node-native-win-utils 1.1.1 → 1.3.0

package/include/opencv2/core/cuda/detail/vec_distance_detail.hpp

@@ -0,0 +1,121 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                           License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_CUDA_VEC_DISTANCE_DETAIL_HPP
+#define OPENCV_CUDA_VEC_DISTANCE_DETAIL_HPP
+
+#include "../datamov_utils.hpp"
+
+//! @cond IGNORED
+
+namespace cv { namespace cuda { namespace device
+{
+    namespace vec_distance_detail
+    {
+        template <int THREAD_DIM, int N> struct UnrollVecDiffCached
+        {
+            template <typename Dist, typename T1, typename T2>
+            static __device__ void calcCheck(const T1* vecCached, const T2* vecGlob, int len, Dist& dist, int ind)
+            {
+                if (ind < len)
+                {
+                    T1 val1 = *vecCached++;
+
+                    T2 val2;
+                    ForceGlob<T2>::Load(vecGlob, ind, val2);
+
+                    dist.reduceIter(val1, val2);
+
+                    UnrollVecDiffCached<THREAD_DIM, N - 1>::calcCheck(vecCached, vecGlob, len, dist, ind + THREAD_DIM);
+                }
+            }
+
+            template <typename Dist, typename T1, typename T2>
+            static __device__ void calcWithoutCheck(const T1* vecCached, const T2* vecGlob, Dist& dist)
+            {
+                T1 val1 = *vecCached++;
+
+                T2 val2;
+                ForceGlob<T2>::Load(vecGlob, 0, val2);
+                vecGlob += THREAD_DIM;
+
+                dist.reduceIter(val1, val2);
+
+                UnrollVecDiffCached<THREAD_DIM, N - 1>::calcWithoutCheck(vecCached, vecGlob, dist);
+            }
+        };
+        template <int THREAD_DIM> struct UnrollVecDiffCached<THREAD_DIM, 0>
+        {
+            template <typename Dist, typename T1, typename T2>
+            static __device__ __forceinline__ void calcCheck(const T1*, const T2*, int, Dist&, int)
+            {
+            }
+
+            template <typename Dist, typename T1, typename T2>
+            static __device__ __forceinline__ void calcWithoutCheck(const T1*, const T2*, Dist&)
+            {
+            }
+        };
+
+        template <int THREAD_DIM, int MAX_LEN, bool LEN_EQ_MAX_LEN> struct VecDiffCachedCalculator;
+        template <int THREAD_DIM, int MAX_LEN> struct VecDiffCachedCalculator<THREAD_DIM, MAX_LEN, false>
+        {
+            template <typename Dist, typename T1, typename T2>
+            static __device__ __forceinline__ void calc(const T1* vecCached, const T2* vecGlob, int len, Dist& dist, int tid)
+            {
+                UnrollVecDiffCached<THREAD_DIM, MAX_LEN / THREAD_DIM>::calcCheck(vecCached, vecGlob, len, dist, tid);
+            }
+        };
+        template <int THREAD_DIM, int MAX_LEN> struct VecDiffCachedCalculator<THREAD_DIM, MAX_LEN, true>
+        {
+            template <typename Dist, typename T1, typename T2>
+            static __device__ __forceinline__ void calc(const T1* vecCached, const T2* vecGlob, int len, Dist& dist, int tid)
+            {
+                UnrollVecDiffCached<THREAD_DIM, MAX_LEN / THREAD_DIM>::calcWithoutCheck(vecCached, vecGlob + tid, dist);
+            }
+        };
+    } // namespace vec_distance_detail
+}}} // namespace cv { namespace cuda { namespace cudev
+
+//! @endcond
+
+#endif // OPENCV_CUDA_VEC_DISTANCE_DETAIL_HPP

