Frederik Beaujean
Frederik Beaujean
Thanks for reporting. I will look into it. I would rather not require the latest version of cmake for everyone so perhaps we can add the appropriate `FindCUDA.cmake` file to...
If you use conda, a simpler way to get a current version of cmake is ``` conda install -c conda-forge cmake ```
If we require 3.9, cmake natively supports cuda on all platforms and 3.9 is already quite old. But would that also be enough for cuda 10? https://devblogs.nvidia.com/building-cuda-applications-cmake/
> Here two example recipes that use cuda 8: I need to check this out, we should always use the latest cuda version so people with the latest hardware benefit....
> I hope the conda forge system allows us to use the latest cuda automatically Of course nothing comes for free but it seems that conda is supported by nvidia...
I think it would be easiest to always build on ci using `cudatoolkit-dev` but not require that for installation. Instead, the user would have to have cuda installed as usual....
Relates to #158
Checking the docs mentioned above, it seems cuda is supported natively since cmake 3.10 (first occurence of the deprecation warning)
I also ran into problems due to the scarce documentation. I haven't checked if the first argument is right (probably isn't) because I didn't get to install yet. But `py.test.sh`...
Why don't the test builds run? I wanted to see this way if it still works with root5 but travis isn't called. Has something major changed since 2018 @oschulz ?