src/core/CL/cl_kernels/winograd.cl

/*
 * Copyright (c) 2018 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(NUM_CHANNELS)

/** This OpenCL kernel performs Winograd filter transform 3x3 when the data format is NCHW and the output tile is 2x2
 *
 * @note The number of channels must be passed at compile time using -DNUM_CHANNELS: e.g. -DNUM_CHANNELS=64
 *
 * @param[in]  src_ptr                           Pointer to the source tensor. Supported data types: 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_stride_w                      Stride of the source tensor in W dimension (in bytes)
 * @param[in]  src_step_w                        src_stride_w * number of elements along W 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]  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]  dst_offset_first_element_in_bytes The offset of the first element in the destination tensor
 */
__kernel void winograd_filter_transform_2x2_3x3_nchw(
    TENSOR4D_DECLARATION(src),
    TENSOR3D_DECLARATION(dst))
{
    Tensor4D src = CONVERT_TO_TENSOR4D_STRUCT(src, NUM_CHANNELS);

    const __global uchar *src_addr = tensor4D_offset(&src, 0, 0, 0, 0);

    // Load the values from the input tensor
    float3 w0 = vload3(0, (__global float *)(src_addr + 0 * src_stride_y));
    float3 w1 = vload3(0, (__global float *)(src_addr + 1 * src_stride_y));
    float3 w2 = vload3(0, (__global float *)(src_addr + 2 * src_stride_y));

    // Transform the 3x3 tile in a 4x4 tile
    float4 out0 = 0.0f;
    float4 out1 = 0.0f;
    float4 out2 = 0.0f;
    float4 out3 = 0.0f;

    // Row 0
    out0.s0 = (w0.s0);
    out0.s1 = (w0.s0 + w0.s1 + w0.s2) * 0.5f;
    out0.s2 = (w0.s0 + w0.s2 - w0.s1) * 0.5f;
    out0.s3 = (w0.s2);

    // Row 1
    out1.s0 = (w0.s0 + w1.s0 + w2.s0) * 0.5f;
    out1.s1 = (w0.s0 + w1.s0 + w2.s0 + w0.s1 + w1.s1 + w2.s1 + w0.s2 + w1.s2 + w2.s2) * 0.25f;
    out1.s2 = (w0.s0 + w1.s0 + w2.s0 + w0.s2 + w1.s2 + w2.s2 - w0.s1 - w1.s1 - w2.s1) * 0.25f;
    out1.s3 = (w0.s2 + w1.s2 + w2.s2) * 0.5f;

    // Row 2
    out2.s0 = (w0.s0 + w2.s0 - w1.s0) * 0.5f;
    out2.s1 = (w0.s0 + w2.s0 + w0.s1 + w2.s1 + w0.s2 + w2.s2 - w1.s0 - w1.s1 - w1.s2) * 0.25f;
    out2.s2 = (w0.s0 + w2.s0 + w1.s1 + w0.s2 + w2.s2 - w1.s0 - w0.s1 - w2.s1 - w1.s2) * 0.25f;
    out2.s3 = (w0.s2 + w2.s2 - w1.s2) * 0.5f;

    // Row 3
    out3.s0 = (w2.s0);
    out3.s1 = (w2.s0 + w2.s1 + w2.s2) * 0.5f;
    out3.s2 = (w2.s0 + w2.s2 - w2.s1) * 0.5f;
    out3.s3 = (w2.s2);

    int z  = get_global_id(2);
    int x0 = z / NUM_CHANNELS; // idx filter
    int y0 = z % NUM_CHANNELS; // idx channel

    // Get output address
    __global uchar *dst_addr = dst_ptr + dst_offset_first_element_in_bytes + x0 * dst_stride_x + y0 * dst_stride_y;

