saturate_cast.hpp 9.76 KB
Newer Older
wester committed
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42
/*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*/

wester committed
43 44
#ifndef __OPENCV_GPU_SATURATE_CAST_HPP__
#define __OPENCV_GPU_SATURATE_CAST_HPP__
wester committed
45 46 47

#include "common.hpp"

wester committed
48
namespace cv { namespace gpu { namespace device
wester committed
49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97
{
    template<typename _Tp> __device__ __forceinline__ _Tp saturate_cast(uchar v) { return _Tp(v); }
    template<typename _Tp> __device__ __forceinline__ _Tp saturate_cast(schar v) { return _Tp(v); }
    template<typename _Tp> __device__ __forceinline__ _Tp saturate_cast(ushort v) { return _Tp(v); }
    template<typename _Tp> __device__ __forceinline__ _Tp saturate_cast(short v) { return _Tp(v); }
    template<typename _Tp> __device__ __forceinline__ _Tp saturate_cast(uint v) { return _Tp(v); }
    template<typename _Tp> __device__ __forceinline__ _Tp saturate_cast(int v) { return _Tp(v); }
    template<typename _Tp> __device__ __forceinline__ _Tp saturate_cast(float v) { return _Tp(v); }
    template<typename _Tp> __device__ __forceinline__ _Tp saturate_cast(double v) { return _Tp(v); }

    template<> __device__ __forceinline__ uchar saturate_cast<uchar>(schar v)
    {
        uint res = 0;
        int vi = v;
        asm("cvt.sat.u8.s8 %0, %1;" : "=r"(res) : "r"(vi));
        return res;
    }
    template<> __device__ __forceinline__ uchar saturate_cast<uchar>(short v)
    {
        uint res = 0;
        asm("cvt.sat.u8.s16 %0, %1;" : "=r"(res) : "h"(v));
        return res;
    }
    template<> __device__ __forceinline__ uchar saturate_cast<uchar>(ushort v)
    {
        uint res = 0;
        asm("cvt.sat.u8.u16 %0, %1;" : "=r"(res) : "h"(v));
        return res;
    }
    template<> __device__ __forceinline__ uchar saturate_cast<uchar>(int v)
    {
        uint res = 0;
        asm("cvt.sat.u8.s32 %0, %1;" : "=r"(res) : "r"(v));
        return res;
    }
    template<> __device__ __forceinline__ uchar saturate_cast<uchar>(uint v)
    {
        uint res = 0;
        asm("cvt.sat.u8.u32 %0, %1;" : "=r"(res) : "r"(v));
        return res;
    }
    template<> __device__ __forceinline__ uchar saturate_cast<uchar>(float v)
    {
        uint res = 0;
        asm("cvt.rni.sat.u8.f32 %0, %1;" : "=r"(res) : "f"(v));
        return res;
    }
    template<> __device__ __forceinline__ uchar saturate_cast<uchar>(double v)
    {
a  
Kai Westerkamp committed
98
    #if __CUDA_ARCH__ >= 130
wester committed
99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145
        uint res = 0;
        asm("cvt.rni.sat.u8.f64 %0, %1;" : "=r"(res) : "d"(v));
        return res;
    #else
        return saturate_cast<uchar>((float)v);
    #endif
    }

    template<> __device__ __forceinline__ schar saturate_cast<schar>(uchar v)
    {
        uint res = 0;
        uint vi = v;
        asm("cvt.sat.s8.u8 %0, %1;" : "=r"(res) : "r"(vi));
        return res;
    }
    template<> __device__ __forceinline__ schar saturate_cast<schar>(short v)
    {
        uint res = 0;
        asm("cvt.sat.s8.s16 %0, %1;" : "=r"(res) : "h"(v));
        return res;
    }
    template<> __device__ __forceinline__ schar saturate_cast<schar>(ushort v)
    {
        uint res = 0;
        asm("cvt.sat.s8.u16 %0, %1;" : "=r"(res) : "h"(v));
        return res;
    }
    template<> __device__ __forceinline__ schar saturate_cast<schar>(int v)
    {
        uint res = 0;
        asm("cvt.sat.s8.s32 %0, %1;" : "=r"(res) : "r"(v));
        return res;
    }
    template<> __device__ __forceinline__ schar saturate_cast<schar>(uint v)
    {
        uint res = 0;
        asm("cvt.sat.s8.u32 %0, %1;" : "=r"(res) : "r"(v));
        return res;
    }
    template<> __device__ __forceinline__ schar saturate_cast<schar>(float v)
    {
        uint res = 0;
        asm("cvt.rni.sat.s8.f32 %0, %1;" : "=r"(res) : "f"(v));
        return res;
    }
    template<> __device__ __forceinline__ schar saturate_cast<schar>(double v)
    {
a  
Kai Westerkamp committed
146
    #if __CUDA_ARCH__ >= 130
wester committed
147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187
        uint res = 0;
        asm("cvt.rni.sat.s8.f64 %0, %1;" : "=r"(res) : "d"(v));
        return res;
    #else
        return saturate_cast<schar>((float)v);
    #endif
    }

