56fd68e7dd
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
108 lines
3.8 KiB
C++
108 lines
3.8 KiB
C++
// -*- C++ -*-
|
|
//===----------------------------------------------------------------------===//
|
|
//
|
|
// Part of the LLVM Project, 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) 2024 NVIDIA CORPORATION & AFFILIATES.
|
|
//
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
#ifndef _LIBCUDACXX_SOURCE_LOCATION
|
|
#define _LIBCUDACXX_SOURCE_LOCATION
|
|
|
|
#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/cstdint>
|
|
#include <cuda/std/version>
|
|
|
|
#include <cuda/std/__cccl/prologue.h>
|
|
|
|
#if _CCCL_HAS_BUILTIN(__builtin_FILE) || _CCCL_COMPILER(GCC) || _CCCL_COMPILER(MSVC, >=, 19, 27)
|
|
# define _CCCL_BUILTIN_FILE() __builtin_FILE()
|
|
#else // ^^^ _CCCL_HAS_BUILTIN(__builtin_FILE) ^^^ / vvv !_CCCL_HAS_BUILTIN(__builtin_FILE) vvv
|
|
# define _CCCL_BUILTIN_FILE() __FILE__
|
|
#endif // !_CCCL_HAS_BUILTIN(__builtin_LINE)
|
|
|
|
#if _CCCL_HAS_BUILTIN(__builtin_FUNCTION) || _CCCL_COMPILER(GCC) || _CCCL_COMPILER(MSVC, >=, 19, 27)
|
|
# define _CCCL_BUILTIN_FUNCTION() __builtin_FUNCTION()
|
|
#else // ^^^ _CCCL_HAS_BUILTIN(__builtin_FUNCTION) ^^^ / vvv !_CCCL_HAS_BUILTIN(__builtin_FUNCTION) vvv
|
|
# define _CCCL_BUILTIN_FUNCTION() "__builtin_FUNCTION is unsupported"
|
|
#endif // !_CCCL_HAS_BUILTIN(__builtin_FUNCTION)
|
|
|
|
#if _CCCL_HAS_BUILTIN(__builtin_LINE) || _CCCL_COMPILER(GCC) || _CCCL_COMPILER(MSVC, >=, 19, 27)
|
|
# define _CCCL_BUILTIN_LINE() __builtin_LINE()
|
|
#else // ^^^ _CCCL_HAS_BUILTIN(__builtin_LINE) ^^^ / vvv !_CCCL_HAS_BUILTIN(__builtin_LINE) vvv
|
|
# define _CCCL_BUILTIN_LINE() __LINE__
|
|
#endif // !_CCCL_HAS_BUILTIN(__builtin_LINE)
|
|
|
|
#if _CCCL_HAS_BUILTIN(__builtin_COLUMN) || _CCCL_COMPILER(MSVC, >=, 19, 27)
|
|
# define _CCCL_BUILTIN_COLUMN() __builtin_COLUMN()
|
|
#else // ^^^ _CCCL_HAS_BUILTIN(__builtin_COLUMN) ^^^ / vvv !_CCCL_HAS_BUILTIN(__builtin_COLUMN) vvv
|
|
# define _CCCL_BUILTIN_COLUMN() 0
|
|
#endif // !_CCCL_HAS_BUILTIN(__builtin_COLUMN)
|
|
|
|
_CCCL_BEGIN_NAMESPACE_CUDA_STD
|
|
|
|
//! NOTE: we cannot use `__builtin_source_location` as that requires us to use `::std::source_location::impl`
|
|
class source_location
|
|
{
|
|
private:
|
|
const char* __file_name_ = "";
|
|
const char* __function_name_ = "";
|
|
uint_least32_t __line_ = 0;
|
|
uint_least32_t __column_ = 0;
|
|
|
|
// Needed to make C++11 happy
|
|
_CCCL_API explicit constexpr source_location(
|
|
const char* __file, const char* __func, uint_least32_t __line, uint_least32_t __column) noexcept
|
|
: __file_name_(__file)
|
|
, __function_name_(__func)
|
|
, __line_(__line)
|
|
, __column_(__column)
|
|
{}
|
|
|
|
public:
|
|
[[nodiscard]] _CCCL_API static constexpr source_location current(
|
|
const char* __file = _CCCL_BUILTIN_FILE(),
|
|
const char* __func = _CCCL_BUILTIN_FUNCTION(),
|
|
uint_least32_t __line = _CCCL_BUILTIN_LINE(),
|
|
uint_least32_t __column = _CCCL_BUILTIN_COLUMN()) noexcept
|
|
{
|
|
return source_location{__file, __func, __line, __column};
|
|
}
|
|
|
|
_CCCL_HIDE_FROM_ABI constexpr source_location() = default;
|
|
|
|
[[nodiscard]] _CCCL_API constexpr uint_least32_t line() const noexcept
|
|
{
|
|
return __line_;
|
|
}
|
|
[[nodiscard]] _CCCL_API constexpr uint_least32_t column() const noexcept
|
|
{
|
|
return __column_;
|
|
}
|
|
[[nodiscard]] _CCCL_API constexpr const char* file_name() const noexcept
|
|
{
|
|
return __file_name_;
|
|
}
|
|
[[nodiscard]] _CCCL_API constexpr const char* function_name() const noexcept
|
|
{
|
|
return __function_name_;
|
|
}
|
|
};
|
|
|
|
_CCCL_END_NAMESPACE_CUDA_STD
|
|
|
|
#include <cuda/std/__cccl/epilogue.h>
|
|
|
|
#endif // _LIBCUDACXX_SOURCE_LOCATION
|