• Sayed Adel's avatar
    core: reimplement SIMD arithmetic, logic and comparison operations into wide universal intrinsics · 93ffebc2
    Sayed Adel authored
      - initialize arithmetic dispatcher
      - add new universal intrinsic v_absdiffs
      - add new universal intrinsic v_pack_b
      - add accumulate version of universal intrinsic v_round
      - fix sse/avx2:uint8 multiplication overflow
      - reimplement arithmetic, logic and comparison operations into wide universal intrinsics
        with full support for all types
      - reimplement IPP arithmetic, logic and comparison operations in a sperate file arithm_ipp.hpp
      - avoid scalar multiplication if scaling factor eq 1 and use integer multiplication
      - move C arithmetic operations to precomp.hpp and delete [arithm_simd|arithm_core].hpp
      - add compatibility with new opencv4 divide policy
    93ffebc2
arithm_ipp.hpp 16.6 KB
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 43 44 45 46 47 48 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 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417
// This file is part of OpenCV project.
// It is subject to the license terms in the LICENSE file found in the top-level directory
// of this distribution and at http://opencv.org/license.html
#if ARITHM_USE_IPP

namespace cv { namespace hal {

//=======================================
// Arithmetic and logical operations
// +, -, *, /, &, |, ^, ~, abs ...
//=======================================

#define ARITHM_IPP_BIN(fun, ...)                        \
do {                                                    \
    if (!CV_IPP_CHECK_COND)                             \
        return 0;                                       \
    if (height == 1)                                    \
        step1 = step2 = step = width * sizeof(dst[0]);  \
    if (0 <= CV_INSTRUMENT_FUN_IPP(fun, __VA_ARGS__))   \
    {                                                   \
        CV_IMPL_ADD(CV_IMPL_IPP);                       \
        return 1;                                       \
    }                                                   \
    setIppErrorStatus();                                \
    return 0;                                           \
} while(0)

//=======================================
// Addition
//=======================================

inline int arithm_ipp_add8u(const uchar* src1, size_t step1, const uchar* src2, size_t step2,
                            uchar* dst, size_t step, int width, int height)
{
    ARITHM_IPP_BIN(ippiAdd_8u_C1RSfs, src1, (int)step1, src2, (int)step2, dst, (int)step, ippiSize(width, height), 0);
}

inline int arithm_ipp_add16u(const ushort* src1, size_t step1, const ushort* src2, size_t step2,
                             ushort* dst, size_t step, int width, int height)
{
    ARITHM_IPP_BIN(ippiAdd_16u_C1RSfs, src1, (int)step1, src2, (int)step2, dst, (int)step, ippiSize(width, height), 0);
}

inline int arithm_ipp_add16s(const short* src1, size_t step1, const short* src2, size_t step2,
                             short* dst, size_t step, int width, int height)
{
    ARITHM_IPP_BIN(ippiAdd_16s_C1RSfs, src1, (int)step1, src2, (int)step2, dst, (int)step, ippiSize(width, height), 0);
}

inline int arithm_ipp_add32f(const float* src1, size_t step1, const float* src2, size_t step2,
                             float* dst, size_t step, int width, int height)
{
    ARITHM_IPP_BIN(ippiAdd_32f_C1R, src1, (int)step1, src2, (int)step2, dst, (int)step, ippiSize(width, height));
}

#define arithm_ipp_add8s(...)  0
#define arithm_ipp_add32s(...) 0
#define arithm_ipp_add64f(...) 0

//=======================================
// Subtract
//=======================================

inline int arithm_ipp_sub8u(const uchar* src1, size_t step1, const uchar* src2, size_t step2,
                            uchar* dst, size_t step, int width, int height)
{
    ARITHM_IPP_BIN(ippiSub_8u_C1RSfs, src2, (int)step2, src1, (int)step1, dst, (int)step, ippiSize(width, height), 0);
}

inline int arithm_ipp_sub16u(const ushort* src1, size_t step1, const ushort* src2, size_t step2,
                             ushort* dst, size_t step, int width, int height)
{
    ARITHM_IPP_BIN(ippiSub_16u_C1RSfs, src2, (int)step2, src1, (int)step1, dst, (int)step, ippiSize(width, height), 0);
}

inline int arithm_ipp_sub16s(const short* src1, size_t step1, const short* src2, size_t step2,
                            short* dst, size_t step, int width, int height)
{
    ARITHM_IPP_BIN(ippiSub_16s_C1RSfs, src2, (int)step2, src1, (int)step1, dst, (int)step, ippiSize(width, height), 0);
}

inline int arithm_ipp_sub32f(const float* src1, size_t step1, const float* src2, size_t step2,
                            float* dst, size_t step, int width, int height)
{
    ARITHM_IPP_BIN(ippiSub_32f_C1R, src2, (int)step2, src1, (int)step1, dst, (int)step, ippiSize(width, height));
}

