/external/opencv3/modules/core/src/opencl/ |
D | fft.cl | 34 void butterfly2(CT a0, CT a1, __local CT* smem, __global const CT* twiddles, 41 smem[dst_ind] = a0 + a1; 42 smem[dst_ind+block_size] = a0 - a1; 46 void butterfly4(CT a0, CT a1, CT a2, CT a3, __local CT* smem, __global const CT* twiddles, 61 smem[dst_ind] = b0 + b1; 62 smem[dst_ind + block_size] = a2 + a3; 63 smem[dst_ind + 2*block_size] = b0 - b1; 64 smem[dst_ind + 3*block_size] = a2 - a3; 68 void butterfly3(CT a0, CT a1, CT a2, __local CT* smem, __global const CT* twiddles, 80 smem[dst_ind] = a0 + b1; [all …]
|
/external/opencv3/modules/cudev/include/opencv2/cudev/warp/ |
D | scan.hpp | 59 __device__ T warpScanInclusive(T data, volatile T* smem, uint tid) in warpScanInclusive() argument 62 (void) smem; in warpScanInclusive() 79 smem[pos] = 0; in warpScanInclusive() 82 smem[pos] = data; in warpScanInclusive() 84 smem[pos] += smem[pos - 1]; in warpScanInclusive() 85 smem[pos] += smem[pos - 2]; in warpScanInclusive() 86 smem[pos] += smem[pos - 4]; in warpScanInclusive() 87 smem[pos] += smem[pos - 8]; in warpScanInclusive() 88 smem[pos] += smem[pos - 16]; in warpScanInclusive() 90 return smem[pos]; in warpScanInclusive() [all …]
|
/external/opencv3/modules/objdetect/src/opencl/ |
D | objdetect_hog.cl | 72 __global float* block_hists, __local float* smem) 87 __local float* hists = smem + lp * cnbins * (CELLS_PER_BLOCK_X * 172 __local float* smem = squares + boffset; 173 float sum = smem[hid]; 175 smem[hid] = sum = sum + smem[hid + 18]; 178 smem[hid] = sum = sum + smem[hid + 9]; 181 smem[hid] = sum + smem[hid + 4]; 183 sum = smem[0] + smem[1] + smem[2] + smem[3] + smem[8]; 192 sum = smem[hid]; 194 smem[hid] = sum = sum + smem[hid + 18]; [all …]
|
/external/opencv3/modules/core/include/opencv2/core/cuda/detail/ |
D | reduce.hpp | 74 … static __device__ void loadToSmem(const PointerTuple& smem, const ValTuple& val, unsigned int tid) in loadToSmem() 76 thrust::get<I>(smem)[tid] = thrust::get<I>(val); in loadToSmem() local 78 For<I + 1, N>::loadToSmem(smem, val, tid); in loadToSmem() 81 …static __device__ void loadFromSmem(const PointerTuple& smem, const ValTuple& val, unsigned int ti… in loadFromSmem() 83 thrust::get<I>(val) = thrust::get<I>(smem)[tid]; in loadFromSmem() 85 For<I + 1, N>::loadFromSmem(smem, val, tid); in loadFromSmem() 89 …static __device__ void merge(const PointerTuple& smem, const ValTuple& val, unsigned int tid, unsi… in merge() 91 …pename thrust::tuple_element<I, PointerTuple>::type>::type reg = thrust::get<I>(smem)[tid + delta]; in merge() 92 … thrust::get<I>(smem)[tid] = thrust::get<I>(val) = thrust::get<I>(op)(thrust::get<I>(val), reg); in merge() local 94 For<I + 1, N>::merge(smem, val, tid, delta, op); in merge() [all …]
|
/external/opencv3/modules/cudev/include/opencv2/cudev/block/detail/ |
D | reduce.hpp | 83 __device__ static void loadToSmem(const PointerTuple& smem, const ValTuple& val, uint tid) in loadToSmem() 85 get<I>(smem)[tid] = get<I>(val); in loadToSmem() local 87 For<I + 1, N>::loadToSmem(smem, val, tid); in loadToSmem() 91 __device__ static void loadFromSmem(const PointerTuple& smem, const ValTuple& val, uint tid) in loadFromSmem() 93 get<I>(val) = get<I>(smem)[tid]; in loadFromSmem() 95 For<I + 1, N>::loadFromSmem(smem, val, tid); in loadFromSmem() 99 …__device__ static void merge(const PointerTuple& smem, const ValTuple& val, uint tid, uint delta, … in merge() 101 …typename GetType<typename tuple_element<I, PointerTuple>::type>::type reg = get<I>(smem)[tid + del… in merge() 102 get<I>(smem)[tid] = get<I>(val) = get<I>(op)(get<I>(val), reg); in merge() local 104 For<I + 1, N>::merge(smem, val, tid, delta, op); in merge() [all …]
|
D | reduce_key_val.hpp | 82 … __device__ static void loadToSmem(const PointerTuple& smem, const ReferenceTuple& data, uint tid) in loadToSmem() 84 get<I>(smem)[tid] = get<I>(data); in loadToSmem() local 86 For<I + 1, N>::loadToSmem(smem, data, tid); in loadToSmem() 90 …__device__ static void loadFromSmem(const PointerTuple& smem, const ReferenceTuple& data, uint tid) in loadFromSmem() 92 get<I>(data) = get<I>(smem)[tid]; in loadFromSmem() 94 For<I + 1, N>::loadFromSmem(smem, data, tid); in loadFromSmem() 152 __device__ __forceinline__ void loadToSmem(volatile T* smem, T& data, uint tid) in loadToSmem() argument 154 smem[tid] = data; in loadToSmem() 158 __device__ __forceinline__ void loadFromSmem(volatile T* smem, T& data, uint tid) in loadFromSmem() argument 160 data = smem[tid]; in loadFromSmem() [all …]
|
/external/opencv3/modules/imgproc/src/opencl/ |
D | clahe.cl | 50 inline int calc_lut(__local int* smem, int val, int tid) 52 smem[tid] = val; 57 smem[i] += smem[i - 1]; 60 return smem[tid]; 64 inline void reduce(volatile __local int* smem, int val, int tid) 66 smem[tid] = val; 70 smem[tid] = val += smem[tid + 128]; 74 smem[tid] = val += smem[tid + 64]; 78 smem[tid] += smem[tid + 32]; 82 smem[tid] += smem[tid + 16]; [all …]
|
D | pyr_down.cl | 104 smem[0][col_lcl] = sum0; \ 107 smem[1][col_lcl] = sum1; 123 vstore4(sum40, col_lcl, (__local float*) &smem[0][2]); \ 126 vstore4(sum41, col_lcl, (__local float*) &smem[1][2]); 137 __local FT smem[2][LOCAL_SIZE + 4]; 229 …FT sum = dot(vload4(0, (__local float*) (&smem) + tid2 + (yin - y) * (LOCAL_SIZE + 4)), (float4)(c… 231 …FT sum = dot(vload4(0, (__local double*) (&smem) + tid2 + (yin - y) * (LOCAL_SIZE + 4)), (double4)… 234 FT sum = co3 * smem[yin - y][2 + tid2 - 2]; 235 sum = MAD(co2, smem[yin - y][2 + tid2 - 1], sum); 236 sum = MAD(co1, smem[yin - y][2 + tid2 ], sum); [all …]
|
D | canny.cl | 80 inline float3 sobel(int idx, __local const floatN *smem) 85 floatN dx = fma(2, smem[idx + GRP_SIZEX + 6] - smem[idx + GRP_SIZEX + 4], 86 smem[idx + 2] - smem[idx] + smem[idx + 2 * GRP_SIZEX + 10] - smem[idx + 2 * GRP_SIZEX + 8]); 88 floatN dy = fma(2, smem[idx + 1] - smem[idx + 2 * GRP_SIZEX + 9], 89 smem[idx + 2] - smem[idx + 2 * GRP_SIZEX + 10] + smem[idx] - smem[idx + 2 * GRP_SIZEX + 8]); 123 __local floatN smem[(GRP_SIZEX + 4) * (GRP_SIZEY + 4)]; 136 smem[j] = loadpix(src + mad24(y, src_step, mad24(x, cn * (int)sizeof(TYPE), src_offset))); 152 mag[i] = (sobel(i, smem)).z; 153 mag[i + grp_sizey * (GRP_SIZEX + 2)] = (sobel(i + grp_sizey * (GRP_SIZEX + 4), smem)).z; 158 mag[i * (GRP_SIZEX + 2)] = (sobel(i * (GRP_SIZEX + 4), smem)).z; [all …]
|
/external/opencv3/modules/cudev/include/opencv2/cudev/block/ |
