| 1 | /// @ref core |
| 2 | |
| 3 | #include "_vectorize.hpp" |
| 4 | #if(GLM_ARCH & GLM_ARCH_X86 && GLM_COMPILER & GLM_COMPILER_VC) |
| 5 | # include <intrin.h> |
| 6 | # pragma intrinsic(_BitScanReverse) |
| 7 | #endif//(GLM_ARCH & GLM_ARCH_X86 && GLM_COMPILER & GLM_COMPILER_VC) |
| 8 | #include <limits> |
| 9 | |
| 10 | namespace glm{ |
| 11 | namespace detail |
| 12 | { |
| 13 | template<typename T> |
| 14 | GLM_FUNC_QUALIFIER T mask(T Bits) |
| 15 | { |
| 16 | return Bits >= static_cast<T>(sizeof(T) * 8) ? ~static_cast<T>(0) : (static_cast<T>(1) << Bits) - static_cast<T>(1); |
| 17 | } |
| 18 | |
| 19 | template<length_t L, typename T, qualifier Q, bool Aligned, bool EXEC> |
| 20 | struct compute_bitfieldReverseStep |
| 21 | { |
| 22 | GLM_FUNC_QUALIFIER static vec<L, T, Q> call(vec<L, T, Q> const& v, T, T) |
| 23 | { |
| 24 | return v; |
| 25 | } |
| 26 | }; |
| 27 | |
| 28 | template<length_t L, typename T, qualifier Q, bool Aligned> |
| 29 | struct compute_bitfieldReverseStep<L, T, Q, Aligned, true> |
| 30 | { |
| 31 | GLM_FUNC_QUALIFIER static vec<L, T, Q> call(vec<L, T, Q> const& v, T Mask, T Shift) |
| 32 | { |
| 33 | return (v & Mask) << Shift | (v & (~Mask)) >> Shift; |
| 34 | } |
| 35 | }; |
| 36 | |
| 37 | template<length_t L, typename T, qualifier Q, bool Aligned, bool EXEC> |
| 38 | struct compute_bitfieldBitCountStep |
| 39 | { |
| 40 | GLM_FUNC_QUALIFIER static vec<L, T, Q> call(vec<L, T, Q> const& v, T, T) |
| 41 | { |
| 42 | return v; |
| 43 | } |
| 44 | }; |
| 45 | |
| 46 | template<length_t L, typename T, qualifier Q, bool Aligned> |
| 47 | struct compute_bitfieldBitCountStep<L, T, Q, Aligned, true> |
| 48 | { |
| 49 | GLM_FUNC_QUALIFIER static vec<L, T, Q> call(vec<L, T, Q> const& v, T Mask, T Shift) |
| 50 | { |
| 51 | return (v & Mask) + ((v >> Shift) & Mask); |
| 52 | } |
| 53 | }; |
| 54 | |
| 55 | template<typename genIUType, size_t Bits> |
| 56 | struct compute_findLSB |
| 57 | { |
| 58 | GLM_FUNC_QUALIFIER static int call(genIUType Value) |
| 59 | { |
| 60 | if(Value == 0) |
| 61 | return -1; |
| 62 | |
| 63 | return glm::bitCount(~Value & (Value - static_cast<genIUType>(1))); |
| 64 | } |
| 65 | }; |
| 66 | |
| 67 | # if GLM_HAS_BITSCAN_WINDOWS |
| 68 | template<typename genIUType> |
| 69 | struct compute_findLSB<genIUType, 32> |
| 70 | { |
| 71 | GLM_FUNC_QUALIFIER static int call(genIUType Value) |
| 72 | { |
| 73 | unsigned long Result(0); |
| 74 | unsigned char IsNotNull = _BitScanForward(&Result, *reinterpret_cast<unsigned long*>(&Value)); |
| 75 | return IsNotNull ? int(Result) : -1; |
| 76 | } |
| 77 | }; |
| 78 | |
| 79 | # if !((GLM_COMPILER & GLM_COMPILER_VC) && (GLM_MODEL == GLM_MODEL_32)) |
| 80 | template<typename genIUType> |
| 81 | struct compute_findLSB<genIUType, 64> |
| 82 | { |
| 83 | GLM_FUNC_QUALIFIER static int call(genIUType Value) |
| 84 | { |
| 85 | unsigned long Result(0); |
| 86 | unsigned char IsNotNull = _BitScanForward64(&Result, *reinterpret_cast<unsigned __int64*>(&Value)); |
