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_SPLIT_MERGE_DETAIL_HPP__ 47 #define __OPENCV_CUDEV_GRID_SPLIT_MERGE_DETAIL_HPP__ 48 49 #include "../../common.hpp" 50 #include "../../util/saturate_cast.hpp" 51 #include "../../util/tuple.hpp" 52 #include "../../util/vec_traits.hpp" 53 #include "../../ptr2d/glob.hpp" 54 #include "../../ptr2d/traits.hpp" 55 56 namespace cv { namespace cudev { 57 58 namespace grid_split_merge_detail 59 { 60 // merge 61 62 template <class Src1Ptr, class Src2Ptr, typename DstType, class MaskPtr> 63 __global__ void mergeC2(const Src1Ptr src1, const Src2Ptr src2, GlobPtr<DstType> dst, const MaskPtr mask, const int rows, const int cols) 64 { 65 typedef typename VecTraits<DstType>::elem_type dst_elem_type; 66 67 const int x = blockIdx.x * blockDim.x + threadIdx.x; 68 const int y = blockIdx.y * blockDim.y + threadIdx.y; 69 70 if (x >= cols || y >= rows || !mask(y, x)) 71 return; 72 73 dst(y, x) = VecTraits<DstType>::make( 74 saturate_cast<dst_elem_type>(src1(y, x)), 75 saturate_cast<dst_elem_type>(src2(y, x)) 76 ); 77 } 78 79 template <class Policy, class Src1Ptr, class Src2Ptr, typename DstType, class MaskPtr> 80 __host__ void mergeC2(const Src1Ptr& src1, const Src2Ptr& src2, const GlobPtr<DstType>& dst, const MaskPtr& mask, int rows, int cols, cudaStream_t stream) 81 { 82 const dim3 block(Policy::block_size_x, Policy::block_size_y); 83 const dim3 grid(divUp(cols, block.x), divUp(rows, block.y)); 84 85 mergeC2<<<grid, block, 0, stream>>>(src1, src2, dst, mask, rows, cols); 86 CV_CUDEV_SAFE_CALL( cudaGetLastError() ); 87 88 if (stream == 0) 89 CV_CUDEV_SAFE_CALL(cudaDeviceSynchronize()); 90 } 91 92 template <class Src1Ptr, class Src2Ptr, class Src3Ptr, typename DstType, class MaskPtr> 93 __global__ void mergeC3(const Src1Ptr src1, const Src2Ptr src2, const Src3Ptr src3, GlobPtr<DstType> dst, const MaskPtr mask, const int rows, const int cols) 94 { 95 typedef typename VecTraits<DstType>::elem_type dst_elem_type; 96 97 const int x = blockIdx.x * blockDim.x + threadIdx.x; 98 const int y = blockIdx.y * blockDim.y + threadIdx.y; 99 100 if (x >= cols || y >= rows || !mask(y, x)) 101 return; 102 103 dst(y, x) = VecTraits<DstType>::make( 104 saturate_cast<dst_elem_type>(src1(y, x)), 105 saturate_cast<dst_elem_type>(src2(y, x)), 106 saturate_cast<dst_elem_type>(src3(y, x)) 107 ); 108 } 109 110 template <class Policy, class Src1Ptr, class Src2Ptr, class Src3Ptr, typename DstType, class MaskPtr> 111 __host__ void mergeC3(const Src1Ptr& src1, const Src2Ptr& src2, const Src3Ptr& src3, const GlobPtr<DstType>& dst, const MaskPtr& mask, int rows, int cols, cudaStream_t stream) 112 { 113 const dim3 block(Policy::block_size_x, Policy::block_size_y); 114 const dim3 grid(divUp(cols, block.x), divUp(rows, block.y)); 115 116 mergeC3<<<grid, block, 0, stream>>>(src1, src2, src3, dst, mask, rows, cols); 117 CV_CUDEV_SAFE_CALL( cudaGetLastError() ); 118 119 if (stream == 0) 120 CV_CUDEV_SAFE_CALL(cudaDeviceSynchronize()); 121 } 122 123 template <class Src1Ptr, class Src2Ptr, class Src3Ptr, class Src4Ptr, typename DstType, class MaskPtr> 124 __global__ void mergeC4(const Src1Ptr src1, const Src2Ptr