build error:error: command '/usr/local/cuda/bin/nvcc' failed with exit status 1 #25
Comments
|
Hi, I'll look into this. Could you please post the PyTorch version that you used? |
|
@fmassa |
|
This is very weird. I've pulled the same version of the pytorch build as you, I have the same nvcc version but compilation went fine. I'm investigating a bit further, but I wonder if it might be related to facebookarchive/caffe2#1898 Can you try doing something like |
|
Oh, wait, what's your gcc version? It seems that CUDA 9.0 doesn't support gcc 6.4.0 |
|
Hi, my gcc version is 6.4.0! so, which gcc version I should use? |
|
I'm using gcc 5.4.0 and it works fine here. |
|
@fmassa I think CUDA9.0 just support gcc6 but not gcc7 which is default in Ubuntu18.04. I got this error to but no clue why |
|
Currently my gcc version is 6.4 |
|
@jinfagang according to the list I posted, CUDA 9.0 doesn't support gcc 6.4 either. |
|
@fmassa , |
|
@fmassa Using gcc5 by how? |
|
@gf19880710 if the environment flag doesn't work, you can add a line in or the path to your gcc5 |
|
@fmassa OK, feedback to you later. |
|
@fmassa I add that line to setup.py, but still got this: |
|
Just to double check, could you |
|
@fmassa rm build folder does'nt help, I tried just now. might be any other issue? I can see it still include some functions from /usr/include/c++/6 ... |
|
@jinfagang could you try ? |
|
@fmassa Then I did the operation according to your advice: rm -rd build/ Still failed but gcc failed this time. |
|
@gf19880710 from looking around in the internet, seems that your gcc 5 is broken https://stackoverflow.com/questions/41554900/gcc-and-g-error-error-trying-to-exec-cc1plus-execvp-no-such-file-or-direc What's your OS? There might be some OS-specific information online that we could use to try to make this work out. |
|
@fmassa after this I think I was successful! Thank you great man! |
|
Awesome! @gf19880710 could you summarize again what were the steps that you did so that it worked correctly? |
|
@fmassa
Wish these steps are useful ! |
|
I ran into the same issue. I switched from CUDA 9.0 to CUDA 9.2 and then it built fine. I tried this on Debian testing and Ubuntu 18.04. |
|
I am using CUDA 8.0 and ubuntu14.04......I have tried to change the version of gcc from 4.8 to 5.4, But the CUDA 8 can not support this gcc version... Maybe it can hardly been compiled in this environment? T T |
|
@gd2016229035 the solution in your case is to compile PyTorch from source. In this case, you can use CUDA 8 with gcc 4.8, and you can compile the Let me know if you have more questions. Instructions on how to compile PyTorch from source can be found in https://github.com/pytorch/pytorch. Don't forget to first uninstall PyTorch prior to compiling it. |
|
Compilation works fine by symlinking gcc and g++ to version 5 (on Ubuntu 18.10). Would you be willing to provide a conda-forge package for I can help if needed. |
|
Hi @fmassa ,I have encountered a problem similar to this, my problem can be seen here：open-mmlab/mmdetection#125 import os def readme(): MAJOR = 0 version_file = 'mmdet/version.py' def get_git_hash(): def get_hash(): def write_version_py(): TIME: {}version = '{}' def get_version(): if name == 'main': |
|
@Curry1201 are you having problems compiling |
|
Hi @gf19880710 , I have encountered a problem similar to yours. The main error description is {unable to execute ':/usr/local/cuda-9.0:/usr/local/cuda-9.0/bin/nvcc': No such file or directory. error: command ':/usr/local/cuda-9.0:/usr/local/cuda-9.0/bin/nvcc' failed with exit status 1}. The specific compilation process is as follows, can you give me some advice? ubuntu server 18.04 power@ubuntu:~/mmdetection$ PYTHON=python3 ./compile.sh |
|
@Curry1201 your question concern |
|
I had this issue; instead of manually changing the /usr/bin symlinks, I got it compile (after I had gcc-5) using update-alternatives: |
|
I install maskrcnn-benchmark 0.1 successfully |
|
@AllenAnthony you probably compiled |
|
@gf19880710 Thx so much for your resume. I had the same error when executing I finally solved the problem via switching to gcc/g++ 5. |
|
Somehow I encountered the similar problem. Probably I missed something when I installed cuda... |
that sounds easier than it is |
|
Even after following the steps in the most liked comment, I got the Check this answer in StackOverflow for more details. |
|
Just in setup.py, add this line "-ccbin=/usr/local/bin/gcc" in |
Hi, my cuda is 10.0.130 and my gcc version is 6.5.0, I think it's very suitable but I met the same problem... |
|
I am also encountered this when I install alphapose.Can you help me? Ubuntu 20.4 cuda 10.1 gcc g++ 7.5.0 |
cuda 11.2
and add below code maskrcnn_benchmark/csrc/cuda/deform_conv_cuda.cu |
I am so sad, it seems do not work in CUDA11 with g++ 5.4. |
❓ Questions and Help
Hello Great guys,
When I tried to install maksrcnn-benchmark in my PC, I got the follow compilation error in the last step, anyone can have a look? Thanks
The text was updated successfully, but these errors were encountered: