"TORCH_ERROR("Not compiled with CUDA support")"

[Ashokvardhan]

Hi,

I have been trying to install this for the past couple of days but I am always running into errors. It would be of great help if you can provide some guidance on this. The following is the configuration of my virtual environment:

`python 3.7

pytorch 1.6

cudatoolkit 10.1

ninja 1.10.2 `

The following is the list of errors I get when I type python setup.py install

`running install

running bdist_egg

running egg_info

writing torch_butterfly.egg-info/PKG-INFO

writing dependency_links to torch_butterfly.egg-info/dependency_links.txt

writing top-level names to torch_butterfly.egg-info/top_level.txt

reading manifest file 'torch_butterfly.egg-info/SOURCES.txt'

writing manifest file 'torch_butterfly.egg-info/SOURCES.txt'

installing library code to build/bdist.linux-x86_64/egg

running install_lib

running build_py

creating build/lib.linux-x86_64-3.7

creating build/lib.linux-x86_64-3.7/torch_butterfly

copying torch_butterfly/permutation.py -> build/lib.linux-x86_64-3.7/torch_butterfly

copying torch_butterfly/diagonal.py -> build/lib.linux-x86_64-3.7/torch_butterfly

copying torch_butterfly/multiply.py -> build/lib.linux-x86_64-3.7/torch_butterfly

copying torch_butterfly/combine.py -> build/lib.linux-x86_64-3.7/torch_butterfly

copying torch_butterfly/special.py -> build/lib.linux-x86_64-3.7/torch_butterfly

copying torch_butterfly/complex_utils.py -> build/lib.linux-x86_64-3.7/torch_butterfly

copying torch_butterfly/init.py -> build/lib.linux-x86_64-3.7/torch_butterfly

copying torch_butterfly/butterfly.py -> build/lib.linux-x86_64-3.7/torch_butterfly

running build_ext

building 'torch_butterfly._butterfly' extension

creating build/temp.linux-x86_64-3.7

creating build/temp.linux-x86_64-3.7/home

creating build/temp.linux-x86_64-3.7/home/ashok

creating build/temp.linux-x86_64-3.7/home/ashok/reed_muller

creating build/temp.linux-x86_64-3.7/home/ashok/reed_muller/Neural_Plotkin

creating build/temp.linux-x86_64-3.7/home/ashok/reed_muller/Neural_Plotkin/reed_muller_modules

creating build/temp.linux-x86_64-3.7/home/ashok/reed_muller/Neural_Plotkin/reed_muller_modules/butterfly-master

creating build/temp.linux-x86_64-3.7/home/ashok/reed_muller/Neural_Plotkin/reed_muller_modules/butterfly-master/csrc

creating build/temp.linux-x86_64-3.7/home/ashok/reed_muller/Neural_Plotkin/reed_muller_modules/butterfly-master/csrc/cpu

gcc -pthread -B /home/ashok/anaconda3/envs/torch/compiler_compat -Wl,--sysroot=/ -Wsign-compare -DNDEBUG -g - fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -I/home/ashok/reed_muller/Neural_Plotkin/reed_muller_modules/butterfly-master/csrc -I/home/ashok/anaconda3/envs/torch/lib/python3.7/site-packages/torch/include -

I/home/ashok/anaconda3/envs/torch/lib/python3.7/site-packages/torch/include/torch/csrc/api/include -

I/home/ashok/anaconda3/envs/torch/lib/python3.7/site-packages/torch/include/TH -

I/home/ashok/anaconda3/envs/torch/lib/python3.7/site-packages/torch/include/THC -

I/home/ashok/anaconda3/envs/torch/include/python3.7m -c /home/ashok/reed_muller/Neural_Plotkin/reed_muller_modules/butterfly-master/csrc/butterfly.cpp -o build/temp.linux-x86_64-3.7/home/ashok/reed_muller/Neural_Plotkin/reed_muller_modules/butterfly-master/csrc/butterfly.o -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=_butterfly -D_GLIBCXX_USE_CXX11_ABI=1 -std=c++11

cc1plus: warning: command line option ‘-Wstrict-prototypes’ is valid for C/ObjC but not for C++