src2, const Src3Ptr src3, const Src4Ptr src4, GlobPtr<DstType> dst, const MaskPtr mask, const int rows, const int cols) 125 { 126 typedef typename VecTraits<DstType>::elem_type dst_elem_type; 127 128 const int x = blockIdx.x * blockDim.x + threadIdx.x; 129 const int y = blockIdx.y * blockDim.y + threadIdx.y; 130 131 if (x >= cols || y >= rows || !mask(y, x)) 132 return; 133 134 dst(y, x) = VecTraits<DstType>::make( 135 saturate_cast<dst_elem_type>(src1(y, x)), 136 saturate_cast<dst_elem_type>(src2(y, x)), 137 saturate_cast<dst_elem_type>(src3(y, x)), 138 saturate_cast<dst_elem_type>(src4(y, x)) 139 ); 140 } 141 142 template <class Policy, class Src1Ptr, class Src2Ptr, class Src3Ptr, class Src4Ptr, typename DstType, class MaskPtr> 143 __host__ void mergeC4(const Src1Ptr& src1, const Src2Ptr& src2, const Src3Ptr& src3, const Src4Ptr& src4, const GlobPtr<DstType>& dst, const MaskPtr& mask, int rows, int cols, cudaStream_t stream) 144 { 145 const dim3 block(Policy::block_size_x, Policy::block_size_y); 146 const dim3 grid(divUp(cols, block.x), divUp(rows, block.y)); 147 148 mergeC4<<<grid, block, 0, stream>>>(src1, src2, src3, src4, dst, mask, rows, cols); 149 CV_CUDEV_SAFE_CALL( cudaGetLastError() ); 150 151 if (stream == 0) 152 CV_CUDEV_SAFE_CALL(cudaDeviceSynchronize()); 153 } 154 155 template <int cn, class Policy> struct MergeImpl; 156 157 template <class Policy> struct MergeImpl<2, Policy> 158 { 159 template <class SrcPtrTuple, typename DstType, class MaskPtr> 160 __host__ static void merge(const SrcPtrTuple& src, const GlobPtr<DstType>& dst, const MaskPtr& mask, int rows, int cols, cudaStream_t stream) 161 { 162 mergeC2<Policy>(get<0>(src), get<1>(src), dst, mask, rows, cols, stream); 163 } 164 }; 165 166 template <class Policy> struct MergeImpl<3, Policy> 167 { 168 template <class SrcPtrTuple, typename DstType, class MaskPtr> 169 __host__ static void merge(const SrcPtrTuple& src, const GlobPtr<DstType>& dst, const MaskPtr& mask, int rows, int cols, cudaStream_t stream) 170 { 171 mergeC3<Policy>(get<0>(src), get<1>(src), get<2>(src), dst, mask, rows, cols, stream); 172 } 173 }; 174 175 template <class Policy> struct MergeImpl<4, Policy> 176 { 177 template <class SrcPtrTuple, typename DstType, class MaskPtr> 178 __host__ static void merge(const SrcPtrTuple& src, const GlobPtr<DstType>& dst, const MaskPtr& mask, int rows, int cols, cudaStream_t stream) 179 { 180 mergeC4<Policy>(get<0>(src), get<1>(src), get<2>(src), get<3>(src), dst, mask, rows, cols, stream); 181 } 182 }; 183 184 // split 185 186 template <class SrcPtr, typename DstType, class MaskPtr> 187 __global__ void split(const SrcPtr src, GlobPtr<DstType> dst1, GlobPtr<DstType> dst2, const MaskPtr mask, const int rows, const int cols) 188 { 189 typedef typename PtrTraits<SrcPtr>::value_type src_type; 190 191 const int x = blockIdx.x * blockDim.x + threadIdx.x; 192 const int y = blockIdx.y * blockDim.y + threadIdx.y; 193 194 if (x >= cols || y >= rows || !mask(y, x)) 195 return; 196 197 const src_type src_value = src(y, x); 198 199 dst1(y, x) = src_value.x; 200 dst2(y, x) = src_value.y; 201 } 202 203 template <class Policy, class