Back to home page

EIC code displayed by LXR
 
 

    

Warning, file /include/Vc/global.h was not indexed or was modified since last indexation (in which case cross-reference links may be missing, inaccurate or erroneous).

0001 /*  This file is part of the Vc library. {{{
0002 Copyright © 2009-2015 Matthias Kretz <kretz@kde.org>
0003 
0004 Redistribution and use in source and binary forms, with or without
0005 modification, are permitted provided that the following conditions are met:
0006     * Redistributions of source code must retain the above copyright
0007       notice, this list of conditions and the following disclaimer.
0008     * Redistributions in binary form must reproduce the above copyright
0009       notice, this list of conditions and the following disclaimer in the
0010       documentation and/or other materials provided with the distribution.
0011     * Neither the names of contributing organizations nor the
0012       names of its contributors may be used to endorse or promote products
0013       derived from this software without specific prior written permission.
0014 
0015 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
0016 ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
0017 WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
0018 DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE FOR ANY
0019 DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
0020 (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
0021 LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
0022 ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
0023 (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
0024 SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
0025 
0026 }}}*/
0027 
0028 #ifndef VC_GLOBAL_H_
0029 #define VC_GLOBAL_H_
0030 
0031 #include <cstdint>
0032 #include "fwddecl.h"
0033 
0034 #ifdef DOXYGEN
0035 
0036 /**
0037  * \name Compiler Identification Macros
0038  * \ingroup Utilities
0039  */
0040 //@{
0041 /**
0042  * \ingroup Utilities
0043  * This macro is defined to a number identifying the ICC version if the current
0044  * translation unit is compiled with the Intel compiler.
0045  *
0046  * For any other compiler this macro is not defined.
0047  */
0048 #define Vc_ICC __INTEL_COMPILER_BUILD_DATE
0049 #undef Vc_ICC
0050 /**
0051  * \ingroup Utilities
0052  * This macro is defined to a number identifying the Clang version if the current
0053  * translation unit is compiled with the Clang compiler.
0054  *
0055  * For any other compiler this macro is not defined.
0056  */
0057 #define Vc_CLANG (__clang_major__ * 0x10000 + __clang_minor__ * 0x100 + __clang_patchlevel__)
0058 #undef Vc_CLANG
0059 /**
0060  * \ingroup Utilities
0061  * This macro is defined to a number identifying the Apple Clang version if the current
0062  * translation unit is compiled with the Apple Clang compiler.
0063  *
0064  * For any other compiler this macro is not defined.
0065  */
0066 #define Vc_APPLECLANG (__clang_major__ * 0x10000 + __clang_minor__ * 0x100 + __clang_patchlevel__)
0067 #undef Vc_APPLECLANG
0068 /**
0069  * \ingroup Utilities
0070  * This macro is defined to a number identifying the GCC version if the current
0071  * translation unit is compiled with the GCC compiler.
0072  *
0073  * For any other compiler this macro is not defined.
0074  */
0075 #define Vc_GCC (__GNUC__ * 0x10000 + __GNUC_MINOR__ * 0x100 + __GNUC_PATCHLEVEL__)
0076 /**
0077  * \ingroup Utilities
0078  * This macro is defined to a number identifying the Microsoft Visual C++ version if
0079  * the current translation unit is compiled with the Visual C++ (MSVC) compiler.
0080  *
0081  * For any other compiler this macro is not defined.
0082  */
0083 #define Vc_MSVC _MSC_FULL_VER
0084 #undef Vc_MSVC
0085 //@}
0086 
0087 #else  // DOXYGEN
0088 
0089 // Compiler defines
0090 #ifdef __INTEL_COMPILER
0091 #define Vc_ICC __INTEL_COMPILER_BUILD_DATE
0092 #elif defined(__clang__) && defined(__apple_build_version__)
0093 #define Vc_APPLECLANG (__clang_major__ * 0x10000 + __clang_minor__ * 0x100 + __clang_patchlevel__)
0094 #elif defined(__clang__)
0095 #define Vc_CLANG (__clang_major__ * 0x10000 + __clang_minor__ * 0x100 + __clang_patchlevel__)
0096 #elif defined(__GNUC__)
0097 #define Vc_GCC (__GNUC__ * 0x10000 + __GNUC_MINOR__ * 0x100 + __GNUC_PATCHLEVEL__)
0098 #elif defined(_MSC_VER)
0099 #define Vc_MSVC _MSC_FULL_VER
0100 #else
0101 #define Vc_UNSUPPORTED_COMPILER 1
0102 #endif
0103 
0104 #if defined Vc_GCC && Vc_GCC >= 0x60000
0105 #define Vc_RESET_DIAGNOSTICS _Pragma("GCC diagnostic pop")
0106 #pragma GCC diagnostic push
0107 #pragma GCC diagnostic ignored "-Wignored-attributes"
0108 #else
0109 #define Vc_RESET_DIAGNOSTICS
0110 #endif
0111 
0112 #if defined Vc_ICC
0113 // 'warning #2922: template parameter "<unnamed>" cannot be used because it follows a
0114 // parameter pack and cannot be deduced from the parameters of function template'
0115 // This warning is stupid. The parameter is unnamed because I don't want to use it. I see
0116 // no other workaround than to disable the warning. Sadly, it doesn't suffice to disable
0117 // it for the Vc headers. It must also be disabled at the places Vc types are used.
0118 #pragma warning disable 2922
0119 #endif
0120 
0121 #if __cplusplus < 201103 && (!defined Vc_MSVC || _MSC_VER < 1900)
0122 # error "Vc requires support for C++11."
0123 #elif __cplusplus >= 201402L
0124 # define Vc_CXX14 1
0125 # if __cplusplus > 201700L
0126 #  define Vc_CXX17 1
0127 # endif
0128 #endif
0129 
0130 #if defined(__GNUC__) && !defined(Vc_NO_INLINE_ASM)
0131 #define Vc_GNU_ASM 1
0132 #endif
0133 
0134 #ifdef Vc_GCC
0135 #  if Vc_GCC >= 0x70000 && defined __i386__
0136      // GCC 7 changed alignof(max_align_t) to 16. glibc 2.26 followed with malloc in 2.26.
0137      // 1. If GCC >= 7 and libc is not glibc max_align_t and malloc mismatch
0138      // 2. If GCC >= 7 and libc is glibc < 2.26 max_align_t and malloc mismatch
0139 #    ifdef __GLIBC_PREREQ
0140 #      if __GLIBC_PREREQ(2,26)
0141 #        define Vc_HAVE_STD_MAX_ALIGN_T 1
0142 #      endif
0143 #    endif
0144 #  elif Vc_GCC >= 0x40900
0145 #    define Vc_HAVE_STD_MAX_ALIGN_T 1
0146 #  else
0147 #    define Vc_HAVE_MAX_ALIGN_T 1
0148 #  endif
0149 #elif !defined(Vc_CLANG) && !defined(Vc_ICC)
0150 //   Clang/ICC don't provide max_align_t at all
0151 //   TODO: Clang defines max_align_t since 3.5.0. Whether std::max_align_t is defined depends on the
0152 //   standard library version.
0153 #  define Vc_HAVE_STD_MAX_ALIGN_T 1
0154 #endif
0155 
0156 #if defined(Vc_GCC) || defined(Vc_CLANG) || defined Vc_APPLECLANG
0157 #define Vc_USE_BUILTIN_VECTOR_TYPES 1
0158 #endif
0159 
0160 #ifdef Vc_MSVC
0161 #  define Vc_CDECL __cdecl
0162 #  define Vc_VDECL __vectorcall
0163 #else
0164 #  define Vc_CDECL
0165 #  define Vc_VDECL
0166 #endif
0167 
0168 /* Define the following strings to a unique integer, which is the only type the preprocessor can
0169  * compare. This allows to use -DVc_IMPL=SSE3. The preprocessor will then consider Vc_IMPL and SSE3
0170  * to be equal. Of course, it is important to undefine the strings later on!
0171  */
0172 #define Scalar 0x00100000
0173 #define SSE    0x00200000
0174 #define SSE2   0x00300000
0175 #define SSE3   0x00400000
0176 #define SSSE3  0x00500000
0177 #define SSE4_1 0x00600000
0178 #define SSE4_2 0x00700000
0179 #define AVX    0x00800000
0180 #define AVX2   0x00900000
0181 
0182 #define XOP    0x00000001
0183 #define FMA4   0x00000002
0184 #define F16C   0x00000004
0185 #define POPCNT 0x00000008
0186 #define SSE4a  0x00000010
0187 #define FMA    0x00000020
0188 #define BMI2   0x00000040
0189 
0190 #define IMPL_MASK 0xFFF00000
0191 #define EXT_MASK  0x000FFFFF
0192 
0193 #ifdef Vc_MSVC
0194 # ifdef _M_IX86_FP
0195 #  if _M_IX86_FP >= 1
0196 #   ifndef __SSE__
0197 #    define __SSE__ 1
0198 #   endif
0199 #  endif
0200 #  if _M_IX86_FP >= 2
0201 #   ifndef __SSE2__
0202 #    define __SSE2__ 1
0203 #   endif
0204 #  endif
0205 # elif defined(_M_AMD64)
0206 // If the target is x86_64 then SSE2 is guaranteed
0207 #  ifndef __SSE__
0208 #   define __SSE__ 1
0209 #  endif
0210 #  ifndef __SSE2__
0211 #   define __SSE2__ 1
0212 #  endif
0213 # endif
0214 #endif
0215 
0216 #if defined Vc_ICC && !defined __POPCNT__
0217 # if defined __SSE4_2__ || defined __SSE4A__
0218 #  define __POPCNT__ 1
0219 # endif
0220 #endif
0221 
0222 #ifdef VC_IMPL
0223 #error "You are using the old VC_IMPL macro. Since Vc 1.0 all Vc macros start with Vc_, i.e. a lower-case 'c'"
0224 #endif
0225 
0226 #ifndef Vc_IMPL
0227 
0228 #  if defined(__AVX2__)
0229 #    define Vc_IMPL_AVX2 1
0230 #    define Vc_IMPL_AVX 1
0231 #  elif defined(__AVX__)
0232 #    define Vc_IMPL_AVX 1
0233 #  else
0234 #    if defined(__SSE4_2__)
0235 #      define Vc_IMPL_SSE 1
0236 #      define Vc_IMPL_SSE4_2 1
0237 #    endif
0238 #    if defined(__SSE4_1__)
0239 #      define Vc_IMPL_SSE 1
0240 #      define Vc_IMPL_SSE4_1 1
0241 #    endif
0242 #    if defined(__SSE3__)
0243 #      define Vc_IMPL_SSE 1
0244 #      define Vc_IMPL_SSE3 1
0245 #    endif
0246 #    if defined(__SSSE3__)
0247 #      define Vc_IMPL_SSE 1
0248 #      define Vc_IMPL_SSSE3 1
0249 #    endif
0250 #    if defined(__SSE2__)
0251 #      define Vc_IMPL_SSE 1
0252 #      define Vc_IMPL_SSE2 1
0253 #    endif
0254 
0255 #    if defined(Vc_IMPL_SSE)
0256        // nothing
0257 #    else
0258 #      define Vc_IMPL_Scalar 1
0259 #    endif
0260 #  endif
0261 #  if !defined(Vc_IMPL_Scalar)
0262 #    ifdef __FMA4__
0263 #      define Vc_IMPL_FMA4 1
0264 #    endif
0265 #    ifdef __XOP__
0266 #      define Vc_IMPL_XOP 1
0267 #    endif
0268 #    ifdef __F16C__
0269 #      define Vc_IMPL_F16C 1
0270 #    endif
0271 #    ifdef __POPCNT__
0272 #      define Vc_IMPL_POPCNT 1
0273 #    endif
0274 #    ifdef __SSE4A__
0275 #      define Vc_IMPL_SSE4a 1
0276 #    endif
0277 #    ifdef __FMA__
0278 #      define Vc_IMPL_FMA 1
0279 #    endif
0280 #    ifdef __BMI2__
0281 #      define Vc_IMPL_BMI2 1
0282 #    endif
0283 #  endif
0284 
0285 #else // Vc_IMPL
0286 
0287 #  if (Vc_IMPL & IMPL_MASK) == AVX2 // AVX2 supersedes SSE
0288 #    define Vc_IMPL_AVX2 1
0289 #    define Vc_IMPL_AVX 1
0290 #  elif (Vc_IMPL & IMPL_MASK) == AVX // AVX supersedes SSE
0291 #    define Vc_IMPL_AVX 1
0292 #  elif (Vc_IMPL & IMPL_MASK) == Scalar
0293 #    define Vc_IMPL_Scalar 1
0294 #  elif (Vc_IMPL & IMPL_MASK) == SSE4_2
0295 #    define Vc_IMPL_SSE4_2 1
0296 #    define Vc_IMPL_SSE4_1 1
0297 #    define Vc_IMPL_SSSE3 1
0298 #    define Vc_IMPL_SSE3 1
0299 #    define Vc_IMPL_SSE2 1
0300 #    define Vc_IMPL_SSE 1
0301 #  elif (Vc_IMPL & IMPL_MASK) == SSE4_1
0302 #    define Vc_IMPL_SSE4_1 1
0303 #    define Vc_IMPL_SSSE3 1
0304 #    define Vc_IMPL_SSE3 1
0305 #    define Vc_IMPL_SSE2 1
0306 #    define Vc_IMPL_SSE 1
0307 #  elif (Vc_IMPL & IMPL_MASK) == SSSE3
0308 #    define Vc_IMPL_SSSE3 1
0309 #    define Vc_IMPL_SSE3 1
0310 #    define Vc_IMPL_SSE2 1
0311 #    define Vc_IMPL_SSE 1
0312 #  elif (Vc_IMPL & IMPL_MASK) == SSE3
0313 #    define Vc_IMPL_SSE3 1
0314 #    define Vc_IMPL_SSE2 1
0315 #    define Vc_IMPL_SSE 1
0316 #  elif (Vc_IMPL & IMPL_MASK) == SSE2
0317 #    define Vc_IMPL_SSE2 1
0318 #    define Vc_IMPL_SSE 1
0319 #  elif (Vc_IMPL & IMPL_MASK) == SSE
0320 #    define Vc_IMPL_SSE 1
0321 #    if defined(__SSE4_2__)
0322 #      define Vc_IMPL_SSE4_2 1
0323 #    endif
0324 #    if defined(__SSE4_1__)
0325 #      define Vc_IMPL_SSE4_1 1
0326 #    endif
0327 #    if defined(__SSE3__)
0328 #      define Vc_IMPL_SSE3 1
0329 #    endif
0330 #    if defined(__SSSE3__)
0331 #      define Vc_IMPL_SSSE3 1
0332 #    endif
0333 #    if defined(__SSE2__)
0334 #      define Vc_IMPL_SSE2 1
0335 #    endif
0336 #  elif (Vc_IMPL & IMPL_MASK) == 0 && (Vc_IMPL & SSE4a)
0337      // this is for backward compatibility only where SSE4a was included in the main
0338      // line of available SIMD instruction sets
0339 #    define Vc_IMPL_SSE3 1
0340 #    define Vc_IMPL_SSE2 1
0341 #    define Vc_IMPL_SSE 1
0342 #  endif
0343 #  if (Vc_IMPL & XOP)
0344 #    define Vc_IMPL_XOP 1
0345 #  endif
0346 #  if (Vc_IMPL & FMA4)
0347 #    define Vc_IMPL_FMA4 1
0348 #  endif
0349 #  if (Vc_IMPL & F16C)
0350 #    define Vc_IMPL_F16C 1
0351 #  endif
0352 #  if (!defined(Vc_IMPL_Scalar) && defined(__POPCNT__)) || (Vc_IMPL & POPCNT)
0353 #    define Vc_IMPL_POPCNT 1
0354 #  endif
0355 #  if (Vc_IMPL & SSE4a)
0356 #    define Vc_IMPL_SSE4a 1
0357 #  endif
0358 #  if (Vc_IMPL & FMA)
0359 #    define Vc_IMPL_FMA 1
0360 #  endif
0361 #  if (Vc_IMPL & BMI2)
0362 #    define Vc_IMPL_BMI2 1
0363 #  endif
0364 #  undef Vc_IMPL
0365 
0366 #endif // Vc_IMPL
0367 
0368 // If AVX is enabled in the compiler it will use VEX coding for the SIMD instructions.
