dylanyunlong/project_6
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity
Files
900ae0b1efc19e46cd313eb3ba7eb694fa596f6b
project_6/cccl_upstream/libcudacxx/include/cuda/std/array
EngineX CI 56fd68e7dd [INFRA] Import NVIDIA/CCCL upstream as optimization reference library 
CCCL (CUDA C++ Core Libraries) provides:
- CUB: device/block/warp-level GPU primitives (reduce, scan, sort, topk)
- Thrust: high-level parallel algorithms (transform_reduce, sort, scan)
- libcudacxx: CUDA C++ standard library (atomics, barriers, memory)
- cudax: experimental features (memory resources, allocators)
- Tuning policies: per-SM hardware-specific algorithm parameters

Competition optimization vectors mapped to CCCL:
- Output TPS (83% weight): warp_reduce, block_reduce, device_topk
- Input TPS (14% weight): device_scan, block_load, prefetch
- Cache TPS (3% weight): prefix caching strategy patterns
- Memory (0.9 util): pooled/cached/buddy allocators

Source: https://github.com/NVIDIA/cccl (shallow clone, HEAD only)
License: Apache-2.0
2026-07-30 09:35:51 +00:00

554 lines
17 KiB
Plaintext
Raw Blame History

//===----------------------------------------------------------------------===//
//
// Part of the CUDA Toolkit, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
// SPDX-FileCopyrightText: Copyright (c) 2023 NVIDIA CORPORATION & AFFILIATES.
//
//===----------------------------------------------------------------------===//
#ifndef _CUDA_STD_ARRAY
#define _CUDA_STD_ARRAY
#include <cuda/std/detail/__config>
#if defined(_CCCL_IMPLICIT_SYSTEM_HEADER_GCC)
# pragma GCC system_header
#elif defined(_CCCL_IMPLICIT_SYSTEM_HEADER_CLANG)
# pragma clang system_header
#elif defined(_CCCL_IMPLICIT_SYSTEM_HEADER_MSVC)
# pragma system_header
#endif // no system header
#include <cuda/std/__algorithm/equal.h>
#include <cuda/std/__algorithm/fill_n.h>
#include <cuda/std/__algorithm/lexicographical_compare.h>
#include <cuda/std/__algorithm/swap_ranges.h>
#include <cuda/std/__concepts/concept_macros.h>
#include <cuda/std/__exception/exception_macros.h>
#include <cuda/std/__fwd/array.h>
#include <cuda/std/__host_stdlib/stdexcept>
#include <cuda/std/__iterator/reverse_iterator.h>
#include <cuda/std/__tuple_dir/sfinae_helpers.h>
#include <cuda/std/__type_traits/conditional.h>
#include <cuda/std/__type_traits/is_array.h>
#include <cuda/std/__type_traits/is_const.h>
#include <cuda/std/__type_traits/is_constructible.h>
#include <cuda/std/__type_traits/is_move_constructible.h>
#include <cuda/std/__type_traits/is_nothrow_constructible.h>
#include <cuda/std/__type_traits/is_nothrow_move_constructible.h>
#include <cuda/std/__type_traits/is_same.h>
#include <cuda/std/__type_traits/is_swappable.h>
#include <cuda/std/__type_traits/remove_cv.h>
#include <cuda/std/__utility/integer_sequence.h>
#include <cuda/std/__utility/move.h>
#include <cuda/std/__utility/unreachable.h>
#include <cuda/std/cstdint>
#include <cuda/std/limits>
// standard-mandated includes
#include <cuda/std/version>
// [iterator.range]
#include <cuda/std/__iterator/access.h>
#include <cuda/std/__iterator/data.h>
#include <cuda/std/__iterator/empty.h>
#include <cuda/std/__iterator/reverse_access.h>
#include <cuda/std/__iterator/size.h>
// [array.syn]
#if _LIBCUDACXX_HAS_SPACESHIP_OPERATOR()
# include <compare>
#endif // _LIBCUDACXX_HAS_SPACESHIP_OPERATOR()
#include <cuda/std/initializer_list>
// [tuple.helper]
#include <cuda/std/__tuple_dir/tuple_element.h>
#include <cuda/std/__tuple_dir/tuple_size.h>
#include <cuda/std/__cccl/prologue.h>
_CCCL_BEGIN_NAMESPACE_CUDA_STD
template <class _Tp, size_t _Size>
struct _CCCL_TYPE_VISIBILITY_DEFAULT array
{
// types:
using __self = array;
using value_type = _Tp;
using reference = value_type&;
using const_reference = const value_type&;
using iterator = value_type*;
using const_iterator = const value_type*;
using pointer = value_type*;
using const_pointer = const value_type*;
using size_type = size_t;
using difference_type = ptrdiff_t;
using reverse_iterator = ::cuda::std::reverse_iterator<iterator>;
using const_reverse_iterator = ::cuda::std::reverse_iterator<const_iterator>;
_Tp __elems_[_Size];
// No explicit construct/copy/destroy for aggregate type
_CCCL_API constexpr void fill(const value_type& __u)
{
::cuda::std::fill_n(__elems_, _Size, __u);
}
_CCCL_API constexpr void swap(array& __a) noexcept(is_nothrow_swappable_v<_Tp>)
{
::cuda::std::swap_ranges(__elems_, __elems_ + _Size, __a.data());
}
// iterators:
[[nodiscard]] _CCCL_API constexpr iterator begin() noexcept
{
return iterator(__elems_);
}
[[nodiscard]] _CCCL_API constexpr const_iterator begin() const noexcept
{
return const_iterator(__elems_);
}
[[nodiscard]] _CCCL_API constexpr iterator end() noexcept
{
return iterator(__elems_ + _Size);
}
[[nodiscard]] _CCCL_API constexpr const_iterator end() const noexcept
{
return const_iterator(__elems_ + _Size);
}
[[nodiscard]] _CCCL_API constexpr reverse_iterator rbegin() noexcept
{
return reverse_iterator(end());
}
[[nodiscard]] _CCCL_API constexpr const_reverse_iterator rbegin() const noexcept
{
return const_reverse_iterator(end());
}
[[nodiscard]] _CCCL_API constexpr reverse_iterator rend() noexcept
{
return reverse_iterator(begin());
}
[[nodiscard]] _CCCL_API constexpr const_reverse_iterator rend() const noexcept
{
return const_reverse_iterator(begin());
}
[[nodiscard]] _CCCL_API constexpr const_iterator cbegin() const noexcept
{
return begin();
}
[[nodiscard]] _CCCL_API constexpr const_iterator cend() const noexcept
{
return end();
}
[[nodiscard]] _CCCL_API constexpr const_reverse_iterator crbegin() const noexcept
{
return rbegin();
}
[[nodiscard]] _CCCL_API constexpr const_reverse_iterator crend() const noexcept
{
return rend();
}
// capacity:
[[nodiscard]] _CCCL_API constexpr size_type size() const noexcept
{
return _Size;
}
[[nodiscard]] _CCCL_API constexpr size_type max_size() const noexcept
{
return _Size;
}
[[nodiscard]] _CCCL_API constexpr bool empty() const noexcept
{
return _Size == 0;
}
// element access:
[[nodiscard]] _CCCL_API constexpr reference operator[](size_type __n) noexcept
{
_CCCL_ASSERT(__n < _Size, "out-of-bounds access in std::array<T, N>");
return __elems_[__n];
}
[[nodiscard]] _CCCL_API constexpr const_reference operator[](size_type __n) const noexcept
{
_CCCL_ASSERT(__n < _Size, "out-of-bounds access in std::array<T, N>");
return __elems_[__n];
}
[[nodiscard]] _CCCL_API constexpr reference at(size_type __n)
{
if (__n >= _Size)
{
_CCCL_THROW(::std::out_of_range, "array::at");
}
return __elems_[__n];
}
[[nodiscard]] _CCCL_API constexpr const_reference at(size_type __n) const
{
if (__n >= _Size)
{
_CCCL_THROW(::std::out_of_range, "array::at");
}
return __elems_[__n];
}
[[nodiscard]] _CCCL_API constexpr reference front() noexcept
{
return (*this)[0];
}
[[nodiscard]] _CCCL_API constexpr const_reference front() const noexcept
{
return (*this)[0];
}
[[nodiscard]] _CCCL_API constexpr reference back() noexcept
{
return (*this)[_Size - 1];
}
[[nodiscard]] _CCCL_API constexpr const_reference back() const noexcept
{
return (*this)[_Size - 1];
}
[[nodiscard]] _CCCL_API constexpr value_type* data() noexcept
{
return __elems_;
}
[[nodiscard]] _CCCL_API constexpr const value_type* data() const noexcept
{
return __elems_;
}
};
_CCCL_DIAG_PUSH
_CCCL_DIAG_SUPPRESS_MSVC(4702) // Unreachable code
template <class _Tp>
struct _CCCL_TYPE_VISIBILITY_DEFAULT array<_Tp, 0>
{
// types:
using __self = array;
using value_type = _Tp;
using reference = value_type&;
using const_reference = const value_type&;
using iterator = value_type*;
using const_iterator = const value_type*;
using pointer = value_type*;
using const_pointer = const value_type*;
using size_type = size_t;
using difference_type = ptrdiff_t;
using reverse_iterator = ::cuda::std::reverse_iterator<iterator>;