package/include/opencv2/core/cuda/dynamic_smem.hpp

@@ -0,0 +1,88 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                           License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_CUDA_DYNAMIC_SMEM_HPP
+#define OPENCV_CUDA_DYNAMIC_SMEM_HPP
+
+/** @file
+ * @deprecated Use @ref cudev instead.
+ */
+
+//! @cond IGNORED
+
+namespace cv { namespace cuda { namespace device
+{
+    template<class T> struct DynamicSharedMem
+    {
+        __device__ __forceinline__ operator T*()
+        {
+            extern __shared__ int __smem[];
+            return (T*)__smem;
+        }
+
+        __device__ __forceinline__ operator const T*() const
+        {
+            extern __shared__ int __smem[];
+            return (T*)__smem;
+        }
+    };
+
+    // specialize for double to avoid unaligned memory access compile errors
+    template<> struct DynamicSharedMem<double>
+    {
+        __device__ __forceinline__ operator double*()
+        {
+            extern __shared__ double __smem_d[];
+            return (double*)__smem_d;
+        }
+
+        __device__ __forceinline__ operator const double*() const
+        {
+            extern __shared__ double __smem_d[];
+            return (double*)__smem_d;
+        }
+    };
+}}}
+
+//! @endcond
+
+#endif // OPENCV_CUDA_DYNAMIC_SMEM_HPP

