You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Describe the bug
I was looking around and playing with the pixi environment and I noticed that the COLMAP used and specified in the lockfile is the CPU version so this is no longer GPU accelerated compared to manually compiling it.
Also looking at the docker image the changes on pull rerquest #3283 made it install pycolmap from pip which according to their documentation does not support cuda as it has to be manually installed for that as well as having colmap installed with cuda support. So since then the docker based usages also don't have pycolmap/colmap with cuda support.
I am not sure this makes any difference at all and understand how painful it is to force the cuda versions of those but felt like sharing in case it was not known.
The text was updated successfully, but these errors were encountered:
As far as I can see from PyColmap/COLMAP they do not provide any pre compiled binary that was compiled with CUDA.
COLMAP documentation which says
Note that the COLMAP packages in the default repositories for Linux/Unix/BSD do not come with CUDA support, which requires a manual build from source https://colmap.github.io/install.html#installation
Describe the bug
I was looking around and playing with the pixi environment and I noticed that the COLMAP used and specified in the lockfile is the CPU version so this is no longer GPU accelerated compared to manually compiling it.
Also looking at the docker image the changes on pull rerquest #3283 made it install pycolmap from pip which according to their documentation does not support cuda as it has to be manually installed for that as well as having colmap installed with cuda support. So since then the docker based usages also don't have pycolmap/colmap with cuda support.
I am not sure this makes any difference at all and understand how painful it is to force the cuda versions of those but felt like sharing in case it was not known.
The text was updated successfully, but these errors were encountered: