NVIDIA/nvcomp

[BUG] Can't be compiled on Windows

BarabashkaD opened this issue · 2 comments

  1. CMake file can't be configurate
    CMake Error at CMakeLists.txt:186 (install):
    install FILES given no DESTINATION!

  2. after fixing first issue by adding slash ${CMAKE_INSTALL_INCLUDEDIR}/
    still can't be compiled because of error on all .CU files
    1>C:\Program Files (x86)\Microsoft Visual Studio\2019\Community\VC\Tools\MSVC\14.29.30133\include\xutility(1309): error : expected a "("
    1> detected during instantiation of "void std::_Adl_verify_range(const _Iter &, const _Sentinel &) [with _Iter=const char *, _Sentinel=const char *]"
    1>C:\Program Files (x86)\Microsoft Visual Studio\2019\Community\VC\Tools\MSVC\14.29.30133\include\xlocale(1990): here

A lot of Windows build issues will be fixed in 2.2 release - please retry then and let us know if the issue still exists. We will be also shipping Windows binaries starting from 2.2.

Hi @BarabashkaD nvCOMP 2.2 should now be able to be built on Windows, and we updated the README.md with instructions for Windows in the Getting Started section. It's been built with Visual C++ 2019, so you'll need a compatible toolkit, (which it looks like you're already using, based on your comment above.)

For the CUDA version, you can plug in 10.2, 11.0, or 11.6 on the command line, though with some projects, I hit some issues with CUDA 11.0 that have apparently been addressed in CUDA 11.5 onward, but if you find that you can't build some project for 11.0 and you need to, try manually removing -std=c++14 from the CUDA C/C++ command line in the settings for the project that's failing to build.

If you run into any problems, or if it works for you now, please let us know! 🙂