Open sirmarcel opened 3 months ago
Hello,
Many thanks for reporting this and providing your observations.
It's usually recommended to build software on the compute nodes, rather than the login nodes, at least this is what we recommend at CSCS. I will however investigate adding in a CMake variable where you can set the gencode and archcode manually, so you can choose that sphericart is built for a given architecture (or potentially a list of).
I'll create a PR tomorrow that addresses the above as well as the RAM issue (and modify the necessary documentation accordingly!)
Thank you! Very impressed with the responsiveness here :)
Hello! @nickjbrowning did this PR actually happen? Can we close this issue?
Hi there, I've just gone through the process of installing
sphericart
with CUDA for the first time and I'd suggest some minor improvements to the related docs:Additionally, the docs on building the C++ code could benefit from:
libtorch
is a requirement and where it should be exported (CMAKE_PREFIX_PATH
)cmake
to be at least version3.24
(otherwise the detection of which target GPU to target fails)Overall, it'd be a good idea to make it possible to build for non-present GPUs, but this can't be accomplished by slightly polishing the docs. :)