D | vec_distance.hpp | 75 __device__ __forceinline__ void reduceWarp(result_type* smem, uint tid) in reduceWarp() 77 warpReduce(smem, mySum, tid, plus<result_type>()); in reduceWarp() 80 … template <int THREAD_DIM> __device__ __forceinline__ void reduceBlock(result_type* smem, uint tid) in reduceBlock() 82 blockReduce<THREAD_DIM>(smem, mySum, tid, plus<result_type>()); in reduceBlock() 104 __device__ __forceinline__ void reduceWarp(result_type* smem, uint tid) in reduceWarp() 106 warpReduce(smem, mySum, tid, plus<result_type>()); in reduceWarp() 109 … template <int THREAD_DIM> __device__ __forceinline__ void reduceBlock(result_type* smem, uint tid) in reduceBlock() 111 blockReduce<THREAD_DIM>(smem, mySum, tid, plus<result_type>()); in reduceBlock() 137 __device__ __forceinline__ void reduceWarp(result_type* smem, uint tid) in reduceWarp() 139 warpReduce(smem, mySum, tid, plus<result_type>()); in reduceWarp() [all …]
|
D | scan.hpp | 58 __device__ T blockScanInclusive(T data, volatile T* smem, uint tid) in blockScanInclusive() argument 63 T warpResult = warpScanInclusive(data, smem, tid); in blockScanInclusive() 71 smem[tid >> LOG_WARP_SIZE] = warpResult; in blockScanInclusive() 79 T val = smem[tid]; in blockScanInclusive() 82 smem[tid] = warpScanExclusive(val, smem, tid); in blockScanInclusive() 88 return warpResult + smem[tid >> LOG_WARP_SIZE]; in blockScanInclusive() 92 return warpScanInclusive(data, smem, tid); in blockScanInclusive() 97 __device__ __forceinline__ T blockScanExclusive(T data, volatile T* smem, uint tid) in blockScanExclusive() argument 99 return blockScanInclusive<THREADS_NUM>(data, smem, tid) - data; in blockScanExclusive()
|
/external/opencv3/modules/core/include/opencv2/core/cuda/ |
D | vec_distance.hpp | 70 template <int THREAD_DIM> __device__ __forceinline__ void reduceAll(int* smem, int tid) in reduceAll() 72 reduce<THREAD_DIM>(smem, mySum, tid, plus<int>()); in reduceAll() 94 template <int THREAD_DIM> __device__ __forceinline__ void reduceAll(float* smem, int tid) in reduceAll() 96 reduce<THREAD_DIM>(smem, mySum, tid, plus<float>()); in reduceAll() 120 template <int THREAD_DIM> __device__ __forceinline__ void reduceAll(float* smem, int tid) in reduceAll() 122 reduce<THREAD_DIM>(smem, mySum, tid, plus<float>()); in reduceAll() 145 template <int THREAD_DIM> __device__ __forceinline__ void reduceAll(int* smem, int tid) in reduceAll() 147 reduce<THREAD_DIM>(smem, mySum, tid, plus<int>()); in reduceAll() 160 …bal(const T1* vec1, const T2* vec2, int len, Dist& dist, typename Dist::result_type* smem, int tid) in calcVecDiffGlobal() argument 173 dist.reduceAll<THREAD_DIM>(smem, tid); in calcVecDiffGlobal() [all …]
|
/external/opencv3/modules/cudev/include/opencv2/cudev/warp/detail/ |
D | reduce.hpp | 81 __device__ static void loadToSmem(const PointerTuple& smem, const ValTuple& val, uint tid) in loadToSmem() 83 get<I>(smem)[tid] = get<I>(val); in loadToSmem() local 85 For<I + 1, N>::loadToSmem(smem, val, tid); in loadToSmem() 89 …__device__ static void merge(const PointerTuple& smem, const ValTuple& val, uint tid, uint delta, … in merge() 91 …typename GetType<typename tuple_element<I, PointerTuple>::type>::type reg = get<I>(smem)[tid + del… in merge() 92 get<I>(smem)[tid] = get<I>(val) = get<I>(op)(get<I>(val), reg); in merge() local 94 For<I + 1, N>::merge(smem, val, tid, delta, op); in merge() 132 __device__ __forceinline__ void loadToSmem(volatile T* smem, T& val, uint tid) in loadToSmem() argument 134 smem[tid] = val; in loadToSmem() 139 …device__ __forceinline__ void loadToSmem(const tuple<P0, P1, P2, P3, P4, P5, P6, P7, P8, P9>& smem, in loadToSmem() argument [all …]
