main/api/math_8hpp_source.html

#ifndef TDLS_CORE_MATH_HPP

#define TDLS_CORE_MATH_HPP


#include <cmath>


#include <tdls/core/macros.hpp>

#include <tdls/core/target.hpp>


namespace tdls {


namespace detail {


template<typename T>


[[nodiscard]] TDLS_HOST_DEVICE TDLS_FORCE_INLINE_GPU constexpr T abs(const T x) noexcept {

#if defined(__has_builtin)

#if __has_builtin(__builtin_is_constant_evaluated)

    if (__builtin_is_constant_evaluated()) return x < T(0) ? -x : x;

#endif

#elif defined(TDLS_DETAIL_COMPILER_MSVC) && _MSC_VER >= 1925

    if (__builtin_is_constant_evaluated()) return x < T(0) ? -x : x;

#endif

    return std::fabs(x);

}


TDLS_HOST_DEVICE TDLS_FORCE_INLINE_GPU constexpr int runtime_dimension(int v) noexcept {

#if defined(TDLS_DETAIL_COMPILER_NVCC) && defined(TDLS_DETAIL_PASS_CUDA) && __CUDA_ARCH__ >= 1000

    if (!__builtin_is_constant_evaluated()) asm volatile("" : "+r"(v));

#endif

    return v;

}


} // namespace detail


} // namespace tdls


#endif // TDLS_CORE_MATH_HPP

macros.hpp
Toolchain detection and portability macros.

TDLS_FORCE_INLINE_GPU
#define TDLS_FORCE_INLINE_GPU
Inlining forced on GPU only: __forceinline__ under CUDA, the equivalent attribute under HIP (see TDLS...
Definition macros.hpp:77

TDLS_HOST_DEVICE
#define TDLS_HOST_DEVICE
__host__ __device__ under CUDA, the equivalent attributes under HIP, empty elsewhere.
Definition macros.hpp:50

tdls::detail::abs
TDLS_HOST_DEVICE TDLS_FORCE_INLINE_GPU constexpr T abs(const T x) noexcept
Absolute value usable in constant expressions.
Definition math.hpp:42

tdls::detail::runtime_dimension
TDLS_HOST_DEVICE TDLS_FORCE_INLINE_GPU constexpr int runtime_dimension(int v) noexcept
A dimension hidden from the optimizer of nvcc when it compiles device code for Blackwell (sm_100 and ...
Definition math.hpp:68

target.hpp
Detection of the compiler, of the language and of the side of the compilation pass: the one place whe...