LR

-

Possibly related issue : https://github.com/isl-org/Open3D/issues/2892 However, I am not using any values except x, y, z

@benjaminum Any suggestions as you have worked on this in a past issue before?

To adapt to other sensors: in `laserscanvis.py` change lines 78, 79 to your sensor configuration. in `laserscan.py` change LaserScan `__init__(...)` defaults to your sensor configuration. in `laserscan.py` change LaserScan `open_scan(...)`...

Try `TORCH_CUDA_ARCH_LIST="8.6" MAX_JOBS=1 FORCE_CUDA=1 python3 setup.py install` Change `8.6` to your GPU architecture [according to](https://arnon.dk/matching-sm-architectures-arch-and-gencode-for-various-nvidia-cards/l). sm_86 is equal to 8.6

@francescoromito In all .cpp files within `lib\` comment out `#include `. Apparently, this is an old file thats not included with PyTorch anymore and its not needed.

@ksama777 Sometimes anaconda has issues with CUDA dependent packages. Maybe you can try installing outside of anaconda, then copying the installed files from your `site_packages` into the corresponding folder of...

@yeah334 Ive managed to get it to run on 20.04 LTS with CUDA 11.3. However I installed everything with pip making sure I check the original source for the correct...

@lucidrains I have the same issue, and cannot get a single layer to run with 12GB GPU Memory. Maybe my understanding of the layer is incorrect but I have made...

Regarding [issue 12](https://github.com/cardwing/Codes-for-PVKD/issues/12), the Cylinder3D repository is incomplete. This repro. is missing the same things in the code as the old repository: - Pointwise refinement module - Weighting for loss...