Compile ‘index_max’ fail.

[SkyeFromKindergarten]

Thanks for your working and new uploads ~

When compiling in file 'index_max_ext', I fail like below,

I cannot find why I fail. Do I have to use this new extension? I remember last time I run this code it does not need to be done.

Thanks and regards.

[lijx-nutonomy]

Thanks for point this out. I have updated the index_max_ext module, which is ok on my system with python3.7+cuda10.1+pytorch1.1. Could you please try again and tell me the result?

Previous the cuda operation is implemented with numba. In the latest commit I remove the dependency of numba and introduce the c++ based index_max_ext to do the job.

[SkyeFromKindergarten]

It seems to be the same ERROR with your latest code. My system is : python3.7 + cuda 10.0 + pytorch 1.2.

In the last line, it says command 'gcc' failed with exit status 1. I don't know if there's anything do with gcc's version. Version of my gcc is 7.3.0. I think it should be my environment problem which I couldn't find out TAT

[JeremyCJM]

I also failed to compile in folder 'index_max_ext'. My system is python3.7 + Cuda 9.1 + PyTorch 1.1.

[lijx-nutonomy]

That is weird since it is working fine on several computers in my lab. Anyway, I will try to solve it as soon as possible, but it may take some time. In the meantime, you may switch back to the numba implementation, e.g., the last commit.

[lijx-nutonomy]

@JeremyCJM @SkyeFromKindergarten The index_max_ext module has been updated for torch 1.2. Could you guys try again and give some feedbacks? Thanks!

[SkyeFromKindergarten]

I can successfully compile on my mechine now :)

[JeremyCJM]

Still failed to build...

[lijx10]

Please make sure that the cuda environment is configured properly. For example, proper environment variables. For example, you may consider adding these lines to ~/.bashrc

# cuda-10.1
export CUDA_HOME=/usr/local/cuda-10.1
export PATH=/usr/local/cuda-10.1/bin:${PATH}
export LD_LIBRARY_PATH=/usr/local/cuda-10.1/lib64:${LD_LIBRARY_PATH}

Run nvcc -V to confirm that cuda is working properly.

After making sure that cuda and pytorch is installed properly, delete folder dist, build, and the file index_max.egg-info, then run python setup install to install the index_max module again.

Finally, please make sure that the index_max folder is not in the python path. There may be linking problems.

[robin1chen2]

"Finally, please make sure that the index_max folder is not in the python path. There may be linking problems." You said make sure that the index_max_ext folder ''is not", instead of ''is'' ,in the python path .Did you make a mistake or I misunderstood?

[lijx10]

The index_max_ext folder should NOT in the python path, otherwise it may cause confusion on the one you installed into python, or that "un-installed" folder.