|
/external/opencv3/modules/cudev/include/opencv2/cudev/grid/detail/ |
D | reduce_to_column.hpp | 61 … __device__ __forceinline__ static void call(work_elem_type smem[1][BLOCK_SIZE], work_type& myVal) in call() 64 blockReduce<BLOCK_SIZE>(smem[0], myVal, threadIdx.x, op); in call() 70 … __device__ __forceinline__ static void call(work_elem_type smem[2][BLOCK_SIZE], work_type& myVal) in call() 73 …blockReduce<BLOCK_SIZE>(smem_tuple(smem[0], smem[1]), tie(myVal.x, myVal.y), threadIdx.x, make_tup… in call() 79 … __device__ __forceinline__ static void call(work_elem_type smem[3][BLOCK_SIZE], work_type& myVal) in call() 82 …blockReduce<BLOCK_SIZE>(smem_tuple(smem[0], smem[1], smem[2]), tie(myVal.x, myVal.y, myVal.z), thr… in call() 88 … __device__ __forceinline__ static void call(work_elem_type smem[4][BLOCK_SIZE], work_type& myVal) in call() 91 …blockReduce<BLOCK_SIZE>(smem_tuple(smem[0], smem[1], smem[2], smem[3]), tie(myVal.x, myVal.y, myVa… in call() 102 __shared__ work_elem_type smem[cn][BLOCK_SIZE]; in reduceToColumn() local 118 Reduce<BLOCK_SIZE, work_type, work_elem_type, Reductor, cn>::call(smem, myVal); in reduceToColumn()
|
D | pyr_down.hpp | 68 __shared__ work_type smem[256 + 4]; in pyrDown() local 86 smem[2 + threadIdx.x] = sum; in pyrDown() 101 smem[threadIdx.x] = sum; in pyrDown() 116 smem[4 + threadIdx.x] = sum; in pyrDown() 130 smem[2 + threadIdx.x] = sum; in pyrDown() 145 smem[threadIdx.x] = sum; in pyrDown() 160 smem[4 + threadIdx.x] = sum; in pyrDown() 172 sum = 0.0625f * smem[2 + tid2 - 2]; in pyrDown() 173 sum = sum + 0.25f * smem[2 + tid2 - 1]; in pyrDown() 174 sum = sum + 0.375f * smem[2 + tid2 ]; in pyrDown() [all …]
|
D | integral.hpp | 63 __shared__ D smem[NUM_SCAN_THREADS * 2]; in horizontal_pass() local 84 const D curScanElem = blockScanInclusive<NUM_SCAN_THREADS>(curElem, smem, threadIdx.x); in horizontal_pass() 105 __shared__ D smem[NUM_SCAN_THREADS * 2]; in horizontal_pass() local 127 const D curScanElem = blockScanInclusive<NUM_SCAN_THREADS>(curElem, smem, threadIdx.x); in horizontal_pass() 481 __shared__ T smem[32][32]; in vertical_pass() 484 volatile T* smem_row = &smem[0][0] + 64 * threadIdx.y; in vertical_pass() 505 smem[threadIdx.y + 0][threadIdx.x] = 0.0f; in vertical_pass() 506 smem[threadIdx.y + 8][threadIdx.x] = 0.0f; in vertical_pass() 507 smem[threadIdx.y + 16][threadIdx.x] = 0.0f; in vertical_pass() 508 smem[threadIdx.y + 24][threadIdx.x] = 0.0f; in vertical_pass() [all …]
|
D | histogram.hpp | 60 __shared__ ResType smem[BIN_COUNT]; in histogram() local 66 smem[i] = 0; in histogram() 77 atomicAdd(&smem[data % BIN_COUNT], 1); in histogram() 86 const ResType histVal = smem[i]; in histogram()
|
/external/opencv3/modules/cudaimgproc/src/cuda/ |