| 87 | return IsNotNull ? int(Result) : -1; |
| 88 | } |
| 89 | }; |
| 90 | # endif |
| 91 | # endif//GLM_HAS_BITSCAN_WINDOWS |
| 92 | |
| 93 | template<length_t L, typename T, qualifier Q, bool EXEC = true> |
| 94 | struct compute_findMSB_step_vec |
| 95 | { |
| 96 | GLM_FUNC_QUALIFIER static vec<L, T, Q> call(vec<L, T, Q> const& x, T Shift) |
| 97 | { |
| 98 | return x | (x >> Shift); |
| 99 | } |
| 100 | }; |
| 101 | |
| 102 | template<length_t L, typename T, qualifier Q> |
| 103 | struct compute_findMSB_step_vec<L, T, Q, false> |
| 104 | { |
| 105 | GLM_FUNC_QUALIFIER static vec<L, T, Q> call(vec<L, T, Q> const& x, T) |
| 106 | { |
| 107 | return x; |
| 108 | } |
| 109 | }; |
| 110 | |
| 111 | template<length_t L, typename T, qualifier Q, int> |
| 112 | struct compute_findMSB_vec |
| 113 | { |
| 114 | GLM_FUNC_QUALIFIER static vec<L, int, Q> call(vec<L, T, Q> const& v) |
| 115 | { |
| 116 | vec<L, T, Q> x(v); |
| 117 | x = compute_findMSB_step_vec<L, T, Q, sizeof(T) * 8 >= 8>::call(x, static_cast<T>( 1)); |
| 118 | x = compute_findMSB_step_vec<L, T, Q, sizeof(T) * 8 >= 8>::call(x, static_cast<T>( 2)); |
| 119 | x = compute_findMSB_step_vec<L, T, Q, sizeof(T) * 8 >= 8>::call(x, static_cast<T>( 4)); |
| 120 | x = compute_findMSB_step_vec<L, T, Q, sizeof(T) * 8 >= 16>::call(x, static_cast<T>( 8)); |
| 121 | x = compute_findMSB_step_vec<L, T, Q, sizeof(T) * 8 >= 32>::call(x, static_cast<T>(16)); |
| 122 | x = compute_findMSB_step_vec<L, T, Q, sizeof(T) * 8 >= 64>::call(x, static_cast<T>(32)); |
| 123 | return vec<L, int, Q>(sizeof(T) * 8 - 1) - glm::bitCount(~x); |
| 124 | } |
| 125 | }; |
| 126 | |
| 127 | # if GLM_HAS_BITSCAN_WINDOWS |
| 128 | template<typename genIUType> |
| 129 | GLM_FUNC_QUALIFIER int compute_findMSB_32(genIUType Value) |
| 130 | { |
| 131 | unsigned long Result(0); |
| 132 | unsigned char IsNotNull = _BitScanReverse(&Result, *reinterpret_cast<unsigned long*>(&Value)); |
| 133 | return IsNotNull ? int(Result) : -1; |
| 134 | } |
| 135 | |
| 136 | template<length_t L, typename T, qualifier Q> |
| 137 | struct compute_findMSB_vec<L, T, Q, 32> |
| 138 | { |
| 139 | GLM_FUNC_QUALIFIER static vec<L, int, Q> call(vec<L, T, Q> const& x) |
| 140 | { |
| 141 | return detail::functor1<vec, L, int, T, Q>::call(compute_findMSB_32, x); |
| 142 | } |
| 143 | }; |
| 144 | |
| 145 | # if !((GLM_COMPILER & GLM_COMPILER_VC) && (GLM_MODEL == GLM_MODEL_32)) |
| 146 | template<typename genIUType> |
| 147 | GLM_FUNC_QUALIFIER int compute_findMSB_64(genIUType Value) |
| 148 | { |
| 149 | unsigned long Result(0); |
| 150 | unsigned char IsNotNull = _BitScanReverse64(&Result, *reinterpret_cast<unsigned __int64*>(&Value)); |
| 151 | return IsNotNull ? int(Result) : -1; |
| 152 | } |
| 153 | |
| 154 | template<length_t L, typename T, qualifier Q> |
| 155 | struct compute_findMSB_vec<L, T, Q, 64> |
| 156 | { |
| 157 | GLM_FUNC_QUALIFIER static vec<L, int, Q> call(vec<L, T, Q> const& x) |
