Currently, the C++ API forces C++14 flag during compilation. (-std=c++14)
Are there any plans to support C++17 in the near future?
Cuda 11.1 is already C++17 compatible.
Thx
Currently, the C++ API forces C++14 flag during compilation. (-std=c++14)
Are there any plans to support C++17 in the near future?
Cuda 11.1 is already C++17 compatible.
Thx
Hi All,
after changing “c++14” to “c++17” in all occurrences within /usr/lib/cmake/Caffe2/*.cmake, i can compile and run code successfully in C++17 mode.
It seems that the libtorch code base is already prepared for C++17, its just the compatibility w.r.t the CUDA env. which has to be maintained. In my case, i’m using CUDA 11.1, which is C++17 ready.
I guess it should not be too hard to enable C++17 when the installed CUDA version supports it… (e.g., by checking in cmake scripts…)
Cheers,
Oni