using const_reverse_iterator = ::cuda::std::reverse_iterator<const_iterator>;
using _CharType = conditional_t<is_const_v<_Tp>, const char, char>;
struct _ArrayInStructT
{
_Tp __data_[1];
};
_CCCL_ALIGNAS_TYPE(_ArrayInStructT) _CharType __elems_[sizeof(_ArrayInStructT)];
[[nodiscard]] _CCCL_API constexpr value_type* data() noexcept
{
return nullptr;
}
[[nodiscard]] _CCCL_API constexpr const value_type* data() const noexcept
{
return nullptr;
}
// No explicit construct/copy/destroy for aggregate type
_CCCL_API constexpr void fill(const value_type&)
{
static_assert(!is_const_v<_Tp>, "cannot fill zero-sized array of type 'const T'");
}
_CCCL_API constexpr void swap(array&) noexcept
{
static_assert(!is_const_v<_Tp>, "cannot swap zero-sized array of type 'const T'");
}
// iterators:
[[nodiscard]] _CCCL_API constexpr iterator begin() noexcept
{
return iterator(nullptr);
}
[[nodiscard]] _CCCL_API constexpr const_iterator begin() const noexcept
{
return const_iterator(nullptr);
}
[[nodiscard]] _CCCL_API constexpr iterator end() noexcept
{
return iterator(nullptr);
}
[[nodiscard]] _CCCL_API constexpr const_iterator end() const noexcept
{
return const_iterator(nullptr);
}
[[nodiscard]] _CCCL_API constexpr reverse_iterator rbegin() noexcept
{
return reverse_iterator(end());
}
[[nodiscard]] _CCCL_API constexpr const_reverse_iterator rbegin() const noexcept
{
return const_reverse_iterator(end());
}
[[nodiscard]] _CCCL_API constexpr reverse_iterator rend() noexcept
{
return reverse_iterator(begin());
}
[[nodiscard]] _CCCL_API constexpr const_reverse_iterator rend() const noexcept
{
return const_reverse_iterator(begin());
}
[[nodiscard]] _CCCL_API constexpr const_iterator cbegin() const noexcept
{
return begin();
}
[[nodiscard]] _CCCL_API constexpr const_iterator cend() const noexcept
{
return end();
}
[[nodiscard]] _CCCL_API constexpr const_reverse_iterator crbegin() const noexcept
{
return rbegin();
}
[[nodiscard]] _CCCL_API constexpr const_reverse_iterator crend() const noexcept
{
return rend();
}
// capacity:
[[nodiscard]] _CCCL_API constexpr size_type size() const noexcept
{
return 0;
}
[[nodiscard]] _CCCL_API constexpr size_type max_size() const noexcept
{
return 0;
}
[[nodiscard]] [[nodiscard]] _CCCL_API constexpr bool empty() const noexcept
{
return true;
}
// element access:
[[nodiscard]] _CCCL_API constexpr reference operator[](size_type) noexcept
{
_CCCL_ASSERT(false, "cannot call array<T, 0>::operator[] on a zero-sized array");
_CCCL_UNREACHABLE();
return *data();
}
[[nodiscard]] _CCCL_API constexpr const_reference operator[](size_type) const noexcept
{
_CCCL_ASSERT(false, "cannot call array<T, 0>::operator[] on a zero-sized array");
_CCCL_UNREACHABLE();
return *data();
}
[[nodiscard]] _CCCL_API constexpr reference at(size_type)
{
_CCCL_THROW(::std::out_of_range, "array<T, 0>::at");
_CCCL_UNREACHABLE();
}
[[nodiscard]] _CCCL_API constexpr const_reference at(size_type) const
{
_CCCL_THROW(::std::out_of_range, "array<T, 0>::at");
_CCCL_UNREACHABLE();
}
[[nodiscard]] _CCCL_API constexpr reference front() noexcept
{
_CCCL_ASSERT(false, "cannot call array<T, 0>::front() on a zero-sized array");
_CCCL_UNREACHABLE();
return *data();
}
[[nodiscard]] _CCCL_API constexpr const_reference front() const noexcept
{
_CCCL_ASSERT(false, "cannot call array<T, 0>::front() on a zero-sized array");
_CCCL_UNREACHABLE();
return *data();
}
[[nodiscard]] _CCCL_API constexpr reference back() noexcept
{
_CCCL_ASSERT(false, "cannot call array<T, 0>::back() on a zero-sized array");
_CCCL_UNREACHABLE();
return *data();
}
[[nodiscard]] _CCCL_API constexpr const_reference back() const noexcept
{
_CCCL_ASSERT(false, "cannot call array<T, 0>::back() on a zero-sized array");
_CCCL_UNREACHABLE();
return *data();
}
};
_CCCL_DIAG_POP
_CCCL_TEMPLATE(class _Tp, class... _Args)
_CCCL_REQUIRES((is_same_v<_Tp, _Args> && ...))