| 158 | { |
| 159 | return detail::functor1<vec, L, int, T, Q>::call(compute_findMSB_64, x); |
| 160 | } |
| 161 | }; |
| 162 | # endif |
| 163 | # endif//GLM_HAS_BITSCAN_WINDOWS |
| 164 | }//namespace detail |
| 165 | |
| 166 | // uaddCarry |
| 167 | GLM_FUNC_QUALIFIER uint uaddCarry(uint const& x, uint const& y, uint & Carry) |
| 168 | { |
| 169 | detail::uint64 const Value64(static_cast<detail::uint64>(x) + static_cast<detail::uint64>(y)); |
| 170 | detail::uint64 const Max32((static_cast<detail::uint64>(1) << static_cast<detail::uint64>(32)) - static_cast<detail::uint64>(1)); |
| 171 | Carry = Value64 > Max32 ? 1u : 0u; |
| 172 | return static_cast<uint>(Value64 % (Max32 + static_cast<detail::uint64>(1))); |
| 173 | } |
| 174 | |
| 175 | template<length_t L, qualifier Q> |
| 176 | GLM_FUNC_QUALIFIER vec<L, uint, Q> uaddCarry(vec<L, uint, Q> const& x, vec<L, uint, Q> const& y, vec<L, uint, Q>& Carry) |
| 177 | { |
| 178 | vec<L, detail::uint64, Q> Value64(vec<L, detail::uint64, Q>(x) + vec<L, detail::uint64, Q>(y)); |
| 179 | vec<L, detail::uint64, Q> Max32((static_cast<detail::uint64>(1) << static_cast<detail::uint64>(32)) - static_cast<detail::uint64>(1)); |
| 180 | Carry = mix(vec<L, uint, Q>(0), vec<L, uint, Q>(1), greaterThan(Value64, Max32)); |
| 181 | return vec<L, uint, Q>(Value64 % (Max32 + static_cast<detail::uint64>(1))); |
| 182 | } |
| 183 | |
| 184 | // usubBorrow |
| 185 | GLM_FUNC_QUALIFIER uint usubBorrow(uint const& x, uint const& y, uint & Borrow) |
| 186 | { |
| 187 | Borrow = x >= y ? static_cast<uint>(0) : static_cast<uint>(1); |
| 188 | if(x >= y) |
| 189 | return x - y; |
| 190 | else |
| 191 | return static_cast<uint>((static_cast<detail::int64>(1) << static_cast<detail::int64>(32)) + (static_cast<detail::int64>(x) - static_cast<detail::int64>(y))); |
| 192 | } |
| 193 | |
| 194 | template<length_t L, qualifier Q> |
| 195 | GLM_FUNC_QUALIFIER vec<L, uint, Q> usubBorrow(vec<L, uint, Q> const& x, vec<L, uint, Q> const& y, vec<L, uint, Q>& Borrow) |
| 196 | { |
| 197 | Borrow = mix(vec<L, uint, Q>(1), vec<L, uint, Q>(0), greaterThanEqual(x, y)); |
| 198 | vec<L, uint, Q> const XgeY(x - y); |
| 199 | vec<L, uint, Q> const YgX(vec<L, uint, Q>((static_cast<detail::int64>(1) << static_cast<detail::int64>(32)) + (vec<L, detail::int64, Q>(x) - vec<L, detail::int64, Q>(y)))); |
| 200 | return mix(YgX, XgeY, greaterThanEqual(x, y)); |
| 201 | } |
| 202 | |
| 203 | // umulExtended |
| 204 | GLM_FUNC_QUALIFIER void umulExtended(uint const& x, uint const& y, uint & msb, uint & lsb) |
| 205 | { |
| 206 | detail::uint64 Value64 = static_cast<detail::uint64>(x) * static_cast<detail::uint64>(y); |
| 207 | msb = static_cast<uint>(Value64 >> static_cast<detail::uint64>(32)); |
| 208 | lsb = static_cast<uint>(Value64); |
| 209 | } |
| 210 | |
| 211 | template<length_t L, qualifier Q> |
| 212 | GLM_FUNC_QUALIFIER void umulExtended(vec<L, uint, Q> const& x, vec<L, uint, Q> const& y, vec<L, uint, Q>& msb, vec<L, uint, Q>& lsb) |
