std::isnan 尚未声明,g++-5

std::isnan has not been declared, with g++-5

我正在尝试使用 nvcc 和 g++-5 编译我的程序,但出现此错误。

In file included from /usr/include/glm/detail/func_common.hpp:426:0,
             from /usr/include/glm/detail/func_geometric.inl:5,
             from /usr/include/glm/detail/func_geometric.hpp:113,
             from /usr/include/glm/geometric.hpp:6,
             from /usr/include/glm/detail/func_matrix.inl:4,
             from /usr/include/glm/detail/func_matrix.hpp:149,
             from /usr/include/glm/detail/type_mat2x2.inl:4,
             from /usr/include/glm/detail/type_mat2x2.hpp:182,
             from /usr/include/glm/mat2x2.hpp:6,
             from /usr/include/glm/glm.hpp:71,
             from src/controls.cpp:6:
/usr/include/glm/detail/func_common.inl:623:14: error: ‘std::isnan’   has not been declared
   using std::isnan;
          ^
/usr/include/glm/detail/func_common.inl:659:14: error: ‘std::isinf’ has not been declared
   using std::isinf;

使用 g++ 可以正确编译,但 nvcc 需要 g++-5。有谁知道如何解决这个问题?顺便说一下,一旦我 运行 sudo do-release-upgrade 在 Ubuntu.

上发生这种情况

nvcc --version 输出是这样的:

nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2016 NVIDIA Corporation
Built on Tue_Jan_10_13:22:03_CST_2017
Cuda compilation tools, release 8.0, V8.0.61

g++-5 --version 输出是这样的:

g++-5 (Ubuntu 5.5.0-1ubuntu1) 5.4.1 20171010
Copyright (C) 2015 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

g++ --version 输出是这样的:

g++ (Ubuntu 7.2.0-8ubuntu3) 7.2.0
Copyright (C) 2017 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

我正在使用 c++11 进行编译,但我的源文件中未包含 cmath。这是我的 CMakeLists.txt.

cmake_minimum_required(VERSION 3.7 FATAL_ERROR)
set(CMAKE_C_COMPILER /usr/bin/gcc-5 CACHE PATH "" FORCE)
set(CMAKE_CXX_COMPILER /usr/bin/gcc-5 CACHE PATH "" FORCE)
project(cpu_pcd C CXX CUDA)

add_compile_options(-std=c++11)

file(GLOB SRC src/*.cpp src/*.hpp)

set(EXECUTABLE_OUTPUT_PATH ${PROJECT_BINARY_DIR}/bin)

find_package(OpenGL REQUIRED)
find_package(glfw3 REQUIRED)
find_package(GLEW REQUIRED)
find_package(CUDA REQUIRED)

include_directories(src/)
include_directories(${OPENGL_INCLUDE_DIRS})
include_directories(${GLFW_INCLUDE_DIRS})
include_directories(${CUDA_INCLUDE_DIRS})

cuda_add_library(kernel src/update_particles.cu)
cuda_add_executable(cpu_pcd ${SRC})

set_target_properties(cpu_pcd
  PROPERTIES CUDA_SEPARABLE_COMPILATION on)

target_link_libraries(cpu_pcd kernel)
target_link_libraries(cpu_pcd ${OPENGL_LIBRARIES} glfw  ${GLEW_LIBRARIES} ${CUDA_LIBRARIES} pthread)

std::isnan 和 std::isinf 在 <cmath> 中从 C++11 开始声明。 header 包括在内吗?是否在 C++11 或更高版本的模式下完成编译?

您需要使 nvcc 到 运行 在 C++11 模式下托管 C++ 编译器(将此选项直接传递给 gcc 是不够的):

 nvcc -Xcompiler "-std=c++11"

详情可见here

相关信息:

c++ version supported by cuda 5.0

Error while using CUDA and C++11

对于那些感兴趣的人,我解决了它改变两行 CMakeLists.txt 像这样:

set(CMAKE_C_COMPILER /usr/bin/gcc-5 CACHE PATH "" FORCE)
set(CMAKE_CXX_COMPILER /usr/bin/g++ CACHE PATH "" FORCE)

现在可以正确编译了,感谢所有愿意提供帮助的人。