_CCCL_DEDUCTION_GUIDE_ATTRIBUTES array(_Tp, _Args...) -> array<_Tp, 1 + sizeof...(_Args)>;
template <class _Tp, size_t _Size>
[[nodiscard]] _CCCL_API constexpr bool operator==(const array<_Tp, _Size>& __x, const array<_Tp, _Size>& __y)
{
return ::cuda::std::equal(__x.begin(), __x.end(), __y.begin());
}
template <class _Tp, size_t _Size>
[[nodiscard]] _CCCL_API constexpr bool operator!=(const array<_Tp, _Size>& __x, const array<_Tp, _Size>& __y)
{
return !(__x == __y);
}
template <class _Tp, size_t _Size>
[[nodiscard]] _CCCL_API constexpr bool operator<(const array<_Tp, _Size>& __x, const array<_Tp, _Size>& __y)
{
return ::cuda::std::lexicographical_compare(__x.begin(), __x.end(), __y.begin(), __y.end());
}
template <class _Tp, size_t _Size>
[[nodiscard]] _CCCL_API constexpr bool operator>(const array<_Tp, _Size>& __x, const array<_Tp, _Size>& __y)
{
return __y < __x;
}
template <class _Tp, size_t _Size>
[[nodiscard]] _CCCL_API constexpr bool operator<=(const array<_Tp, _Size>& __x, const array<_Tp, _Size>& __y)
{
return !(__y < __x);
}
template <class _Tp, size_t _Size>
[[nodiscard]] _CCCL_API constexpr bool operator>=(const array<_Tp, _Size>& __x, const array<_Tp, _Size>& __y)
{
return !(__x < __y);
}
_CCCL_TEMPLATE(class _Tp, size_t _Size)
_CCCL_REQUIRES((_Size == 0) || is_nothrow_swappable_v<_Tp>)
_CCCL_API constexpr void swap(array<_Tp, _Size>& __x, array<_Tp, _Size>& __y) noexcept(noexcept(__x.swap(__y)))
{
__x.swap(__y);
}
template <size_t _Ip, class _Tp, size_t _Size>
[[nodiscard]] _CCCL_API constexpr _Tp& get(array<_Tp, _Size>& __a) noexcept
{
static_assert(_Ip < _Size, "Index out of bounds in std::get<> (std::array)");
return __a.__elems_[_Ip];
}
template <size_t _Ip, class _Tp, size_t _Size>
[[nodiscard]] _CCCL_API constexpr const _Tp& get(const array<_Tp, _Size>& __a) noexcept
{
static_assert(_Ip < _Size, "Index out of bounds in std::get<> (const std::array)");
return __a.__elems_[_Ip];
}
template <size_t _Ip, class _Tp, size_t _Size>
[[nodiscard]] _CCCL_API constexpr _Tp&& get(array<_Tp, _Size>&& __a) noexcept
{
static_assert(_Ip < _Size, "Index out of bounds in std::get<> (std::array &&)");
return ::cuda::std::move(__a.__elems_[_Ip]);
}
template <size_t _Ip, class _Tp, size_t _Size>
[[nodiscard]] _CCCL_API constexpr const _Tp&& get(const array<_Tp, _Size>&& __a) noexcept
{
static_assert(_Ip < _Size, "Index out of bounds in std::get<> (const std::array &&)");
return ::cuda::std::move(__a.__elems_[_Ip]);
}
template <typename _Tp, size_t _Size, size_t... _Index>
[[nodiscard]] _CCCL_API constexpr array<remove_cv_t<_Tp>, _Size>
__to_array_lvalue_impl(_Tp (&__arr)[_Size], index_sequence<_Index...>)
{
return {{__arr[_Index]...}};
}
template <typename _Tp, size_t _Size, size_t... _Index>
[[nodiscard]] _CCCL_API constexpr array<remove_cv_t<_Tp>, _Size>
