst_offset_first_element_in_bytes The offset of the first element in the destination tensor
+ * @param[in]  idx_ptr                           Pointer to the index tensor. Supported data types: U32
+ * @param[in]  idx_stride_x                      Stride of the index tensor in X dimension (in bytes)
+ * @param[in]  idx_step_x                        idx_stride_x * number of elements along X processed per workitem(in bytes)
+ * @param[in]  idx_offset_first_element_in_bytes The offset of the first element in the index tensor
+ */
+__kernel void fft_digit_reverse_axis_0(
+    TENSOR3D_DECLARATION(src),
+    TENSOR3D_DECLARATION(dst),
+    VECTOR_DECLARATION(idx))
+{
+    // Get tensor pointers
+    Tensor3D src = CONVERT_TO_TENSOR3D_STRUCT_NO_STEP(src);
+    Tensor3D dst = CONVERT_TO_TENSOR3D_STRUCT(dst);
+    Vector   idx = CONVERT_TO_VECTOR_STRUCT(idx);
+
+    const unsigned int iidx = *((__global uint *)(idx.ptr));
+
+    // Load data
+#if VEC_SIZE == 1
+    DATA_TYPE data = *((__global DATA_TYPE *)tensor3D_offset(&src, iidx, get_global_id(1), get_global_id(2)));
+#elif VEC_SIZE == 2
+    VEC_DATA_TYPE(DATA_TYPE, 2)
+    data = vload2(0, (__global DATA_TYPE *)tensor3D_offset(&src, iidx, get_global_id(1), get_global_id(2)));
+#else // VEC_SIZE == 1
+#error "vec_size of 1 and 2 are supported"
+#endif // VEC_SIZE == 1
+
+    // Create result
+#if VEC_SIZE == 1
+    VEC_DATA_TYPE(DATA_TYPE, 2)
+    res = { data, 0 };
+#elif VEC_SIZE == 2
+    VEC_DATA_TYPE(DATA_TYPE, 2)
+    res = data;
+#else // VEC_SIZE == 1
+#error "vec_size of 1 and 2 are supported"
+#endif // VEC_SIZE == 1
+
+    // Store result
+#if defined(CONJ)
+    vstore2((VEC_DATA_TYPE(DATA_TYPE, 2))(res.s0, -res.s1), 0, (__global DATA_TYPE *)dst.ptr);
+#else  // defined(CONJ)
+    vstore2(res, 0, (__global DATA_TYPE *)dst.ptr);
+#endif // defined(CONJ)
+}
+
+/** Computes the digit reverse stage on axis Y
+ *
+ * @param[in]  src_ptr                           Pointer to the source tensor. Supported data types: F16/F32
+ * @param[in]  src_stride_x                      Stride of the source tensor in X dimension (in bytes)
+ * @param[in]  src_step_x                        src_stride_x * number of elements along X processed per workitem(in bytes)
+ * @param[in]  src_stride_y                      Stride of the source tensor in Y dimension (in bytes)
+ * @param[in]  src_step_y                        src_stride_y * number of elements along Y processed per workitem(in bytes)
+ * @param[in]  src_stride_z                      Stride of the source tensor in Z dimension (in bytes)
+ * @param[in]  src_step_z                        src_stride_z * number of elements along Z processed per workitem(in bytes)
+ * @param[in]  src_offset_first_element_in_bytes The offset of the first element in the source tensor
+ * @param[out] dst_ptr                           Pointer to the destination tensor. Supported data types: same as @p src_ptr
+ * @param[in]  dst_stride_x                      Stride of the destination tensor in X dimension (in bytes)
+ * @param[in]  dst_step_x                        dst_stride_x * number of elements along X processed per workitem(in bytes)
+ * @param[in]  dst_stride_y                      Stride of the destination tensor in Y dimension (in bytes)
+ * @param[in]  dst_step_y                        dst_stride_y * number of elements along Y processed per workitem(in bytes)
+ * @param[in]  dst_stride_z                      Stride of the source tensor in Z dimension (in bytes)
+ * @param[in]  dst_step_z                        dst_stride_z * number of elements along Z processed per workitem(in bytes)
+ * @param[in]  dst_offset_first_element_in_bytes The offset of the first element in the destination tensor
+ * @param[in]  idx_ptr                           Pointer to the index tensor. Supported data types: U32
+ * @param[in]  idx_stride_x                      Stride of the index tensor in X dimension (in bytes)
+ * @param[in]  idx_step_x                        idx_stride_x * number of elements along X processed per workitem(in bytes)
+ * @param[in]  idx_offset_first_element_in_bytes The offset of the first element in the index tensor
+ */
+__kernel void fft_digit_reverse_axis_1(
+    TENSOR3D_DECLARATION(src),
+    TENSOR3D_DECLARATION(dst),
+    VECTOR_DECLARATION(idx))
+{
+    // Get tensor pointers
+    Tensor3D src = CONVERT_TO_TENSOR3D_STRUCT_NO_STEP(src);
+    Tensor3D dst = CONVERT_TO_TENSOR3D_STRUCT(dst);
+    Vector   idx = CONVERT_TO_VECTOR_STRUCT_NO_STEP(idx);
+
+    const unsigned int iidx = *((__global uint *)vector_offset(&idx, (int)(get_global_id(1))));
+
+    // Load data
+#if VEC_SIZE == 1
+    DATA_TYPE data = *((__global DATA_TYPE *)tensor3D_offset(&src, get_global_id(0), iidx, get_global_id(2)));
+#elif VEC_SIZE == 2
+    VEC_DATA_TYPE(DATA_TYPE, 2)
+    data = vload2(0, (__global DATA_TYPE *)tensor3D_offset(&src, get_global_id(0), iidx, get_global_id(2)));
+#else // VEC_SIZE == 1
+#error "vec_size of 1 and 2 are supported"
+#endif // VEC_SIZE == 1
+
+    // Create result
+#if VEC_SIZE == 1
+    VEC_DATA_TYPE(DATA_TYPE, 2)
+    res = { data, 0 };
+#elif VEC_SIZE == 2
+    VEC_DATA_TYPE(DATA_TYPE, 2)
+    res = data;
+#else // VEC_SIZE == 1
+#error "vec_size of 1 and 2 are supported"
+#endif // VEC_SIZE == 1
+
+    // Store result
+#if defined(CONJ)
+    vstore2((VEC_DATA_TYPE(DATA_TYPE, 2))(res.s0, -res.s1), 0, (__global DATA_TYPE *)dst.ptr);
+#else  // defined(CONJ)
+    vstore2(res, 0, (__global DATA_TYPE *)dst.ptr);
+#endif // defined(CONJ)
+}
+#endif // defined(VEC_SIZE) && defined(DATA_TYPE)
\ No newline at end of file