|
Gromacs
2026.0-dev-20251106-2ba968f
|
All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Macros Groups Pages
|
Gromacs
2026.0-dev-20251106-2ba968f
|
#include "config.h"#include <cstddef>#include <cstdint>#include <array>#include <memory>#include <type_traits>#include "gromacs/utility/basedefinitions.h"#include "gromacs/utility/real.h"#include "impl_reference/impl_reference.h"#include "gromacs/simd/hsimd_declarations.h"#include "gromacs/simd/scalar/scalar.h"#include "gromacs/simd/scalar/scalar_math.h"#include "gromacs/simd/scalar/scalar_util.h"#include "gromacs/simd/simd_memory.h"
Include dependency graph for simd.h: This graph shows which files directly or indirectly include this file:Definitions, capabilities, and wrappers for SIMD module.
The macros in this file are intended to be used for writing architecture-independent SIMD intrinsics code. To support a new architecture, adding a new sub-include with macros here should be (nearly) all that is needed.
The defines in this top-level file will set default Gromacs real precision operations to either single or double precision based on whether GMX_DOUBLE is 1. The actual implementation - including e.g. conversion operations specifically between single and double - is documented in impl_reference.h.
SIMD data types | |
The actual storage of these types is implementation dependent. The documentation is generated from the reference implementation, but for normal usage this will likely not be what you are using. | |
| typedef SimdFloat | gmx::SimdReal |
| Real precision floating-point SIMD datatype. More... | |
| typedef SimdFBool | gmx::SimdBool |
| Boolean SIMD type for usage with SimdReal. More... | |
| typedef SimdFInt32 | gmx::SimdInt32 |
| 32-bit integer SIMD type. More... | |
| typedef SimdFIBool | gmx::SimdIBool |
| Boolean SIMD type for usage with SimdInt32. More... | |
| typedef Simd4Float | gmx::Simd4Real |
| Real precision floating-point SIMD4 datatype. More... | |
| typedef Simd4FBool | gmx::Simd4Bool |
| Boolean SIMD4 type for usage with SimdReal. More... | |
| const int | gmx::c_simdBestPairAlignment = c_simdBestPairAlignmentFloat |
High-level SIMD proxy objects to disambiguate load/set operations | |
| #define | GMX_SIMD_HAVE_4NSIMD_UTIL_FLOAT GMX_SIMD_HAVE_LOADU |
| #define | GMX_SIMD4N_FLOAT_WIDTH GMX_SIMD_FLOAT_WIDTH |
| #define | GMX_SIMD_HAVE_4NSIMD_UTIL_DOUBLE GMX_SIMD_HAVE_LOADU |
| #define | GMX_SIMD4N_DOUBLE_WIDTH GMX_SIMD_DOUBLE_WIDTH |
| #define | GMX_SIMD_HAVE_4NSIMD_UTIL_REAL GMX_SIMD_HAVE_4NSIMD_UTIL_FLOAT |
| #define | GMX_SIMD4N_REAL_WIDTH GMX_SIMD4N_FLOAT_WIDTH |
| using | gmx::Simd4NFloat = SimdFloat |
| using | gmx::Simd4NDouble = SimdDouble |
| using | gmx::Simd4NReal = Simd4NFloat |
| template<typename T > | |
| static std::remove_const_t< T > | gmx::load (const internal::SimdTraitsT< T > *m) |
| Load function that returns SIMD or scalar. More... | |
| template<typename T > | |
| static T | gmx::load (const std::enable_if_t< std::is_arithmetic_v< T >, T > *m) |
| template<typename T , size_t N> | |
| static T gmx_simdcall | gmx::load (const AlignedArray< internal::SimdTraitsT< T >, N > &m) |
| template<typename T > | |
| static T | gmx::loadU (const internal::SimdTraitsT< T > *m) |
| Load function that returns SIMD or scalar based on template argument. More... | |
| template<typename T > | |
| static T | gmx::loadU (const std::enable_if_t< std::is_arithmetic_v< T >, T > *m) |
| template<typename T , size_t N> | |