D | canny.cu | 249 __shared__ volatile int smem[18][18]; in edgesHysteresisLocalKernel() local 254 smem[threadIdx.y + 1][threadIdx.x + 1] = checkIdx(y, x, map.rows, map.cols) ? map(y, x) : 0; in edgesHysteresisLocalKernel() 256 smem[0][threadIdx.x + 1] = checkIdx(y - 1, x, map.rows, map.cols) ? map(y - 1, x) : 0; in edgesHysteresisLocalKernel() 258 …smem[blockDim.y + 1][threadIdx.x + 1] = checkIdx(y + 1, x, map.rows, map.cols) ? map(y + 1, x) : 0; in edgesHysteresisLocalKernel() 260 smem[threadIdx.y + 1][0] = checkIdx(y, x - 1, map.rows, map.cols) ? map(y, x - 1) : 0; in edgesHysteresisLocalKernel() 262 …smem[threadIdx.y + 1][blockDim.x + 1] = checkIdx(y, x + 1, map.rows, map.cols) ? map(y, x + 1) : 0; in edgesHysteresisLocalKernel() 264 smem[0][0] = checkIdx(y - 1, x - 1, map.rows, map.cols) ? map(y - 1, x - 1) : 0; in edgesHysteresisLocalKernel() 266 … smem[0][blockDim.x + 1] = checkIdx(y - 1, x + 1, map.rows, map.cols) ? map(y - 1, x + 1) : 0; in edgesHysteresisLocalKernel() 268 … smem[blockDim.y + 1][0] = checkIdx(y + 1, x - 1, map.rows, map.cols) ? map(y + 1, x - 1) : 0; in edgesHysteresisLocalKernel() 270 …smem[blockDim.y + 1][blockDim.x + 1] = checkIdx(y + 1, x + 1, map.rows, map.cols) ? map(y + 1, x +… in edgesHysteresisLocalKernel() [all …]
|
D | clahe.cu | 61 __shared__ int smem[512]; in calcLutKernel() local 67 smem[tid] = 0; in calcLutKernel() 76 Emulation::smem::atomicAdd(&smem[data], 1); in calcLutKernel() 82 int tHistVal = smem[tid]; in calcLutKernel() 99 reduce<256>(smem, clipped, tid, plus<int>()); in calcLutKernel() 119 const int lutVal = blockScanInclusive<256>(tHistVal, smem, tid); in calcLutKernel()
|
D | hough_circles.cu | 184 int* smem = DynamicSharedMem<int>(); in circlesAccumRadius() local 187 smem[i] = 0; in circlesAccumRadius() 210 Emulation::smem::atomicAdd(&smem[r + 1], 1); in circlesAccumRadius() 218 const int curVotes = smem[i + 1]; in circlesAccumRadius() 220 if (curVotes >= threshold && curVotes > smem[i] && curVotes >= smem[i + 2]) in circlesAccumRadius()
|
/external/opencv3/modules/cudawarping/src/cuda/ |
D | pyr_down.cu | 59 __shared__ work_t smem[256 + 4]; in pyrDown() local 77 smem[2 + threadIdx.x] = sum; in pyrDown() 92 smem[threadIdx.x] = sum; in pyrDown() 107 smem[4 + threadIdx.x] = sum; in pyrDown() 121 smem[2 + threadIdx.x] = sum; in pyrDown() 136 smem[threadIdx.x] = sum; in pyrDown() 151 smem[4 + threadIdx.x] = sum; in pyrDown() 163 sum = 0.0625f * smem[2 + tid2 - 2]; in pyrDown() 164 sum = sum + 0.25f * smem[2 + tid2 - 1]; in pyrDown() 165 sum = sum + 0.375f * smem[2 + tid2 ]; in pyrDown() [all …]
|
/external/opencv3/modules/cudafeatures2d/src/cuda/ |
D | bf_match.cu | 141 extern __shared__ int smem[]; in matchUnrolledCached() 145 typename Dist::value_type* s_query = (typename Dist::value_type*)(smem); in matchUnrolledCached() 146 …typename Dist::value_type* s_train = (typename Dist::value_type*)(smem + BLOCK_SIZE * MAX_DESC_LEN… in matchUnrolledCached() 157 float* s_distance = (float*)(smem); in matchUnrolledCached() 158 int* s_trainIdx = (int*)(smem + BLOCK_SIZE * BLOCK_SIZE); in matchUnrolledCached() 190 extern __shared__ int smem[]; in matchUnrolledCached() 194 typename Dist::value_type* s_query = (typename Dist::value_type*)(smem); in matchUnrolledCached() 195 …typename Dist::value_type* s_train = (typename Dist::value_type*)(smem + BLOCK_SIZE * MAX_DESC_LEN… in matchUnrolledCached() 214 float* s_distance = (float*)(smem); in matchUnrolledCached() 215 int* s_trainIdx = (int*)(smem + BLOCK_SIZE * BLOCK_SIZE); in matchUnrolledCached() [all …]