    // Store the 16 values across the 16 channels
    *(__global float *)(dst_addr + 0 * dst_stride_z)  = out0.s0;
    *(__global float *)(dst_addr + 1 * dst_stride_z)  = out0.s1;
    *(__global float *)(dst_addr + 2 * dst_stride_z)  = out0.s2;
    *(__global float *)(dst_addr + 3 * dst_stride_z)  = out0.s3;
    *(__global float *)(dst_addr + 4 * dst_stride_z)  = out1.s0;
    *(__global float *)(dst_addr + 5 * dst_stride_z)  = out1.s1;
    *(__global float *)(dst_addr + 6 * dst_stride_z)  = out1.s2;
    *(__global float *)(dst_addr + 7 * dst_stride_z)  = out1.s3;
    *(__global float *)(dst_addr + 8 * dst_stride_z)  = out2.s0;
    *(__global float *)(dst_addr + 9 * dst_stride_z)  = out2.s1;
    *(__global float *)(dst_addr + 10 * dst_stride_z) = out2.s2;
    *(__global float *)(dst_addr + 11 * dst_stride_z) = out2.s3;
    *(__global float *)(dst_addr + 12 * dst_stride_z) = out3.s0;
    *(__global float *)(dst_addr + 13 * dst_stride_z) = out3.s1;
    *(__global float *)(dst_addr + 14 * dst_stride_z) = out3.s2;
    *(__global float *)(dst_addr + 15 * dst_stride_z) = out3.s3;
}
#endif // defined(NUM_CHANNELS)

#if defined(NUM_TILES_X) && defined(PAD_LEFT) && defined(PAD_TOP)
/** This OpenCL kernel computes the input transform when the kernel size is 3x3 and the output tile is 2x2
 *
 * @note The number of tiles in the x axis must be passed at compile time using -DNUM_TILES_X (i.e.-DNUM_TILES_X=5).
 * @note The pad left and pad top must be passed at compile time using -DPAD_LEFT and -DPAD_TOP (i.e.-DPAD_LEFT=1 and -DPAD_TOP=0).
 *
 * @param[in] src_ptr                           Pointer to the source image. Supported data types: F32
 * @param[in] src_stride_x                      Stride of the source image 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 image 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_offset_first_element_in_bytes The offset of the first element in the source image
 * @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 Y processed per workitem(in bytes)
 * @param[in] dst_ptr                           Pointer to the destination tensor. Supported data types: 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 destination tensor in Z dimension (in bytes)
 * @param[in] dst_step_z                        dst_stride_z * number of elements along Y processed per workitem(in bytes)
 * @param[in] dst_offset_first_element_in_bytes The offset of the first element in the destination tensor
 */
__kernel void winograd_input_transform_2x2_3x3_stepz1_nchw(
    TENSOR3D_DECLARATION(src),
    TENSOR3D_DECLARATION(dst))
{
    int x = get_global_id(0);
    int y = get_global_id(1);
    int z = get_global_id(2);

    // Compute input address
    __global uchar *src_addr = src_ptr + src_offset_first_element_in_bytes + x * 2 * src_stride_x + y * 2 * src_stride_y + z * src_stride_z;

    src_addr = src_addr - ((int)PAD_LEFT * src_stride_x) - ((int)PAD_TOP * src_stride_y);

    float4 in_row0 = vload4(0, (__global float *)(src_addr + 0 * src_stride_y));
    float4 in_row1 = vload4(0, (__global float *)(src_addr + 1 * src_stride_y));
    float4 in_row2 = vload4(0, (__global float *)(src_addr + 2 * src_stride_y));
    float4 in_row3 = vload4(0, (__global float *)(src_addr + 3 * src_stride_y));

    float4 tmp0 = in_row0 - in_row2;
    float4 tmp1 = in_row1 + in_row2;
    float4 tmp2 = in_row2 - in_row1;
    float4 tmp3 = in_row1 - in_row3;

    float out00 = tmp0.s0 - tmp0.s2;
    float out01 = tmp0.s1 + tmp0.s2;
    float out02 = tmp0.s2 - tmp0.s1;
    float out03 = tmp0.s1 - tmp0.s3;

    float out10 = tmp1.s0 - tmp1.s2;
    float out11 = tmp1.s1 + tmp1.s2;
    float out12 = tmp1.s2 - tmp1.s1;
    float out13 = tmp1.s1 - tmp1.s3;

    float out20 = tmp2.s0 - tmp2.s2;
    float out21 = tmp2.s1 + tmp2.s2;
    float out22 = tmp2.s2 - tmp2.s1;
    float out23 = tmp2.s1 - tmp2.s3;

    float out30 = tmp3.s0 - tmp3.s2;
    float out31 = tmp3.s1 + tmp3.s2;
    float out32 = tmp3.s2 - tmp3.s1;
    float out33 = tmp3.s1 - tmp3.s3;