#define arithm_ipp_sub8s(...)  0
#define arithm_ipp_sub32s(...) 0
#define arithm_ipp_sub64f(...) 0

///////////////////////////////////////////////////////////////////////////////////////////////////

#define ARITHM_IPP_MIN_MAX(fun, type)                            \
do {                                                             \
    if (!CV_IPP_CHECK_COND)                                      \
        return 0;                                                \
    type* s1 = (type*)src1;                                      \
    type* s2 = (type*)src2;                                      \
    type* d  = dst;                                              \
    if (height == 1)                                             \
        step1 = step2 = step = width * sizeof(dst[0]);           \
    int i = 0;                                                   \
    for(; i < height; i++)                                       \
    {                                                            \
        if (0 > CV_INSTRUMENT_FUN_IPP(fun, s1, s2, d, width))    \
            break;                                               \
        s1 = (type*)((uchar*)s1 + step1);                        \
        s2 = (type*)((uchar*)s2 + step2);                        \
        d  = (type*)((uchar*)d + step);                          \
    }                                                            \
    if (i == height)                                             \
    {                                                            \
        CV_IMPL_ADD(CV_IMPL_IPP);                                \
        return 1;                                                \
    }                                                            \
    setIppErrorStatus();                                         \
    return 0;                                                    \
} while(0)

//=======================================
// Max
//=======================================

inline int arithm_ipp_max8u(const uchar* src1, size_t step1, const uchar* src2, size_t step2,
                           uchar* dst, size_t step, int width, int height)
{
    ARITHM_IPP_MIN_MAX(ippsMaxEvery_8u, uchar);
}

inline int arithm_ipp_max16u(const ushort* src1, size_t step1, const ushort* src2, size_t step2,
                             ushort* dst, size_t step, int width, int height)
{
    ARITHM_IPP_MIN_MAX(ippsMaxEvery_16u, ushort);
}

inline int arithm_ipp_max32f(const float* src1, size_t step1, const float* src2, size_t step2,
                             float* dst, size_t step, int width, int height)
{
    ARITHM_IPP_MIN_MAX(ippsMaxEvery_32f, float);
}

inline int arithm_ipp_max64f(const double* src1, size_t step1, const double* src2, size_t step2,
                             double* dst, size_t step, int width, int height)
{
    ARITHM_IPP_MIN_MAX(ippsMaxEvery_64f, double);
}

#define arithm_ipp_max8s(...)  0
#define arithm_ipp_max16s(...) 0
#define arithm_ipp_max32s(...) 0

//=======================================
// Min
//=======================================

inline int arithm_ipp_min8u(const uchar* src1, size_t step1, const uchar* src2, size_t step2,
                            uchar* dst, size_t step, int width, int height)
{
    ARITHM_IPP_MIN_MAX(ippsMinEvery_8u, uchar);
}

inline int arithm_ipp_min16u(const ushort* src1, size_t step1, const ushort* src2, size_t step2,
                            ushort* dst, size_t step, int width, int height)
{
    ARITHM_IPP_MIN_MAX(ippsMinEvery_16u, ushort);
}

inline int arithm_ipp_min32f(const float* src1, size_t step1, const float* src2,size_t step2,
                             float* dst, size_t step, int width, int height)
{
    ARITHM_IPP_MIN_MAX(ippsMinEvery_32f, float);
}

inline int arithm_ipp_min64f(const double* src1, size_t step1, const double* src2, size_t step2,
                             double* dst, size_t step, int width, int height)
{
    ARITHM_IPP_MIN_MAX(ippsMinEvery_64f, double);
}

#define arithm_ipp_min8s(...)  0
#define arithm_ipp_min16s(...) 0
#define arithm_ipp_min32s(...) 0

//=======================================
// AbsDiff
//=======================================

inline int arithm_ipp_absdiff8u(const uchar* src1, size_t step1, const uchar* src2, size_t step2,
                                uchar* dst, size_t step, int width, int height)
{
    ARITHM_IPP_BIN(ippiAbsDiff_8u_C1R, src1, (int)step1, src2, (int)step2, dst, (int)step, ippiSize(width, height));
}

inline int arithm_ipp_absdiff16u(const ushort* src1, size_t step1, const ushort* src2, size_t step2,
                                ushort* dst, size_t step, int width, int height)
{
    ARITHM_IPP_BIN(ippiAbsDiff_16u_C1R, src1, (int)step1, src2, (int)step2, dst, (int)step, ippiSize(width, height));
}

inline int arithm_ipp_absdiff32f(const float* src1, size_t step1, const float* src2, size_t step2,
                                float* dst, size_t step, int width, int height)
{
    ARITHM_IPP_BIN(ippiAbsDiff_32f_C1R, src1, (int)step1, src2, (int)step2, dst, (int)step, ippiSize(width, height));
}
#define arithm_ipp_absdiff8s(...)  0
#define arithm_ipp_absdiff16s(...) 0
#define arithm_ipp_absdiff32s(...) 0
#define arithm_ipp_absdiff64f(...) 0