| 213 | { |
| 214 | vec<L, detail::uint64, Q> Value64(vec<L, detail::uint64, Q>(x) * vec<L, detail::uint64, Q>(y)); |
| 215 | msb = vec<L, uint, Q>(Value64 >> static_cast<detail::uint64>(32)); |
| 216 | lsb = vec<L, uint, Q>(Value64); |
| 217 | } |
| 218 | |
| 219 | // imulExtended |
| 220 | GLM_FUNC_QUALIFIER void imulExtended(int x, int y, int& msb, int& lsb) |
| 221 | { |
| 222 | detail::int64 Value64 = static_cast<detail::int64>(x) * static_cast<detail::int64>(y); |
| 223 | msb = static_cast<int>(Value64 >> static_cast<detail::int64>(32)); |
| 224 | lsb = static_cast<int>(Value64); |
| 225 | } |
| 226 | |
| 227 | template<length_t L, qualifier Q> |
| 228 | GLM_FUNC_QUALIFIER void imulExtended(vec<L, int, Q> const& x, vec<L, int, Q> const& y, vec<L, int, Q>& msb, vec<L, int, Q>& lsb) |
| 229 | { |
| 230 | vec<L, detail::int64, Q> Value64(vec<L, detail::int64, Q>(x) * vec<L, detail::int64, Q>(y)); |
| 231 | lsb = vec<L, int, Q>(Value64 & static_cast<detail::int64>(0xFFFFFFFF)); |
| 232 | msb = vec<L, int, Q>((Value64 >> static_cast<detail::int64>(32)) & static_cast<detail::int64>(0xFFFFFFFF)); |
| 233 | } |
| 234 | |
| 235 | // bitfieldExtract |
| 236 | template<typename genIUType> |
| 237 | GLM_FUNC_QUALIFIER genIUType bitfieldExtract(genIUType Value, int Offset, int Bits) |
| 238 | { |
| 239 | return bitfieldExtract(vec<1, genIUType>(Value), Offset, Bits).x; |
| 240 | } |
| 241 | |
| 242 | template<length_t L, typename T, qualifier Q> |
| 243 | GLM_FUNC_QUALIFIER vec<L, T, Q> bitfieldExtract(vec<L, T, Q> const& Value, int Offset, int Bits) |
| 244 | { |
| 245 | static_assert(std::numeric_limits<T>::is_integer, "'bitfieldExtract' only accept integer inputs"); |
| 246 | |
| 247 | return (Value >> static_cast<T>(Offset)) & static_cast<T>(detail::mask(Bits)); |
| 248 | } |
| 249 | |
| 250 | // bitfieldInsert |
| 251 | template<typename genIUType> |
| 252 | GLM_FUNC_QUALIFIER genIUType bitfieldInsert(genIUType const& Base, genIUType const& Insert, int Offset, int Bits) |
| 253 | { |
| 254 | static_assert(std::numeric_limits<genIUType>::is_integer, "'bitfieldInsert' only accept integer values"); |
| 255 | |
| 256 | return bitfieldInsert(vec<1, genIUType>(Base), vec<1, genIUType>(Insert), Offset, Bits).x; |
| 257 | } |
| 258 | |
| 259 | template<length_t L, typename T, qualifier Q> |
| 260 | GLM_FUNC_QUALIFIER vec<L, T, Q> bitfieldInsert(vec<L, T, Q> const& Base, vec<L, T, Q> const& Insert, int Offset, int Bits) |
| 261 | { |
| 262 | static_assert(std::numeric_limits<T>::is_integer, "'bitfieldInsert' only accept integer values"); |
| 263 | |
| 264 | T const Mask = detail::mask(static_cast<T>(Bits)) << Offset; |
| 265 | return (Base & ~Mask) | ((Insert << static_cast<T>(Offset)) & Mask); |
| 266 | } |
| 267 | |
| 268 | #if GLM_COMPILER & GLM_COMPILER_VC |
| 269 | # pragma warning(push) |
| 270 | # pragma warning(disable : 4309) |
| 271 | #endif |
| 272 | |
| 273 | // bitfieldReverse |