    __global uchar *dst_addr = dst_ptr + dst_offset_first_element_in_bytes + z * dst_stride_x + (x + y * (int)NUM_TILES_X) * dst_stride_y;

    *((__global float *)(dst_addr + 0 * dst_stride_z))  = out00;
    *((__global float *)(dst_addr + 1 * dst_stride_z))  = out01;
    *((__global float *)(dst_addr + 2 * dst_stride_z))  = out02;
    *((__global float *)(dst_addr + 3 * dst_stride_z))  = out03;
    *((__global float *)(dst_addr + 4 * dst_stride_z))  = out10;
    *((__global float *)(dst_addr + 5 * dst_stride_z))  = out11;
    *((__global float *)(dst_addr + 6 * dst_stride_z))  = out12;
    *((__global float *)(dst_addr + 7 * dst_stride_z))  = out13;
    *((__global float *)(dst_addr + 8 * dst_stride_z))  = out20;
    *((__global float *)(dst_addr + 9 * dst_stride_z))  = out21;
    *((__global float *)(dst_addr + 10 * dst_stride_z)) = out22;
    *((__global float *)(dst_addr + 11 * dst_stride_z)) = out23;
    *((__global float *)(dst_addr + 12 * dst_stride_z)) = out30;
    *((__global float *)(dst_addr + 13 * dst_stride_z)) = out31;
    *((__global float *)(dst_addr + 14 * dst_stride_z)) = out32;
    *((__global float *)(dst_addr + 15 * dst_stride_z)) = out33;
}

/** This OpenCL kernel computes the input transform when the kernel size is 3x3, the output tile is 2x2 and the number of channels is multiple of 2
 *
 * @note The number of tiles in the x axis must be passed at compile time using -DNUM_TILES_X (i.e.-DNUM_TILES_X=5).
 * @note The pad left and pad top must be passed at compile time using -DPAD_LEFT and -DPAD_TOP (i.e.-DPAD_LEFT=1 and -DPAD_TOP=0).
 *
 * @param[in] src_ptr                           Pointer to the source image. Supported data types: F32
 * @param[in] src_stride_x                      Stride of the source image 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 image 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_offset_first_element_in_bytes The offset of the first element in the source image
 * @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 Y processed per workitem(in bytes)
 * @param[in] dst_ptr                           Pointer to the destination tensor. Supported data types: 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 destination tensor in Z dimension (in bytes)
 * @param[in] dst_step_z                        dst_stride_z * number of elements along Y processed per workitem(in bytes)
 * @param[in] dst_offset_first_element_in_bytes The offset of the first element in the destination tensor
 */
__kernel void winograd_input_transform_2x2_3x3_stepz2_nchw(
    TENSOR3D_DECLARATION(src),
    TENSOR3D_DECLARATION(dst))
{
    int x = get_global_id(0);
    int y = get_global_id(1);
    int z = get_global_id(2) * 2;

    // Compute input address
    __global uchar *src_addr = src_ptr + src_offset_first_element_in_bytes + x * 2 * src_stride_x + y * 2 * src_stride_y + z * src_stride_z;

    src_addr = src_addr - ((int)PAD_LEFT * src_stride_x) - ((int)PAD_TOP * src_stride_y);

    float4 in_row0 = vload4(0, (__global float *)(src_addr + 0 * src_stride_y));
    float4 in_row1 = vload4(0, (__global float *)(src_addr + 1 * src_stride_y));
    float4 in_row2 = vload4(0, (__global float *)(src_addr + 2 * src_stride_y));
    float4 in_row3 = vload4(0, (__global float *)(src_addr + 3 * src_stride_y));

    src_addr += src_stride_z;
    float4 in_row4 = vload4(0, (__global float *)(src_addr + 0 * src_stride_y));
    float4 in_row5 = vload4(0, (__global float *)(src_addr + 1 * src_stride_y));
    float4 in_row6 = vload4(0, (__global float *)(src_addr + 2 * src_stride_y));
    float4 in_row7 = vload4(0, (__global float *)(src_addr + 3 * src_stride_y));

    float4 tmp0 = in_row0 - in_row2;
    float4 tmp1 = in_row1 + in_row2;
    float4 tmp2 = in_row2 - in_row1;
    float4 tmp3 = in_row1 - in_row3;