|
D | bf_knnmatch.cu | 379 extern __shared__ int smem[]; in matchUnrolledCached() 383 typename Dist::value_type* s_query = (typename Dist::value_type*)(smem); in matchUnrolledCached() 384 …typename Dist::value_type* s_train = (typename Dist::value_type*)(smem + BLOCK_SIZE * MAX_DESC_LEN… in matchUnrolledCached() 397 float* s_distance = (float*)(smem); in matchUnrolledCached() 398 int* s_trainIdx = (int*)(smem + BLOCK_SIZE * BLOCK_SIZE); in matchUnrolledCached() 429 extern __shared__ int smem[]; in matchUnrolledCached() 433 typename Dist::value_type* s_query = (typename Dist::value_type*)(smem); in matchUnrolledCached() 434 …typename Dist::value_type* s_train = (typename Dist::value_type*)(smem + BLOCK_SIZE * MAX_DESC_LEN… in matchUnrolledCached() 456 float* s_distance = (float*)(smem); in matchUnrolledCached() 457 int* s_trainIdx = (int*)(smem + BLOCK_SIZE * BLOCK_SIZE); in matchUnrolledCached() [all …]
|
/external/opencv3/modules/photo/src/cuda/ |
D | nlm.cu | 189 … __device__ __forceinline__ thrust::tuple<volatile float*, volatile float*> smem_tuple(float* smem) in smem_tuple() 191 return cv::cuda::device::smem_tuple(smem, smem + BLOCK_SIZE); in smem_tuple() 208 …ceinline__ thrust::tuple<volatile float*, volatile float*, volatile float*> smem_tuple(float* smem) in smem_tuple() 210 return cv::cuda::device::smem_tuple(smem, smem + BLOCK_SIZE, smem + 2 * BLOCK_SIZE); in smem_tuple() 227 …::tuple<volatile float*, volatile float*, volatile float*, volatile float*> smem_tuple(float* smem) in smem_tuple() 229 …return cv::cuda::device::smem_tuple(smem, smem + BLOCK_SIZE, smem + 2 * BLOCK_SIZE, smem + 3 * BLO… in smem_tuple() 246 …float*, volatile float*, volatile float*, volatile float*, volatile float*> smem_tuple(float* smem) in smem_tuple() 248 …return cv::cuda::device::smem_tuple(smem, smem + BLOCK_SIZE, smem + 2 * BLOCK_SIZE, smem + 3 * BLO… in smem_tuple() 502 int smem = search_window * search_window * sizeof(int); in nlm_fast_gpu() local 505 fast_nlm_kernel<<<grid, block, smem>>>(fnlm, (PtrStepSz<T>)dst); in nlm_fast_gpu()
|
/external/opencv3/modules/cudaoptflow/src/cuda/ |
D | farneback.cu | 75 extern __shared__ float smem[]; in polynomialExpansion() 76 volatile float *row = smem + tx; in polynomialExpansion() 140 int smem = 3 * block.x * sizeof(float); in polynomialExpansionGpu() local 143 polynomialExpansion<5><<<grid, block, smem, stream>>>(src.rows, src.cols, src, dst); in polynomialExpansionGpu() 145 polynomialExpansion<7><<<grid, block, smem, stream>>>(src.rows, src.cols, src, dst); in polynomialExpansionGpu() 364 extern __shared__ float smem[]; in boxFilter5() 367 volatile float *row = smem + 5 * ty * smw; in boxFilter5() 422 int smem = (block.x + 2*ksizeHalf) * 5 * block.y * sizeof(float); in boxFilter5Gpu() local 425 boxFilter5<<<grid, block, smem, stream>>>(height, width, src, ksizeHalf, boxAreaInv, dst); in boxFilter5Gpu() 441 int smem = (block.x + 2*ksizeHalf) * 5 * block.y * sizeof(float); in boxFilter5Gpu_CC11() local [all …]
|