| 274 | template<typename genIUType> |
| 275 | GLM_FUNC_QUALIFIER genIUType bitfieldReverse(genIUType x) |
| 276 | { |
| 277 | static_assert(std::numeric_limits<genIUType>::is_integer, "'bitfieldReverse' only accept integer values"); |
| 278 | |
| 279 | return bitfieldReverse(glm::vec<1, genIUType, glm::defaultp>(x)).x; |
| 280 | } |
| 281 | |
| 282 | template<length_t L, typename T, qualifier Q> |
| 283 | GLM_FUNC_QUALIFIER vec<L, T, Q> bitfieldReverse(vec<L, T, Q> const& v) |
| 284 | { |
| 285 | static_assert(std::numeric_limits<T>::is_integer, "'bitfieldReverse' only accept integer values"); |
| 286 | |
| 287 | vec<L, T, Q> x(v); |
| 288 | x = detail::compute_bitfieldReverseStep<L, T, Q, detail::is_aligned<Q>::value, sizeof(T) * 8>= 2>::call(x, static_cast<T>(0x5555555555555555ull), static_cast<T>( 1)); |
| 289 | x = detail::compute_bitfieldReverseStep<L, T, Q, detail::is_aligned<Q>::value, sizeof(T) * 8>= 4>::call(x, static_cast<T>(0x3333333333333333ull), static_cast<T>( 2)); |
| 290 | x = detail::compute_bitfieldReverseStep<L, T, Q, detail::is_aligned<Q>::value, sizeof(T) * 8>= 8>::call(x, static_cast<T>(0x0F0F0F0F0F0F0F0Full), static_cast<T>( 4)); |
| 291 | x = detail::compute_bitfieldReverseStep<L, T, Q, detail::is_aligned<Q>::value, sizeof(T) * 8>= 16>::call(x, static_cast<T>(0x00FF00FF00FF00FFull), static_cast<T>( 8)); |
| 292 | x = detail::compute_bitfieldReverseStep<L, T, Q, detail::is_aligned<Q>::value, sizeof(T) * 8>= 32>::call(x, static_cast<T>(0x0000FFFF0000FFFFull), static_cast<T>(16)); |
| 293 | x = detail::compute_bitfieldReverseStep<L, T, Q, detail::is_aligned<Q>::value, sizeof(T) * 8>= 64>::call(x, static_cast<T>(0x00000000FFFFFFFFull), static_cast<T>(32)); |
| 294 | return x; |
| 295 | } |
| 296 | |
| 297 | # if GLM_COMPILER & GLM_COMPILER_VC |
| 298 | # pragma warning(pop) |
| 299 | # endif |
| 300 | |
| 301 | // bitCount |
| 302 | template<typename genIUType> |
| 303 | GLM_FUNC_QUALIFIER int bitCount(genIUType x) |
| 304 | { |
| 305 | static_assert(std::numeric_limits<genIUType>::is_integer, "'bitCount' only accept integer values"); |
| 306 | |
| 307 | return bitCount(glm::vec<1, genIUType, glm::defaultp>(x)).x; |
| 308 | } |
| 309 | |
| 310 | template<length_t L, typename T, qualifier Q> |
| 311 | GLM_FUNC_QUALIFIER vec<L, int, Q> bitCount(vec<L, T, Q> const& v) |
| 312 | { |
| 313 | static_assert(std::numeric_limits<T>::is_integer, "'bitCount' only accept integer values"); |
| 314 | |
| 315 | # if GLM_COMPILER & GLM_COMPILER_VC |
| 316 | # pragma warning(push) |
| 317 | # pragma warning(disable : 4310) //cast truncates constant value |
| 318 | # endif |
| 319 | |
| 320 | vec<L, typename std::make_unsigned<T>::type, Q> x(v); |
| 321 | x = detail::compute_bitfieldBitCountStep<L, typename std::make_unsigned<T>::type, Q, detail::is_aligned<Q>::value, sizeof(T) * 8>= 2>::call(x, typename std::make_unsigned<T>::type(0x5555555555555555ull), typename std::make_unsigned<T>::type( 1)); |
