ripser-plusplus
ripser-plusplus copied to clipboard
Error installing ripser-plusplus
Trying yo install ripster-plusplus on my system and got this error:
Building NVCC (Device) object CMakeFiles/pyripser++.dir/ripserplusplus/pyripser++_generated_ripser++.cu.o
error: "__or_v" is not a function or static data member
100 errors detected in the compilation of "/tmp/pip-req-build-csf0lpi2/./ripserplusplus/ripser++.cu".
Together with similar errors like this, any idea what this might be due to?
Thank you!
What are your versions for cuda and gcc? As is shown here:
CUDA version | max supported GCC version |
---|---|
11.1, 11.2, 11.3 | 10 |
11 | 9 |
10.1, 10.2 | 8 |
you need to have compatible GCC and CUDA versions. You can check your gcc and cuda by typing in:
gcc --version
and
nvcc --version
yes, I already confirmed this ... I have the right version . Ripser installs perfectly too. 10.4.0 for gcc and Cuda 11.1
Does it mean for 11.1 it has to be strictly 10 .
Thank you!
Yes, you should probably try lowering the gcc version. If that doesn't work, try lowering the version of both CUDA and gcc.
ok. I see. Thank you!
Have you been able to resolve this issue? I have the same problem when compiling, I have tried multiple cuda versions (11.1, 11.2, 11.3), and multiple GCC versions - 8.5, 9.5, 10.2, and also multiple CMake versions, but I have the same type of errors as highlighted above:
<gcc_conda_environment>/x86_64-conda-linux-gnu/include/c++/9.5.0/type_traits(148): error: "constexpr" is not valid here
<gcc_conda_environment>/x86_64-conda-linux-gnu/include/c++/9.5.0/type_traits(150): error: "__and_v" is not a function or static data member
<gcc_conda_environment>/x86_64-conda-linux-gnu/include/c++/9.5.0/type_traits(150): error: "constexpr" is not valid here
<gcc_conda_environment>/x86_64-conda-linux-gnu/include/c++/9.5.0/type_traits(170): error: "conjunction_v" is not a function or static data member
<gcc_conda_environment>/x86_64-conda-linux-gnu/include/c++/9.5.0/type_traits(170): error: "constexpr" is not valid here
Everything was installed in a fresh conda environment, with a dedicated gcc and gxx download. @simonzhang00 can you maybe provide a configuration that you know is working?
Okay, I was not able to resolve this within the supported CUDA versions, but the combination from google colab (GCC/GXX ver. 11.4.0, CUDA ver. 11.8.89) seems to work. Looks like the readme could use an update, but I'm not sure what are the exact compatibilities.