//=======================================
// Logical
//=======================================

inline int arithm_ipp_and8u(const uchar* src1, size_t step1, const uchar* src2, size_t step2,
                            uchar* dst, size_t step, int width, int height)
{
    ARITHM_IPP_BIN(ippiAnd_8u_C1R, src1, (int)step1, src2, (int)step2, dst, (int)step, ippiSize(width, height));
}

inline int arithm_ipp_or8u(const uchar* src1, size_t step1, const uchar* src2, size_t step2,
                           uchar* dst, size_t step, int width, int height)
{
    ARITHM_IPP_BIN(ippiOr_8u_C1R, src1, (int)step1, src2, (int)step2, dst, (int)step, ippiSize(width, height));
}

inline int arithm_ipp_xor8u(const uchar* src1, size_t step1, const uchar* src2, size_t step2,
                            uchar* dst, size_t step, int width, int height)
{
    ARITHM_IPP_BIN(ippiXor_8u_C1R, src1, (int)step1, src2, (int)step2, dst, (int)step, ippiSize(width, height));
}

inline int arithm_ipp_not8u(const uchar* src1, size_t step1, uchar* dst, size_t step, int width, int height)
{
    if (!CV_IPP_CHECK_COND)
        return 0;
    if (height == 1)
        step1 = step = width * sizeof(dst[0]);
    if (0 <= CV_INSTRUMENT_FUN_IPP(ippiNot_8u_C1R, src1, (int)step1, dst, (int)step, ippiSize(width, height)))
    {
        CV_IMPL_ADD(CV_IMPL_IPP);
        return 1;
    }
    setIppErrorStatus();
    return 0;
}

//=======================================
// Compare
//=======================================

#define ARITHM_IPP_CMP(fun, ...)                          \
do {                                                      \
    if (!CV_IPP_CHECK_COND)                               \
        return 0;                                         \
    IppCmpOp op = arithm_ipp_convert_cmp(cmpop);          \
    if (op < 0)                                           \
        return 0;                                         \
    if (height == 1)                                      \
        step1 = step2 = step = width * sizeof(dst[0]);    \
    if (0 <= CV_INSTRUMENT_FUN_IPP(fun, __VA_ARGS__, op)) \
    {                                                     \
        CV_IMPL_ADD(CV_IMPL_IPP);                         \
        return 1;                                         \
    }                                                     \
    setIppErrorStatus();                                  \
    return 0;                                             \
} while(0)

inline IppCmpOp arithm_ipp_convert_cmp(int cmpop)
{
    switch(cmpop)
    {
        case CMP_EQ: return ippCmpEq;
        case CMP_GT: return ippCmpGreater;
        case CMP_GE: return ippCmpGreaterEq;
        case CMP_LT: return ippCmpLess;
        case CMP_LE: return ippCmpLessEq;
        default:     return (IppCmpOp)-1;
    }
}

inline int arithm_ipp_cmp8u(const uchar* src1, size_t step1, const uchar* src2, size_t step2,
                            uchar* dst, size_t step, int width, int height, int cmpop)
{
    ARITHM_IPP_CMP(ippiCompare_8u_C1R, src1, (int)step1, src2, (int)step2, dst, (int)step, ippiSize(width, height));
}

inline int arithm_ipp_cmp16u(const ushort* src1, size_t step1, const ushort* src2, size_t step2,
                            uchar* dst, size_t step, int width, int height, int cmpop)
{
    ARITHM_IPP_CMP(ippiCompare_16u_C1R, src1, (int)step1, src2, (int)step2, dst, (int)step, ippiSize(width, height));
}

inline int arithm_ipp_cmp16s(const short* src1, size_t step1, const short* src2, size_t step2,
                             uchar* dst, size_t step, int width, int height, int cmpop)
{
    ARITHM_IPP_CMP(ippiCompare_16s_C1R, src1, (int)step1, src2, (int)step2, dst, (int)step, ippiSize(width, height));
}

inline int arithm_ipp_cmp32f(const float* src1, size_t step1, const float* src2, size_t step2,
                             uchar* dst, size_t step, int width, int height, int cmpop)
{
    ARITHM_IPP_CMP(ippiCompare_32f_C1R, src1, (int)step1, src2, (int)step2, dst, (int)step, ippiSize(width, height));
}

#define arithm_ipp_cmp8s(...)  0
#define arithm_ipp_cmp32s(...) 0
#define arithm_ipp_cmp64f(...) 0

//=======================================
// Multiply
//=======================================

#define ARITHM_IPP_MUL(fun, ...)                      \
do {                                                  \
    if (!CV_IPP_CHECK_COND)                           \
        return 0;                                     \
    float fscale = (float)scale;                      \
    if (std::fabs(fscale - 1) > FLT_EPSILON)          \
        return 0;                                     \
    if (0 <= CV_INSTRUMENT_FUN_IPP(fun, __VA_ARGS__)) \
    {                                                 \
        CV_IMPL_ADD(CV_IMPL_IPP);                     \
        return 1;                                     \
    }                                                 \
    setIppErrorStatus();                              \
    return 0;                                         \
} while(0)

inline int arithm_ipp_mul8u(const uchar *src1, size_t step1, const uchar *src2, size_t step2,
                            uchar *dst, size_t step, int width, int height, double scale)
{
    ARITHM_IPP_MUL(ippiMul_8u_C1RSfs, src1, (int)step1, src2, (int)step2,dst, (int)step, ippiSize(width, height), 0);
}
inline int arithm_ipp_mul16u(const ushort *src1, size_t step1, const ushort *src2, size_t step2,
                            ushort *dst, size_t step, int width, int height, double scale)
{
    ARITHM_IPP_MUL(ippiMul_16u_C1RSfs, src1, (int)step1, src2, (int)step2,dst, (int)step, ippiSize(width, height), 0);
}

inline int arithm_ipp_mul16s(const short *src1, size_t step1, const short *src2, size_t step2,
                            short *dst, size_t step, int width, int height, double scale)
{
    ARITHM_IPP_MUL(ippiMul_16s_C1RSfs, src1, (int)step1, src2, (int)step2,dst, (int)step, ippiSize(width, height), 0);
}

inline int arithm_ipp_mul32f(const float *src1, size_t step1, const float *src2, size_t step2,
                            float *dst, size_t step, int width, int height, double scale)
{
    ARITHM_IPP_MUL(ippiMul_32f_C1R, src1, (int)step1, src2, (int)step2,dst, (int)step, ippiSize(width, height));
}

#define arithm_ipp_mul8s(...)  0
#define arithm_ipp_mul32s(...) 0
#define arithm_ipp_mul64f(...) 0

//=======================================
// Div
//=======================================

#define arithm_ipp_div8u(...)  0
#define arithm_ipp_div8s(...)  0
#define arithm_ipp_div16u(...) 0
#define arithm_ipp_div16s(...) 0
#define arithm_ipp_div32s(...) 0
#define arithm_ipp_div32f(...) 0
#define arithm_ipp_div64f(...) 0

//=======================================
// AddWeighted
//=======================================

#define arithm_ipp_addWeighted8u(...)  0
#define arithm_ipp_addWeighted8s(...)  0
#define arithm_ipp_addWeighted16u(...) 0
#define arithm_ipp_addWeighted16s(...) 0
#define arithm_ipp_addWeighted32s(...) 0
#define arithm_ipp_addWeighted32f(...) 0
#define arithm_ipp_addWeighted64f(...) 0

//=======================================
// Reciprocial
//=======================================

#define arithm_ipp_recip8u(...)  0
#define arithm_ipp_recip8s(...)  0
#define arithm_ipp_recip16u(...) 0
#define arithm_ipp_recip16s(...) 0
#define arithm_ipp_recip32s(...) 0
#define arithm_ipp_recip32f(...) 0
#define arithm_ipp_recip64f(...) 0

/** empty block in case if you have "fun"
#define arithm_ipp_8u(...)  0
#define arithm_ipp_8s(...)  0
#define arithm_ipp_16u(...) 0
#define arithm_ipp_16s(...) 0
#define arithm_ipp_32s(...) 0
#define arithm_ipp_32f(...) 0
#define arithm_ipp_64f(...) 0
**/

}} // cv::hal::

#define ARITHM_CALL_IPP(fun, ...)       \
{                                       \
    if (__CV_EXPAND(fun(__VA_ARGS__)))  \
        return;                         \
}

#endif // ARITHM_USE_IPP


#if !ARITHM_USE_IPP
#define ARITHM_CALL_IPP(...)
#endif