    float4 tmp4 = in_row4 - in_row6;
    float4 tmp5 = in_row5 + in_row6;
    float4 tmp6 = in_row6 - in_row5;
    float4 tmp7 = in_row5 - in_row7;

    float2 out00 = (float2)(tmp0.s0 - tmp0.s2, tmp4.s0 - tmp4.s2);
    float2 out01 = (float2)(tmp0.s1 + tmp0.s2, tmp4.s1 + tmp4.s2);
    float2 out02 = (float2)(tmp0.s2 - tmp0.s1, tmp4.s2 - tmp4.s1);
    float2 out03 = (float2)(tmp0.s1 - tmp0.s3, tmp4.s1 - tmp4.s3);

    float2 out10 = (float2)(tmp1.s0 - tmp1.s2, tmp5.s0 - tmp5.s2);
    float2 out11 = (float2)(tmp1.s1 + tmp1.s2, tmp5.s1 + tmp5.s2);
    float2 out12 = (float2)(tmp1.s2 - tmp1.s1, tmp5.s2 - tmp5.s1);
    float2 out13 = (float2)(tmp1.s1 - tmp1.s3, tmp5.s1 - tmp5.s3);

    float2 out20 = (float2)(tmp2.s0 - tmp2.s2, tmp6.s0 - tmp6.s2);
    float2 out21 = (float2)(tmp2.s1 + tmp2.s2, tmp6.s1 + tmp6.s2);
    float2 out22 = (float2)(tmp2.s2 - tmp2.s1, tmp6.s2 - tmp6.s1);
    float2 out23 = (float2)(tmp2.s1 - tmp2.s3, tmp6.s1 - tmp6.s3);

    float2 out30 = (float2)(tmp3.s0 - tmp3.s2, tmp7.s0 - tmp7.s2);
    float2 out31 = (float2)(tmp3.s1 + tmp3.s2, tmp7.s1 + tmp7.s2);
    float2 out32 = (float2)(tmp3.s2 - tmp3.s1, tmp7.s2 - tmp7.s1);
    float2 out33 = (float2)(tmp3.s1 - tmp3.s3, tmp7.s1 - tmp7.s3);

    __global uchar *dst_addr = dst_ptr + dst_offset_first_element_in_bytes + z * dst_stride_x + (x + y * (int)NUM_TILES_X) * dst_stride_y;

    vstore2(out00, 0, (__global float *)(dst_addr + 0 * dst_stride_z));
    vstore2(out01, 0, (__global float *)(dst_addr + 1 * dst_stride_z));
    vstore2(out02, 0, (__global float *)(dst_addr + 2 * dst_stride_z));
    vstore2(out03, 0, (__global float *)(dst_addr + 3 * dst_stride_z));
    vstore2(out10, 0, (__global float *)(dst_addr + 4 * dst_stride_z));
    vstore2(out11, 0, (__global float *)(dst_addr + 5 * dst_stride_z));
    vstore2(out12, 0, (__global float *)(dst_addr + 6 * dst_stride_z));
    vstore2(out13, 0, (__global float *)(dst_addr + 7 * dst_stride_z));
    vstore2(out20, 0, (__global float *)(dst_addr + 8 * dst_stride_z));
    vstore2(out21, 0, (__global float *)(dst_addr + 9 * dst_stride_z));
    vstore2(out22, 0, (__global float *)(dst_addr + 10 * dst_stride_z));
    vstore2(out23, 0, (__global float *)(dst_addr + 11 * dst_stride_z));
    vstore2(out30, 0, (__global float *)(dst_addr + 12 * dst_stride_z));
    vstore2(out31, 0, (__global float *)(dst_addr + 13 * dst_stride_z));
    vstore2(out32, 0, (__global float *)(dst_addr + 14 * dst_stride_z));
    vstore2(out33, 0, (__global float *)(dst_addr + 15 * dst_stride_z));
}
#endif // defined(NUM_TILES_X) && defined(PAD_LEFT) && defined(PAD_TOP)