0369 #ifdef __AVX__
0370 #  define Vc_USE_VEX_CODING 1
0371 #endif
0372 
0373 #ifdef Vc_IMPL_AVX
0374 // if we have AVX then we also have all SSE intrinsics
0375 #    define Vc_IMPL_SSE4_2 1
0376 #    define Vc_IMPL_SSE4_1 1
0377 #    define Vc_IMPL_SSSE3 1
0378 #    define Vc_IMPL_SSE3 1
0379 #    define Vc_IMPL_SSE2 1
0380 #    define Vc_IMPL_SSE 1
0381 #endif
0382 
0383 #if defined(Vc_CLANG) && Vc_CLANG >= 0x30600 && Vc_CLANG < 0x30700
0384 #    if defined(Vc_IMPL_AVX)
0385 #        warning "clang 3.6.x miscompiles AVX code, frequently losing 50% of the data. Vc will fall back to SSE4 instead."
0386 #        undef Vc_IMPL_AVX
0387 #        if defined(Vc_IMPL_AVX2)
0388 #            undef Vc_IMPL_AVX2
0389 #        endif
0390 #    endif
0391 #endif
0392 
0393 # if !defined(Vc_IMPL_Scalar) && !defined(Vc_IMPL_SSE) && !defined(Vc_IMPL_AVX)
0394 #  error "No suitable Vc implementation was selected! Probably Vc_IMPL was set to an invalid value."
0395 # elif defined(Vc_IMPL_SSE) && !defined(Vc_IMPL_SSE2)
0396 #  error "SSE requested but no SSE2 support. Vc needs at least SSE2!"
0397 # endif
0398 
0399 #undef Scalar
0400 #undef SSE
0401 #undef SSE2
0402 #undef SSE3
0403 #undef SSSE3
0404 #undef SSE4_1
0405 #undef SSE4_2
0406 #undef AVX
0407 #undef AVX2
0408 
0409 #undef XOP
0410 #undef FMA4
0411 #undef F16C
0412 #undef POPCNT
0413 #undef SSE4a
0414 #undef FMA
0415 #undef BMI2
0416 
0417 #undef IMPL_MASK
0418 #undef EXT_MASK
0419 
0420 #if defined Vc_IMPL_AVX2
0421 #define Vc_DEFAULT_IMPL_AVX2
0422 #elif defined Vc_IMPL_AVX
0423 #define Vc_DEFAULT_IMPL_AVX
0424 #elif defined Vc_IMPL_SSE
0425 #define Vc_DEFAULT_IMPL_SSE
0426 #elif defined Vc_IMPL_Scalar
0427 #define Vc_DEFAULT_IMPL_Scalar
0428 #else
0429 #error "Preprocessor logic broken. Please report a bug."
