1 /*M///////////////////////////////////////////////////////////////////////////////////////
2 //
3 //  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
4 //
5 //  By downloading, copying, installing or using the software you agree to this license.
6 //  If you do not agree to this license, do not download, install,
7 //  copy or use the software.
8 //
9 //
10 //                          License Agreement
11 //                For Open Source Computer Vision Library
12 //
13 // Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
14 // Copyright (C) 2009, Willow Garage Inc., all rights reserved.
15 // Copyright (C) 2013, OpenCV Foundation, all rights reserved.
16 // Third party copyrights are property of their respective owners.
17 //
18 // Redistribution and use in source and binary forms, with or without modification,
19 // are permitted provided that the following conditions are met:
20 //
21 //   * Redistribution's of source code must retain the above copyright notice,
22 //     this list of conditions and the following disclaimer.
23 //
24 //   * Redistribution's in binary form must reproduce the above copyright notice,
25 //     this list of conditions and the following disclaimer in the documentation
26 //     and/or other materials provided with the distribution.
27 //
28 //   * The name of the copyright holders may not be used to endorse or promote products
29 //     derived from this software without specific prior written permission.
30 //
31 // This software is provided by the copyright holders and contributors "as is" and
32 // any express or implied warranties, including, but not limited to, the implied
33 // warranties of merchantability and fitness for a particular purpose are disclaimed.
34 // In no event shall the Intel Corporation or contributors be liable for any direct,
35 // indirect, incidental, special, exemplary, or consequential damages
36 // (including, but not limited to, procurement of substitute goods or services;
37 // loss of use, data, or profits; or business interruption) however caused
38 // and on any theory of liability, whether in contract, strict liability,
39 // or tort (including negligence or otherwise) arising in any way out of
40 // the use of this software, even if advised of the possibility of such damage.
41 //
42 //M*/
43 
44 #pragma once
45 
46 #ifndef __OPENCV_CUDEV_GRID_REDUCE_TO_COLUMN_DETAIL_HPP__
47 #define __OPENCV_CUDEV_GRID_REDUCE_TO_COLUMN_DETAIL_HPP__
48 
49 #include "../../common.hpp"
50 #include "../../util/saturate_cast.hpp"
51 #include "../../block/reduce.hpp"
52 
53 namespace cv { namespace cudev {
54 
55 namespace grid_reduce_to_vec_detail
56 {
57     template <int BLOCK_SIZE, typename work_type, typename work_elem_type, class Reductor, int cn> struct Reduce;
58 
59     template <int BLOCK_SIZE, typename work_type, typename work_elem_type, class Reductor> struct Reduce<BLOCK_SIZE, work_type, work_elem_type, Reductor, 1>
60     {
callcv::cudev::grid_reduce_to_vec_detail::Reduce61         __device__ __forceinline__ static void call(work_elem_type smem[1][BLOCK_SIZE], work_type& myVal)
62         {
63             typename Reductor::template rebind<work_elem_type>::other op;
64             blockReduce<BLOCK_SIZE>(smem[0], myVal, threadIdx.x, op);
65         }
66     };
67 
68     template <int BLOCK_SIZE, typename work_type, typename work_elem_type, class Reductor> struct Reduce<BLOCK_SIZE, work_type, work_elem_type, Reductor, 2>
69     {
callcv::cudev::grid_reduce_to_vec_detail::Reduce70         __device__ __forceinline__ static void call(work_elem_type smem[2][BLOCK_SIZE], work_type& myVal)
71         {
72             typename Reductor::template rebind<work_elem_type>::other op;
73             blockReduce<BLOCK_SIZE>(smem_tuple(smem[0], smem[1]), tie(myVal.x, myVal.y), threadIdx.x, make_tuple(op, op));
74         }
75     };
76 
77     template <int BLOCK_SIZE, typename work_type, typename work_elem_type, class Reductor> struct Reduce<BLOCK_SIZE, work_type, work_elem_type, Reductor, 3>
78     {
callcv::cudev::grid_reduce_to_vec_detail::Reduce79         __device__ __forceinline__ static void call(work_elem_type smem[3][BLOCK_SIZE], work_type& myVal)
80         {
81             typename Reductor::template rebind<work_elem_type>::other op;
82             blockReduce<BLOCK_SIZE>(smem_tuple(smem[0], smem[1], smem[2]), tie(myVal.x, myVal.y, myVal.z), threadIdx.x, make_tuple(op, op, op));
83         }
84     };
85 
86     template <int BLOCK_SIZE, typename work_type, typename work_elem_type, class Reductor> struct Reduce<BLOCK_SIZE, work_type, work_elem_type, Reductor, 4>
87     {
callcv::cudev::grid_reduce_to_vec_detail::Reduce88         __device__ __forceinline__ static void call(work_elem_type smem[4][BLOCK_SIZE], work_type& myVal)
89         {
90             typename Reductor::template rebind<work_elem_type>::other op;
91             blockReduce<BLOCK_SIZE>(smem_tuple(smem[0], smem[1], smem[2], smem[3]), tie(myVal.x, myVal.y, myVal.z, myVal.w), threadIdx.x, make_tuple(op, op, op, op));
92         }
93     };
94 
95     template <class Reductor, int BLOCK_SIZE, class SrcPtr, typename ResType, class MaskPtr>
reduceToColumn(const SrcPtr src,ResType * dst,const MaskPtr mask,const int cols)96     __global__ void reduceToColumn(const SrcPtr src, ResType* dst, const MaskPtr mask, const int cols)
97     {
98         typedef typename Reductor::work_type work_type;
99         typedef typename VecTraits<work_type>::elem_type work_elem_type;
100         const int cn = VecTraits<work_type>::cn;
101 
102         __shared__ work_elem_type smem[cn][BLOCK_SIZE];
103 
104         const int y = blockIdx.x;
105 
106         work_type myVal = Reductor::initialValue();
107 
108         Reductor op;
109 
110         for (int x = threadIdx.x; x < cols; x += BLOCK_SIZE)
111         {
112             if (mask(y, x))
113             {
114                 myVal = op(myVal, saturate_cast<work_type>(src(y, x)));
115             }
116         }
117 
118         Reduce<BLOCK_SIZE, work_type, work_elem_type, Reductor, cn>::call(smem, myVal);
119 
120         if (threadIdx.x == 0)
121             dst[y] = saturate_cast<ResType>(Reductor::result(myVal, cols));
122     }
123 
124     template <class Reductor, class Policy, class SrcPtr, typename ResType, class MaskPtr>
reduceToColumn(const SrcPtr & src,ResType * dst,const MaskPtr & mask,int rows,int cols,cudaStream_t stream)125     __host__ void reduceToColumn(const SrcPtr& src, ResType* dst, const MaskPtr& mask, int rows, int cols, cudaStream_t stream)
126     {
127         const int BLOCK_SIZE_X = Policy::block_size_x;
128         const int BLOCK_SIZE_Y = Policy::block_size_y;
129 
130         const int BLOCK_SIZE = BLOCK_SIZE_X * BLOCK_SIZE_Y;
131 
132         const dim3 block(BLOCK_SIZE);
133         const dim3 grid(rows);
134 
135         reduceToColumn<Reductor, BLOCK_SIZE><<<grid, block, 0, stream>>>(src, dst, mask, cols);
136         CV_CUDEV_SAFE_CALL( cudaGetLastError() );
137 
138         if (stream == 0)
139             CV_CUDEV_SAFE_CALL( cudaDeviceSynchronize() );
140 
141     }
142 }
143 
144 }}
145 
146 #endif
147