#if defined(NUM_TILES_X)
/** This OpenCL kernel performs Winograd output transform when the output tile is 2x2, the filter size 3x3 and the data format is NCHW
 *
 * @note The number of tiles along the X direction must be passed at compile time using -DNUM_TILES_X: e.g. -DNUM_TILES_X=16
 *
 * @param[in]  src_ptr                           Pointer to the source tensor. Supported data types: 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]  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]  dst_offset_first_element_in_bytes The offset of the first element in the destination tensor
 */
__kernel void winograd_output_transform_2x2_3x3_nchw(
    TENSOR3D_DECLARATION(src),
    TENSOR3D_DECLARATION(dst)
#if defined(HAS_BIAS)
    ,
    VECTOR_DECLARATION(bias)
#endif // defined(HAS_BIAS)
)
{
    // Each thread stores a 2x2 tile
    Tensor3D src = CONVERT_TO_TENSOR3D_STRUCT(src);

    const __global uchar *src_addr = tensor3D_offset(&src, 0, 0, 0);

    // Load the values across the 16 channels to compose the 4x4 tile
    float d00 = *((__global float *)(src_addr + 0 * src_stride_z));
    float d01 = *((__global float *)(src_addr + 1 * src_stride_z));
    float d02 = *((__global float *)(src_addr + 2 * src_stride_z));
    float d03 = *((__global float *)(src_addr + 3 * src_stride_z));

    float d10 = *((__global float *)(src_addr + 4 * src_stride_z));
    float d11 = *((__global float *)(src_addr + 5 * src_stride_z));
    float d12 = *((__global float *)(src_addr + 6 * src_stride_z));
    float d13 = *((__global float *)(src_addr + 7 * src_stride_z));

    float d20 = *((__global float *)(src_addr + 8 * src_stride_z));
    float d21 = *((__global float *)(src_addr + 9 * src_stride_z));
    float d22 = *((__global float *)(src_addr + 10 * src_stride_z));
    float d23 = *((__global float *)(src_addr + 11 * src_stride_z));

    float d30 = *((__global float *)(src_addr + 12 * src_stride_z));
    float d31 = *((__global float *)(src_addr + 13 * src_stride_z));
    float d32 = *((__global float *)(src_addr + 14 * src_stride_z));
    float d33 = *((__global float *)(src_addr + 15 * src_stride_z));

    // Compute the 2x2 output tile
    float k0 = d01 + d11 + d21;
    float k1 = d02 + d12 + d22;
    float k2 = d11 - d21 - d31;
    float k3 = d12 - d22 - d32;

    // out00 = d00 + d10 + d20 + d01 + d11 + d21 + d02 + d12 + d22
    // out01 = d01 + d11 + d21 - (d02 + d12 + d22) - (d03 + d13 + d23)
    // out10 = d10 - d20 - d30 + (d11 - d21 - d31) + (d12 - d22 - d32)
    // out11 = d11 - d21 - d31 - (d12 - d22 - d32) - (d13 - d23 - d33)

    float out00 = d10;
    float out01 = -d13;
    float out10 = d10;
    float out11 = -d13;

    out00 += d00 + d20 + k0 + k1;
    out01 += k0 - k1 - (d03 + d23);
    out10 += -d20 - d30 + k2 + k3;
    out11 += k2 - k3 + d23 + d33;

    int y_in  = get_global_id(1);
    int x_out = (y_in % NUM_TILES_X) * 2;
    int y_out = (y_in / NUM_TILES_X) * 2;
    int z_out = get_global_id(0);

#if defined(HAS_BIAS)
    // Add bias
    Vector bias = CONVERT_TO_VECTOR_STRUCT_NO_STEP(bias);

    float b = (float) * ((__global float *)(vector_offset(&bias, z_out)));

    out00 += (float)b;
    out01 += (float)b;
    out10 += (float)b;
    out11 += (float)b;
#endif // defined(HAS_BIAS)

    // Get output address
    __global uchar *dst_addr = dst_ptr + dst_offset_first_element_in_bytes + x_out * dst_stride_x + y_out * dst_stride_y + z_out * dst_stride_z;

    // Store the 2x2 output tile
    vstore2((float2)(out00, out01), 0, (__global float *)(dst_addr + 0 * dst_stride_y));
    vstore2((float2)(out10, out11), 0, (__global float *)(dst_addr + 1 * dst_stride_y));
}
#endif // defined(NUM_TILES_X)