__to_array_rvalue_impl(_Tp (&&__arr)[_Size], index_sequence<_Index...>)
{
return {{::cuda::std::move(__arr[_Index])...}};
}
template <typename _Tp, size_t _Size>
[[nodiscard]] _CCCL_API constexpr array<remove_cv_t<_Tp>, _Size>
to_array(_Tp (&__arr)[_Size]) noexcept(is_nothrow_constructible_v<_Tp, _Tp&>)
{
static_assert(!is_array_v<_Tp>, "[array.creation]/1: to_array does not accept multidimensional arrays.");
static_assert(is_constructible_v<_Tp, _Tp&>, "[array.creation]/1: to_array requires copy constructible elements.");
return ::cuda::std::__to_array_lvalue_impl(__arr, make_index_sequence<_Size>());
}
template <typename _Tp, size_t _Size>
[[nodiscard]] _CCCL_API constexpr array<remove_cv_t<_Tp>, _Size>
to_array(_Tp (&&__arr)[_Size]) noexcept(is_nothrow_move_constructible_v<_Tp>)
{
static_assert(!is_array_v<_Tp>, "[array.creation]/4: to_array does not accept multidimensional arrays.");
static_assert(is_move_constructible_v<_Tp>, "[array.creation]/4: to_array requires move constructible elements.");
return ::cuda::std::__to_array_rvalue_impl(::cuda::std::move(__arr), make_index_sequence<_Size>());
}
// specialize cuda::std::tuple_size and cuda::std::tuple_element for both std::array and cuda::std::array
#if _CCCL_HAS_HOST_STD_LIB()
template <class _Tp, size_t _Np>
struct _CCCL_TYPE_VISIBILITY_DEFAULT tuple_size<::std::array<_Tp, _Np>> : integral_constant<size_t, _Np>
{};
template <size_t _Ip, class _Tp, size_t _Np>
struct _CCCL_TYPE_VISIBILITY_DEFAULT tuple_element<_Ip, ::std::array<_Tp, _Np>>
{
static_assert(_Ip < _Np, "Index out of bounds in cuda::std::tuple_element<> (std::array)");
using type _CCCL_NODEBUG_ALIAS = _Tp;
};
#endif // _CCCL_HAS_HOST_STD_LIB()
template <class _Tp, size_t _Np>
struct _CCCL_TYPE_VISIBILITY_DEFAULT tuple_size<array<_Tp, _Np>> : integral_constant<size_t, _Np>
{};
template <size_t _Ip, class _Tp, size_t _Np>
struct _CCCL_TYPE_VISIBILITY_DEFAULT tuple_element<_Ip, array<_Tp, _Np>>
{
static_assert(_Ip < _Np, "Index out of bounds in cuda::std::tuple_element<> (cuda::std::array)");
using type _CCCL_NODEBUG_ALIAS = _Tp;
};
_CCCL_END_NAMESPACE_CUDA_STD
// tuple protocol for cuda::std::array
_CCCL_BEGIN_NAMESPACE_STD
template <class _Tp, ::cuda::std::size_t _Np>
struct _CCCL_TYPE_VISIBILITY_DEFAULT tuple_size<::cuda::std::array<_Tp, _Np>>
: ::cuda::std::integral_constant<::cuda::std::size_t, _Np>
{};
template <::cuda::std::size_t _Ip, class _Tp, ::cuda::std::size_t _Np>
struct _CCCL_TYPE_VISIBILITY_DEFAULT tuple_element<_Ip, ::cuda::std::array<_Tp, _Np>>
{
static_assert(_Ip < _Np, "Index out of bounds in std::tuple_element<> (cuda::std::array)");
using type _CCCL_NODEBUG_ALIAS = _Tp;
};
_CCCL_END_NAMESPACE_STD
#include <cuda/std/__cccl/epilogue.h>
#endif // _CUDA_STD_ARRAY
Reference in New Issue View Git Blame Copy Permalink