SrcPtr, typename DstType, class MaskPtr> 204 __host__ void split(const SrcPtr& src, const GlobPtr<DstType>& dst1, const GlobPtr<DstType>& dst2, const MaskPtr& mask, int rows, int cols, cudaStream_t stream) 205 { 206 const dim3 block(Policy::block_size_x, Policy::block_size_y); 207 const dim3 grid(divUp(cols, block.x), divUp(rows, block.y)); 208 209 split<<<grid, block, 0, stream>>>(src, dst1, dst2, mask, rows, cols); 210 CV_CUDEV_SAFE_CALL( cudaGetLastError() ); 211 212 if (stream == 0) 213 CV_CUDEV_SAFE_CALL(cudaDeviceSynchronize()); 214 } 215 216 template <class SrcPtr, typename DstType, class MaskPtr> 217 __global__ void split(const SrcPtr src, GlobPtr<DstType> dst1, GlobPtr<DstType> dst2, GlobPtr<DstType> dst3, const MaskPtr mask, const int rows, const int cols) 218 { 219 typedef typename PtrTraits<SrcPtr>::value_type src_type; 220 221 const int x = blockIdx.x * blockDim.x + threadIdx.x; 222 const int y = blockIdx.y * blockDim.y + threadIdx.y; 223 224 if (x >= cols || y >= rows || !mask(y, x)) 225 return; 226 227 const src_type src_value = src(y, x); 228 229 dst1(y, x) = src_value.x; 230 dst2(y, x) = src_value.y; 231 dst3(y, x) = src_value.z; 232 } 233 234 template <class Policy, class SrcPtr, typename DstType, class MaskPtr> 235 __host__ void split(const SrcPtr& src, const GlobPtr<DstType>& dst1, const GlobPtr<DstType>& dst2, const GlobPtr<DstType>& dst3, const MaskPtr& mask, int rows, int cols, cudaStream_t stream) 236 { 237 const dim3 block(Policy::block_size_x, Policy::block_size_y); 238 const dim3 grid(divUp(cols, block.x), divUp(rows, block.y)); 239 240 split<<<grid, block, 0, stream>>>(src, dst1, dst2, dst3, mask, rows, cols); 241 CV_CUDEV_SAFE_CALL( cudaGetLastError() ); 242 243 if (stream == 0) 244 CV_CUDEV_SAFE_CALL(cudaDeviceSynchronize()); 245 } 246 247 template <class SrcPtr, typename DstType, class MaskPtr> 248 __global__ void split(const SrcPtr src, GlobPtr<DstType> dst1, GlobPtr<DstType> dst2, GlobPtr<DstType> dst3, GlobPtr<DstType> dst4, const MaskPtr mask, const int rows, const int cols) 249 { 250 typedef typename PtrTraits<SrcPtr>::value_type src_type; 251 252 const int x = blockIdx.x * blockDim.x + threadIdx.x; 253 const int y = blockIdx.y * blockDim.y + threadIdx.y; 254 255 if (x >= cols || y >= rows || !mask(y, x)) 256 return; 257 258 const src_type src_value = src(y, x); 259 260 dst1(y, x) = src_value.x; 261 dst2(y, x) = src_value.y; 262 dst3(y, x) = src_value.z; 263 dst4(y, x) = src_value.w; 264 } 265 266 template <class Policy, class SrcPtr, typename DstType, class MaskPtr> 267 __host__ void split(const SrcPtr& src, const GlobPtr<DstType>& dst1, const GlobPtr<DstType>& dst2, const GlobPtr<DstType>& dst3, const GlobPtr<DstType>& dst4, const MaskPtr& mask, int rows, int cols, cudaStream_t stream) 268 { 269 const dim3 block(Policy::block_size_x, Policy::block_size_y); 270 const dim3 grid(divUp(cols, block.x), divUp(rows, block.y)); 271 272 split<<<grid, block, 0, stream>>>(src, dst1, dst2, dst3, dst4, mask, rows, cols); 273 CV_CUDEV_SAFE_CALL( cudaGetLastError() ); 274 275 if (stream == 0) 276 CV_CUDEV_SAFE_CALL(cudaDeviceSynchronize()); 277 } 278 } 279 280 }} 281 282 #endif 283