0430 #endif
0431 
0432 #endif // DOXYGEN
0433 
0434 namespace Vc_VERSIONED_NAMESPACE
0435 {
0436 
0437 typedef   signed char        int8_t;
0438 typedef unsigned char       uint8_t;
0439 typedef   signed short      int16_t;
0440 typedef unsigned short     uint16_t;
0441 typedef   signed int        int32_t;
0442 typedef unsigned int       uint32_t;
0443 typedef   signed long long  int64_t;
0444 typedef unsigned long long uint64_t;
0445 
0446 /**
0447  * \ingroup Utilities
0448  *
0449  * Enum that specifies the alignment and padding restrictions to use for memory allocation with
0450  * Vc::malloc.
0451  */
0452 enum MallocAlignment {
0453     /**
0454      * Align on boundary of vector sizes (e.g. 16 Bytes on SSE platforms) and pad to allow
0455      * vector access to the end. Thus the allocated memory contains a multiple of
0456      * VectorAlignment bytes.
0457      */
0458     AlignOnVector,
0459     /**
0460      * Align on boundary of cache line sizes (e.g. 64 Bytes on x86) and pad to allow
0461      * full cache line access to the end. Thus the allocated memory contains a multiple of
0462      * 64 bytes.
0463      */
0464     AlignOnCacheline,
0465     /**
0466      * Align on boundary of page sizes (e.g. 4096 Bytes on x86) and pad to allow
0467      * full page access to the end. Thus the allocated memory contains a multiple of
0468      * 4096 bytes.
0469      */
0470     AlignOnPage
0471 };
0472 
0473 /**
0474  * \ingroup Utilities
0475  *
0476  * Enum to identify a certain SIMD instruction set.
0477  *
0478  * You can use \ref CurrentImplementation for the currently active implementation.
0479  *
0480  * \see ExtraInstructions
0481  */
0482 enum Implementation : std::uint_least32_t { // TODO: make enum class
0483     /// uses only fundamental types
0484     ScalarImpl,
0485     /// x86 SSE + SSE2
0486     SSE2Impl,
0487     /// x86 SSE + SSE2 + SSE3
0488     SSE3Impl,
0489     /// x86 SSE + SSE2 + SSE3 + SSSE3
0490     SSSE3Impl,
0491     /// x86 SSE + SSE2 + SSE3 + SSSE3 + SSE4.1
0492     SSE41Impl,
0493     /// x86 SSE + SSE2 + SSE3 + SSSE3 + SSE4.1 + SSE4.2
0494     SSE42Impl,
0495     /// x86 AVX
0496     AVXImpl,
0497     /// x86 AVX + AVX2
0498     AVX2Impl,
0499     /// Intel Xeon Phi
0500     MICImpl,
0501     ImplementationMask = 0xfff
0502 };
0503 
0504 /**
0505  * \ingroup Utilities
0506  *
0507  * The list of available instructions is not easily described by a linear list of instruction sets.
0508  * On x86 the following instruction sets always include their predecessors:
0509  * SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, AVX, AVX2
0510  *
0511  * But there are additional instructions that are not necessarily required by this list. These are
0512  * covered in this enum.
0513  */
0514 enum ExtraInstructions : std::uint_least32_t { // TODO: make enum class
0515     //! Support for float16 conversions in hardware
0516     Float16cInstructions  = 0x01000,
0517     //! Support for FMA4 instructions
0518     Fma4Instructions      = 0x02000,
0519     //! Support for XOP instructions
0520     XopInstructions       = 0x04000,
0521     //! Support for the population count instruction
0522     PopcntInstructions    = 0x08000,
0523     //! Support for SSE4a instructions
0524     Sse4aInstructions     = 0x10000,
0525     //! Support for FMA instructions (3 operand variant)
0526     FmaInstructions       = 0x20000,
0527     //! Support for ternary instruction coding (VEX)
0528     VexInstructions       = 0x40000,
0529     //! Support for BMI2 instructions
0530     Bmi2Instructions      = 0x80000,
0531     // PclmulqdqInstructions,
0532     // AesInstructions,
0533     // RdrandInstructions
0534     ExtraInstructionsMask = 0xfffff000u
0535 };
0536 
0537 /**
0538  * \ingroup Utilities
0539  * This class identifies the specific implementation %Vc uses in the current translation
0540  * unit in terms of a type.