/home/ashok/reed_muller/Neural_Plotkin/reed_muller_modules/butterfly-master/csrc/butterfly.cpp: In function ‘at::Tensor butterfly_multiply_fw(at::Tensor, at::Tensor, bool)’: /home/ashok/reed_muller/Neural_Plotkin/reed_muller_modules/butterfly-master/csrc/butterfly.cpp:43:49: error: ‘TORCH_ERROR’ was not declared in this scope TORCH_ERROR("Not compiled with CUDA support"); ^ /home/ashok/reed_muller/Neural_Plotkin/reed_muller_modules/butterfly-master/csrc/butterfly.cpp: In function ‘std::tuple<at::Tensor, at::Tensor> butterfly_multiply_bw(at::Tensor, at::Tensor, at::Tensor, bool)’: /home/ashok/reed_muller/Neural_Plotkin/reed_muller_modules/butterfly-master/csrc/butterfly.cpp:81:49: error: ‘TORCH_ERROR’ was not declared in this scope TORCH_ERROR("Not compiled with CUDA support"); ^ /home/ashok/reed_muller/Neural_Plotkin/reed_muller_modules/butterfly-master/csrc/butterfly.cpp: At global scope: /home/ashok/reed_muller/Neural_Plotkin/reed_muller_modules/butterfly-master/csrc/butterfly.cpp:118:14: error: expected constructor, destructor, or type conversion before ‘(’ token TORCH_LIBRARY(torch_butterfly, m) { ^ /home/ashok/reed_muller/Neural_Plotkin/reed_muller_modules/butterfly-master/csrc/butterfly.cpp: In function ‘at::Tensor butterfly_multiply_fw(at::Tensor, at::Tensor, bool)’: /home/ashok/reed_muller/Neural_Plotkin/reed_muller_modules/butterfly-master/csrc/butterfly.cpp:48:1: warning: control reaches end of non-void function [-Wreturn-type] } ^ /home/ashok/reed_muller/Neural_Plotkin/reed_muller_modules/butterfly-master/csrc/butterfly.cpp: In function ‘std::tuple<at::Tensor, at::Tensor> butterfly_multiply_bw(at::Tensor, at::Tensor, at::Tensor, bool)’: /home/ashok/reed_muller/Neural_Plotkin/reed_muller_modules/butterfly-master/csrc/butterfly.cpp:86:1: warning: control reaches end of non-void function [-Wreturn-type] } ^ error: command 'gcc' failed with exit status 1 `

[tridao]

Are you compiling the code from the latest commit on master? Which OS and gcc version are you using? Was Pytorch installed with conda or pip?

[tridao]

One possibility is incorrect Pytorch version. Old Pytorch versions don't have TORCH_CHECK and that might cause the error. When I compile with Pytorch 1.6 it defaults to -std=c++14 (https://github.com/pytorch/pytorch/blob/v1.6.0/torch/utils/cpp_extension.py#L360) but in your case it's using -std=c++11. Can you check that you have Pytorch 1.6 in your eonvironment?

[tridao]

I've pushed a commit to change TORCH_ERROR to support non-CUDA compilation. Can you try that out?

[Ashokvardhan]

I've pushed a commit to change TORCH_ERROR to support non-CUDA compilation. Can you try that out?

Yes. I am able to successfully compile this. I see some warnings here and there but no errors.

[Ashokvardhan]

One possibility is incorrect Pytorch version. Old Pytorch versions don't have TORCH_CHECK and that might cause the error. When I compile with Pytorch 1.6 it defaults to -std=c++14 (https://github.com/pytorch/pytorch/blob/v1.6.0/torch/utils/cpp_extension.py#L360) but in your case it's using -std=c++11. Can you check that you have Pytorch 1.6 in your eonvironment?

Yes. I checked the version. When I typed import torch; print(torch.__version__) it outputs 1.6.0. Also I installed it with conda only.

I am using Linux OS only. More specifically, when I type uname -srm, it gives Linux 4.4.0-98-generic x86_64.

For the GCC, I checked that the version is gcc (Ubuntu 5.4.0-6ubuntu1~16.04.9) 5.4.0 20160609 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.

[tridao]

Yes. I am able to successfully compile this. I see some warnings here and there but no errors.

That's great. Thanks for trying that out.

[Ashokvardhan]

Yes. I am able to successfully compile this. I see some warnings here and there but no errors.

That's great. Thanks for trying that out.

Thanks a lot for your help! Just wondering if this is slower than the original version.

[tridao]

No, it was just an error on my part. Thanks for reporting it.