Closed peremato closed 1 year ago
There was an update in one of the tools used. I pushed a hotfix 3dac008 to use a previous version of that tool, which should allow you to build it.
The import error seems to be identical with #13. I can't say much about this as I'm not an expert on Apple products and I also don't have access to any Apple machines apart from the github CI runner (which seems to work fine).
Thanks. But I am not sure how to use the modified pyproject.toml
with pip3 install geant4-pybind
It is fixed. I just had to clone with recursive submodules. Thanks again.
In case you are interested I run some very simple benchmarks.
B2a (C++) | B2a.jl | B2a.py | |
---|---|---|---|
events = 1 | 0.9 s | 6 s | 1 s |
events =100k | 106 s | 79 s | 176 s |
events =100k (MT) | 23 s | 22 s | -- |
The reason why the python is slower is understandable because the user actions and sensitive detectors are implemented in Python and , it is many factors slower than C++ or Julia. If more work is done in these user actions, as it will be the case on more realistic applications, the performance will deteriorate even further. I do to have an explanation why Julia should be faster than C++ in the serial case. I was expecting within a few percent the same.
The OSX import error should be fixed too with b89bb3afd11672f903f0e59e3db9557bdd023116 by building for an older OSX version.
I am trying to run some simple G4 examples to compare with https://github.com/JuliaHEP/Geant4.jl but I am having problems installing the extension on MacOS.
pip3 install geant4-pybind
:Do you know how to fix these installation issues? Thanks in advance.