| static T gmx_simdcall | gmx::loadU (const AlignedArray< internal::SimdTraitsT< T >, N > &m) |
| static SimdSetZeroProxy gmx_simdcall | gmx::setZero () |
| Helper function to set any SIMD or scalar variable to zero. More... | |
| template<typename T > | |
| T | gmx::load (const internal::Simd4TraitsT< T > *m) |
| template<typename T > | |
| T | gmx::loadU (const internal::Simd4TraitsT< T > *m) |
| static Simd4NFloat gmx_simdcall | gmx::loadUNDuplicate4 (const float *f) |
| static Simd4NFloat gmx_simdcall | gmx::load4DuplicateN (const float *f) |
| static Simd4NFloat gmx_simdcall | gmx::loadU4NOffset (const float *f, int) |
| static Simd4NDouble gmx_simdcall | gmx::loadUNDuplicate4 (const double *f) |
| static Simd4NDouble gmx_simdcall | gmx::load4DuplicateN (const double *f) |
| static Simd4NDouble gmx_simdcall | gmx::loadU4NOffset (const double *f, int) |
Classes | |
| struct | gmx::SimdFloatTag |
| Tag type to select to load SimdFloat with simdLoad(U) More... | |
| struct | gmx::SimdDoubleTag |
| Tag type to select to load SimdDouble with simdLoad(U) More... | |
| struct | gmx::SimdFInt32Tag |
| Tag type to select to load SimdFInt32 with simdLoad(U) More... | |
| struct | gmx::SimdDInt32Tag |
| Tag type to select to load SimdDInt32 with simdLoad(U) More... | |
| struct | gmx::AlignedArray< float, N > |
| Identical to std::array with GMX_SIMD_FLOAT_WIDTH alignment. Should not be deleted through base pointer (destructor is non-virtual). More... | |
| struct | gmx::AlignedArray< double, N > |
| Identical to std::array with GMX_SIMD_DOUBLE_WIDTH alignment. Should not be deleted through base pointer (destructor is non-virtual). More... | |
| struct | gmx::internal::SimdTraits< T > |
| Simd traits. More... | |
| class | gmx::SimdSetZeroProxy |
| Proxy object to enable setZero() for SIMD and real types. More... | |
Macros | |
SIMD predefined macros to describe high-level capabilities | |
These macros are used to describe the features available in default Gromacs real precision. They are set from the lower-level implementation files that have macros describing single and double precision individually, as well as the implementation details. | |
| #define | GMX_SIMD_HAVE_REAL GMX_SIMD_HAVE_FLOAT |
| 1 if SimdReal is available, otherwise 0. More... | |
| #define | GMX_SIMD_REAL_WIDTH GMX_SIMD_FLOAT_WIDTH |
| Width of SimdReal. More... | |
| #define | GMX_SIMD_HAVE_INT32_EXTRACT GMX_SIMD_HAVE_FINT32_EXTRACT |
| 1 if support is available for extracting elements from SimdInt32, otherwise 0 More... | |
| #define | GMX_SIMD_HAVE_INT32_LOGICAL GMX_SIMD_HAVE_FINT32_LOGICAL |
| 1 if logical ops are supported on SimdInt32, otherwise 0. More... | |
| #define | GMX_SIMD_HAVE_INT32_ARITHMETICS GMX_SIMD_HAVE_FINT32_ARITHMETICS |
| 1 if arithmetic ops are supported on SimdInt32, otherwise 0. More... | |
| #define | GMX_SIMD_HAVE_GATHER_LOADU_BYSIMDINT_TRANSPOSE_REAL GMX_SIMD_HAVE_GATHER_LOADU_BYSIMDINT_TRANSPOSE_FLOAT |
| 1 if gmx::simdGatherLoadUBySimdIntTranspose is present, otherwise 0 More... | |
| #define | GMX_SIMD_HAVE_HSIMD_UTIL_REAL GMX_SIMD_HAVE_HSIMD_UTIL_FLOAT |
| 1 if real half-register load/store/reduce utils present, otherwise 0 More... | |
| #define | GMX_SIMD4_HAVE_REAL GMX_SIMD4_HAVE_FLOAT |
| 1 if Simd4Real is available, otherwise 0. More... | |
Typedefs | |
| template<typename T > | |
| using | gmx::internal::SimdTraitsT = typename SimdTraits< T >::type |
| template<typename T > | |
| using | gmx::internal::Simd4TraitsT = typename Simd4Traits< T >::type |
1.8.5