Hello, I’m trying to write the hello world program and following all the instructions in “Installing C++ Distributions of PyTorch — PyTorch master documentation”. I can manege to build all the files with cmake with following prints:
-- The C compiler identification is AppleClang 8.0.0.8000042
-- The CXX compiler identification is AppleClang 8.0.0.8000042
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working C compiler: /Library/Developer/CommandLineTools/usr/bin/cc - skipped
-- Detecting C compile features
-- Detecting C compile features - done
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Check for working CXX compiler: /Library/Developer/CommandLineTools/usr/bin/c++ - skipped
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- Looking for pthread.h
-- Looking for pthread.h - found
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD - Success
-- Found Threads: TRUE
-- MKL_ARCH: None, set to ` intel64` by default
-- MKL_ROOT /Users/kayttaja-sepe/anaconda
-- MKL_LINK: None, set to ` dynamic` by default
-- MKL_INTERFACE_FULL: None, set to ` intel_ilp64` by default
-- MKL_THREADING: None, set to ` intel_thread` by default
-- MKL_MPI: None, set to ` mpich` by default
-- Found Torch: /MAIN/KOODI/C/lib/libtorch/lib/libtorch.dylib
-- Configuring done
-- Generating done
-- Build files have been written to: /MAIN/KOODI/C/torch_test/out
Do MKL_ARCH / MKL_ROOT etc. have something to do with GPU or what are they? I would appreciate any info about these variables.
Next when I compile, I get the following error:
In file included from /MAIN/KOODI/C/torch_test/main.cpp:1:
In file included from /MAIN/KOODI/C/lib/libtorch/include/torch/csrc/api/include/torch/torch.h:3:
In file included from /MAIN/KOODI/C/lib/libtorch/include/torch/csrc/api/include/torch/all.h:8:
In file included from /MAIN/KOODI/C/lib/libtorch/include/torch/csrc/api/include/torch/data.h:3:
In file included from /MAIN/KOODI/C/lib/libtorch/include/torch/csrc/api/include/torch/data/dataloader.h:3:
In file included from /MAIN/KOODI/C/lib/libtorch/include/torch/csrc/api/include/torch/data/dataloader/stateful.h:3:
In file included from /MAIN/KOODI/C/lib/libtorch/include/torch/csrc/api/include/torch/data/dataloader/base.h:3:
In file included from /MAIN/KOODI/C/lib/libtorch/include/torch/csrc/api/include/torch/data/dataloader_options.h:4:
In file included from /MAIN/KOODI/C/lib/libtorch/include/torch/csrc/api/include/torch/types.h:3:
In file included from /MAIN/KOODI/C/lib/libtorch/include/ATen/ATen.h:11:
In file included from /MAIN/KOODI/C/lib/libtorch/include/ATen/DeviceGuard.h:4:
In file included from /MAIN/KOODI/C/lib/libtorch/include/ATen/core/Tensor.h:3:
In file included from /MAIN/KOODI/C/lib/libtorch/include/ATen/core/TensorBody.h:16:
/MAIN/KOODI/C/lib/libtorch/include/c10/core/Scalar.h:208:37: error: inline declaration of 'to<unsigned char>' follows non-inline definition
AT_FORALL_SCALAR_TYPES_WITH_COMPLEX(DEFINE_TO)
And the basically same error following repeatedly. The path to libtorch is 100% correct and the library is up-to-date and my compiler supports c++14. I’m running Mac OS X 10.12.6 though and don’t want to update for several reasons.
I would appreciate all the help to debug and understand this issue. I personally have no clue what to do next. I have been using PyTorch on python for years but wanted to experiment with c++ version.
EDIT: And I forgot to mention that I’m using CPU