package/include/opencv2/core/cuda/emulation.hpp

@@ -0,0 +1,269 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                           License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_CUDA_EMULATION_HPP_
+#define OPENCV_CUDA_EMULATION_HPP_
+
+#include "common.hpp"
+#include "warp_reduce.hpp"
+
+/** @file
+ * @deprecated Use @ref cudev instead.
+ */
+
+//! @cond IGNORED
+
+namespace cv { namespace cuda { namespace device
+{
+    struct Emulation
+    {
+
+        static __device__ __forceinline__ int syncthreadsOr(int pred)
+        {
+#if defined (__CUDA_ARCH__) && (__CUDA_ARCH__ < 200)
+                // just campilation stab
+                return 0;
+#else
+                return __syncthreads_or(pred);
+#endif
+        }
+
+        template<int CTA_SIZE>
+        static __forceinline__ __device__ int Ballot(int predicate)
+        {
+#if defined (__CUDA_ARCH__) && (__CUDA_ARCH__ >= 200)
+            return __ballot(predicate);
+#else
+            __shared__ volatile int cta_buffer[CTA_SIZE];
+
+            int tid = threadIdx.x;
+            cta_buffer[tid] = predicate ? (1 << (tid & 31)) : 0;
+            return warp_reduce(cta_buffer);
+#endif
+        }
+
+        struct smem
+        {
+            enum { TAG_MASK = (1U << ( (sizeof(unsigned int) << 3) - 5U)) - 1U };
+
+            template<typename T>
+            static __device__ __forceinline__ T atomicInc(T* address, T val)
+            {
+#if defined (__CUDA_ARCH__) && (__CUDA_ARCH__ < 120)
+                T count;
+                unsigned int tag = threadIdx.x << ( (sizeof(unsigned int) << 3) - 5U);
+                do
+                {
+                    count = *address & TAG_MASK;
+                    count = tag | (count + 1);
+                    *address = count;
+                } while (*address != count);
+
+                return (count & TAG_MASK) - 1;
+#else
+                return ::atomicInc(address, val);
+#endif
+            }
+
+            template<typename T>
+            static __device__ __forceinline__ T atomicAdd(T* address, T val)
+            {
+#if defined (__CUDA_ARCH__) && (__CUDA_ARCH__ < 120)
+                T count;
+                unsigned int tag = threadIdx.x << ( (sizeof(unsigned int) << 3) - 5U);
+                do
+                {
+                    count = *address & TAG_MASK;
+                    count = tag | (count + val);
+                    *address = count;
+                } while (*address != count);
+
+                return (count & TAG_MASK) - val;
+#else
+                return ::atomicAdd(address, val);
+#endif
+            }
+
+            template<typename T>
+            static __device__ __forceinline__ T atomicMin(T* address, T val)
+            {
+#if defined (__CUDA_ARCH__) && (__CUDA_ARCH__ < 120)
+                T count = ::min(*address, val);
+                do
+                {
+                    *address = count;
+                } while (*address > count);
+
+                return count;
+#else
+                return ::atomicMin(address, val);
+#endif
+            }
+        }; // struct cmem
+
+        struct glob
+        {
+            static __device__ __forceinline__ int atomicAdd(int* address, int val)
+            {
+                return ::atomicAdd(address, val);
+            }
+            static __device__ __forceinline__ unsigned int atomicAdd(unsigned int* address, unsigned int val)
+            {
+                return ::atomicAdd(address, val);
+            }
+            static __device__ __forceinline__ float atomicAdd(float* address, float val)
+            {
+            #if __CUDA_ARCH__ >= 200
+                return ::atomicAdd(address, val);
+            #else
+                int* address_as_i = (int*) address;
+                int old = *address_as_i, assumed;
+                do {
+                    assumed = old;
+                    old = ::atomicCAS(address_as_i, assumed,
+                        __float_as_int(val + __int_as_float(assumed)));
+                } while (assumed != old);
+                return __int_as_float(old);
+            #endif
+            }
+            static __device__ __forceinline__ double atomicAdd(double* address, double val)
+            {
+            #if __CUDA_ARCH__ >= 130
+                unsigned long long int* address_as_ull = (unsigned long long int*) address;
+                unsigned long long int old = *address_as_ull, assumed;
+                do {
+                    assumed = old;
+                    old = ::atomicCAS(address_as_ull, assumed,
+                        __double_as_longlong(val + __longlong_as_double(assumed)));
+                } while (assumed != old);
+                return __longlong_as_double(old);
+            #else
+                CV_UNUSED(address);
+                CV_UNUSED(val);
+                return 0.0;
+            #endif
+            }
+
+            static __device__ __forceinline__ int atomicMin(int* address, int val)
+            {
+                return ::atomicMin(address, val);
+            }
+            static __device__ __forceinline__ float atomicMin(float* address, float val)
+            {
+            #if __CUDA_ARCH__ >= 120
+                int* address_as_i = (int*) address;
+                int old = *address_as_i, assumed;
+                do {
+                    assumed = old;
+                    old = ::atomicCAS(address_as_i, assumed,
+                        __float_as_int(::fminf(val, __int_as_float(assumed))));
+                } while (assumed != old);
+                return __int_as_float(old);
+            #else
+                CV_UNUSED(address);
+                CV_UNUSED(val);
+                return 0.0f;
+            #endif
+            }
+            static __device__ __forceinline__ double atomicMin(double* address, double val)
+            {
+            #if __CUDA_ARCH__ >= 130
+                unsigned long long int* address_as_ull = (unsigned long long int*) address;
+                unsigned long long int old = *address_as_ull, assumed;
+                do {
+                    assumed = old;
+                    old = ::atomicCAS(address_as_ull, assumed,
+                        __double_as_longlong(::fmin(val, __longlong_as_double(assumed))));
+                } while (assumed != old);
+                return __longlong_as_double(old);
+            #else
+                CV_UNUSED(address);
+                CV_UNUSED(val);
+                return 0.0;
+            #endif
+            }
+
+            static __device__ __forceinline__ int atomicMax(int* address, int val)
+            {
+                return ::atomicMax(address, val);
+            }
+            static __device__ __forceinline__ float atomicMax(float* address, float val)
+            {
+            #if __CUDA_ARCH__ >= 120
+                int* address_as_i = (int*) address;
+                int old = *address_as_i, assumed;
+                do {
+                    assumed = old;
+                    old = ::atomicCAS(address_as_i, assumed,
+                        __float_as_int(::fmaxf(val, __int_as_float(assumed))));
+                } while (assumed != old);
+                return __int_as_float(old);
+            #else
+                CV_UNUSED(address);
+                CV_UNUSED(val);
+                return 0.0f;
+            #endif
+            }
+            static __device__ __forceinline__ double atomicMax(double* address, double val)
+            {
+            #if __CUDA_ARCH__ >= 130
+                unsigned long long int* address_as_ull = (unsigned long long int*) address;
+                unsigned long long int old = *address_as_ull, assumed;
+                do {
+                    assumed = old;
+                    old = ::atomicCAS(address_as_ull, assumed,
+                        __double_as_longlong(::fmax(val, __longlong_as_double(assumed))));
+                } while (assumed != old);
+                return __longlong_as_double(old);
+            #else
+                CV_UNUSED(address);
+                CV_UNUSED(val);
+                return 0.0;
+            #endif
+            }
+        };
+    }; //struct Emulation
+}}} // namespace cv { namespace cuda { namespace cudev
+
+//! @endcond
+
+#endif /* OPENCV_CUDA_EMULATION_HPP_ */