| 322 | x = detail::compute_bitfieldBitCountStep<L, typename std::make_unsigned<T>::type, Q, detail::is_aligned<Q>::value, sizeof(T) * 8>= 4>::call(x, typename std::make_unsigned<T>::type(0x3333333333333333ull), typename std::make_unsigned<T>::type( 2)); |
| 323 | x = detail::compute_bitfieldBitCountStep<L, typename std::make_unsigned<T>::type, Q, detail::is_aligned<Q>::value, sizeof(T) * 8>= 8>::call(x, typename std::make_unsigned<T>::type(0x0F0F0F0F0F0F0F0Full), typename std::make_unsigned<T>::type( 4)); |
| 324 | x = detail::compute_bitfieldBitCountStep<L, typename std::make_unsigned<T>::type, Q, detail::is_aligned<Q>::value, sizeof(T) * 8>= 16>::call(x, typename std::make_unsigned<T>::type(0x00FF00FF00FF00FFull), typename std::make_unsigned<T>::type( 8)); |
| 325 | x = detail::compute_bitfieldBitCountStep<L, typename std::make_unsigned<T>::type, Q, detail::is_aligned<Q>::value, sizeof(T) * 8>= 32>::call(x, typename std::make_unsigned<T>::type(0x0000FFFF0000FFFFull), typename std::make_unsigned<T>::type(16)); |
| 326 | x = detail::compute_bitfieldBitCountStep<L, typename std::make_unsigned<T>::type, Q, detail::is_aligned<Q>::value, sizeof(T) * 8>= 64>::call(x, typename std::make_unsigned<T>::type(0x00000000FFFFFFFFull), typename std::make_unsigned<T>::type(32)); |
| 327 | return vec<L, int, Q>(x); |
| 328 | |
| 329 | # if GLM_COMPILER & GLM_COMPILER_VC |
| 330 | # pragma warning(pop) |
| 331 | # endif |
| 332 | } |
| 333 | |
| 334 | // findLSB |
| 335 | template<typename genIUType> |
| 336 | GLM_FUNC_QUALIFIER int findLSB(genIUType Value) |
| 337 | { |
| 338 | static_assert(std::numeric_limits<genIUType>::is_integer, "'findLSB' only accept integer values"); |
| 339 | |
| 340 | return detail::compute_findLSB<genIUType, sizeof(genIUType) * 8>::call(Value); |
| 341 | } |
| 342 | |
| 343 | template<length_t L, typename T, qualifier Q> |
| 344 | GLM_FUNC_QUALIFIER vec<L, int, Q> findLSB(vec<L, T, Q> const& x) |
| 345 | { |
| 346 | static_assert(std::numeric_limits<T>::is_integer, "'findLSB' only accept integer values"); |
| 347 | |
| 348 | return detail::functor1<vec, L, int, T, Q>::call(findLSB, x); |
| 349 | } |
| 350 | |
| 351 | // findMSB |
| 352 | template<typename genIUType> |
| 353 | GLM_FUNC_QUALIFIER int findMSB(genIUType v) |
| 354 | { |
| 355 | static_assert(std::numeric_limits<genIUType>::is_integer, "'findMSB' only accept integer values"); |
| 356 | |
| 357 | return findMSB(vec<1, genIUType>(v)).x; |
| 358 | } |
| 359 | |
| 360 | template<length_t L, typename T, qualifier Q> |
| 361 | GLM_FUNC_QUALIFIER vec<L, int, Q> findMSB(vec<L, T, Q> const& v) |
| 362 | { |
| 363 | static_assert(std::numeric_limits<T>::is_integer, "'findMSB' only accept integer values"); |
| 364 | |
| 365 | return detail::compute_findMSB_vec<L, T, Q, static_cast<int>(sizeof(T) * 8)>::call(v); |
| 366 | } |
| 367 | }//namespace glm |
| 368 | |
| 369 | #if GLM_CONFIG_SIMD == GLM_ENABLE |
| 370 | # include "func_integer_simd.inl" |
| 371 | #endif |
| 372 | |