0541  *
0542  * Most importantantly, the type \ref CurrentImplementation instantiates the class
0543  * template with the bitmask identifying the current implementation. The contents of the
0544  * bitmask can be queried with the static member functions of the class.
0545  */
0546 template <unsigned int Features> struct ImplementationT {
0547     /// Returns the currently used Vc::Implementation.
0548     static constexpr Implementation current()
0549     {
0550         return static_cast<Implementation>(Features & ImplementationMask);
0551     }
0552     /// Returns whether \p impl is the current Vc::Implementation.
0553     static constexpr bool is(Implementation impl)
0554     {
0555         return static_cast<unsigned int>(impl) == current();
0556     }
0557     /**
0558      * Returns whether the current Vc::Implementation implements at least \p low and at
0559      * most \p high.
0560      */
0561     static constexpr bool is_between(Implementation low, Implementation high)
0562     {
0563         return static_cast<unsigned int>(low) <= current() &&
0564                static_cast<unsigned int>(high) >= current();
0565     }
0566     /**
0567      * Returns whether the current code would run on a CPU providing \p extraInstructions.
0568      */
0569     static constexpr bool runs_on(unsigned int extraInstructions)
0570     {
0571         return (extraInstructions & Features & ExtraInstructionsMask) ==
0572                (Features & ExtraInstructionsMask);
0573     }
0574 };
0575 /**
0576  * \ingroup Utilities
0577  * Identifies the %Vc implementation used in the current translation unit.
0578  *
0579  * \see ImplementationT
0580  */
0581 using CurrentImplementation = ImplementationT<
0582 #ifdef Vc_IMPL_Scalar
0583     ScalarImpl
0584 #elif defined(Vc_IMPL_AVX2)
0585     AVX2Impl
0586 #elif defined(Vc_IMPL_AVX)
0587     AVXImpl
0588 #elif defined(Vc_IMPL_SSE4_2)
0589     SSE42Impl
0590 #elif defined(Vc_IMPL_SSE4_1)
0591     SSE41Impl
0592 #elif defined(Vc_IMPL_SSSE3)
0593     SSSE3Impl
0594 #elif defined(Vc_IMPL_SSE3)
0595     SSE3Impl
0596 #elif defined(Vc_IMPL_SSE2)
0597     SSE2Impl
0598 #endif
0599 #ifdef Vc_IMPL_SSE4a
0600     + Vc::Sse4aInstructions
0601 #ifdef Vc_IMPL_XOP
0602     + Vc::XopInstructions
0603 #ifdef Vc_IMPL_FMA4
0604     + Vc::Fma4Instructions
0605 #endif
0606 #endif
0607 #endif
0608 #ifdef Vc_IMPL_POPCNT
0609     + Vc::PopcntInstructions
0610 #endif
0611 #ifdef Vc_IMPL_FMA
0612     + Vc::FmaInstructions
0613 #endif
0614 #ifdef Vc_IMPL_BMI2
0615     + Vc::Bmi2Instructions
0616 #endif
0617 #ifdef Vc_USE_VEX_CODING
0618     + Vc::VexInstructions
0619 #endif
0620     >;
0621 
0622 }  // namespace Vc
0623 
0624 #include "version.h"
0625 
0626 #endif // VC_GLOBAL_H_
0627 
0628 // vim: foldmethod=marker
[ Source navigation ] [ Diff markup ] [ Identifier search ] [ general search ]

This page was automatically generated by the 2.3.7 LXR engine.

The LXR team
Valid CSS 2.1! Valid HTML 4.01!