    template<> __device__ __forceinline__ ushort saturate_cast<ushort>(schar v)
    {
        ushort res = 0;
        int vi = v;
        asm("cvt.sat.u16.s8 %0, %1;" : "=h"(res) : "r"(vi));
        return res;
    }
    template<> __device__ __forceinline__ ushort saturate_cast<ushort>(short v)
    {
        ushort res = 0;
        asm("cvt.sat.u16.s16 %0, %1;" : "=h"(res) : "h"(v));
        return res;
    }
    template<> __device__ __forceinline__ ushort saturate_cast<ushort>(int v)
    {
        ushort res = 0;
        asm("cvt.sat.u16.s32 %0, %1;" : "=h"(res) : "r"(v));
        return res;
    }
    template<> __device__ __forceinline__ ushort saturate_cast<ushort>(uint v)
    {
        ushort res = 0;
        asm("cvt.sat.u16.u32 %0, %1;" : "=h"(res) : "r"(v));
        return res;
    }
    template<> __device__ __forceinline__ ushort saturate_cast<ushort>(float v)
    {
        ushort res = 0;
        asm("cvt.rni.sat.u16.f32 %0, %1;" : "=h"(res) : "f"(v));
        return res;
    }
    template<> __device__ __forceinline__ ushort saturate_cast<ushort>(double v)
    {
a  
Kai Westerkamp committed
188
    #if __CUDA_ARCH__ >= 130
wester committed
189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222
        ushort res = 0;
        asm("cvt.rni.sat.u16.f64 %0, %1;" : "=h"(res) : "d"(v));
        return res;
    #else
        return saturate_cast<ushort>((float)v);
    #endif
    }

    template<> __device__ __forceinline__ short saturate_cast<short>(ushort v)
    {
        short res = 0;
        asm("cvt.sat.s16.u16 %0, %1;" : "=h"(res) : "h"(v));
        return res;
    }
    template<> __device__ __forceinline__ short saturate_cast<short>(int v)
    {
        short res = 0;
        asm("cvt.sat.s16.s32 %0, %1;" : "=h"(res) : "r"(v));
        return res;
    }
    template<> __device__ __forceinline__ short saturate_cast<short>(uint v)
    {
        short res = 0;
        asm("cvt.sat.s16.u32 %0, %1;" : "=h"(res) : "r"(v));
        return res;
    }
    template<> __device__ __forceinline__ short saturate_cast<short>(float v)
    {
        short res = 0;
        asm("cvt.rni.sat.s16.f32 %0, %1;" : "=h"(res) : "f"(v));
        return res;
    }
    template<> __device__ __forceinline__ short saturate_cast<short>(double v)
    {
a  
Kai Westerkamp committed
223
    #if __CUDA_ARCH__ >= 130
wester committed
224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283
        short res = 0;
        asm("cvt.rni.sat.s16.f64 %0, %1;" : "=h"(res) : "d"(v));
        return res;
    #else
        return saturate_cast<short>((float)v);
    #endif
    }

    template<> __device__ __forceinline__ int saturate_cast<int>(uint v)
    {
        int res = 0;
        asm("cvt.sat.s32.u32 %0, %1;" : "=r"(res) : "r"(v));
        return res;
    }
    template<> __device__ __forceinline__ int saturate_cast<int>(float v)
    {
        return __float2int_rn(v);
    }
    template<> __device__ __forceinline__ int saturate_cast<int>(double v)
    {
    #if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 130
        return __double2int_rn(v);
    #else
        return saturate_cast<int>((float)v);
    #endif
    }

    template<> __device__ __forceinline__ uint saturate_cast<uint>(schar v)
    {
        uint res = 0;
        int vi = v;
        asm("cvt.sat.u32.s8 %0, %1;" : "=r"(res) : "r"(vi));
        return res;
    }
    template<> __device__ __forceinline__ uint saturate_cast<uint>(short v)
    {
        uint res = 0;
        asm("cvt.sat.u32.s16 %0, %1;" : "=r"(res) : "h"(v));
        return res;
    }
    template<> __device__ __forceinline__ uint saturate_cast<uint>(int v)
    {
        uint res = 0;
        asm("cvt.sat.u32.s32 %0, %1;" : "=r"(res) : "r"(v));
        return res;
    }
    template<> __device__ __forceinline__ uint saturate_cast<uint>(float v)
    {
        return __float2uint_rn(v);
    }
    template<> __device__ __forceinline__ uint saturate_cast<uint>(double v)
    {
    #if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 130
        return __double2uint_rn(v);
    #else
        return saturate_cast<uint>((float)v);
    #endif
    }
}}}

wester committed
284
#endif /* __OPENCV_GPU_SATURATE_CAST_HPP__ */