Skip to content

Update install.sh to fix Ubuntu compilation and dependencies #3

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
fdenzinger wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Update install.sh to fix Ubuntu compilation and dependencies #3

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
30 changes: 26 additions & 4 deletions install.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,5 +1,8 @@
#!/bin/bash

# Stop script immediately if any command fails
set -e

echo "---------------------------------------------"
echo "----------- Env Installation -----------"
echo "---------------------------------------------"
Expand Down Expand Up @@ -30,7 +33,7 @@ CUDA_VERSION=12.4

echo "Install PyTorch ${TORCH_VERSION} with CUDA ${CUDA_VERSION}"
conda install pytorch=="${TORCH_VERSION}" torchvision=="${TV_VERSION}" torchaudio=="${TA_VERSION}" \
pytorch-cuda="${CUDA_VERSION}" -c pytorch -c nvidia -y
pytorch-cuda="${CUDA_VERSION}" cuda-toolkit="${CUDA_VERSION}" cuda-nvcc="${CUDA_VERSION}" cuda-compiler="${CUDA_VERSION}" gcc_linux-64=13 gxx_linux-64=13 cuda-cudart-dev="${CUDA_VERSION}" cuda-cccl="${CUDA_VERSION}" -c pytorch -c nvidia -c conda-forge -y

python -c "import torch; print('torch:', torch.__version__, 'cuda:', torch.version.cuda, 'is_available:', torch.cuda.is_available())"

Expand All @@ -56,28 +59,47 @@ pip install -r requirements.txt
# 4) Build your cpp_core wrappers
# -------------------------
echo "Build cpp_core wrappers"

# Force C++ extensions to strictly use the Conda environment's CUDA toolkit
export CUDA_HOME=$CONDA_PREFIX

# Install missing build dependencies, forcing PCL 1.13+ and VTK
conda install cmake make swig "pcl>=1.13" vtk libboost-python -c conda-forge -y

pushd cpp_core/pcd_tiling
sed -i 's/boost_python311/boost_python38/g' CMakeLists.txt
bash generate_wraper.sh
popd

# install it when using supervoxel_seegmentation for partitioning
pushd cpp_core/supervoxel_segmentation
sed -i 's/boost_python311/boost_python38/g' CMakeLists.txt
bash generate_wraper.sh
popd

# install packages 5-7 if using superpoint transformer for partitioning, reference: superpoint_transformer/install.sh

# -------------------------
# 5) FRNN (as in your script)
# -------------------------
echo "⭐ Installing FRNN"
git clone git@github.com:zhaoyiww/superpoint_transformer.git
# Remove any existing FRNN artifacts if the script failed previously
rm -rf superpoint_transformer/src/dependencies/FRNN

git clone https://github.com/zhaoyiww/superpoint_transformer.git || true
cd superpoint_transformer
mkdir -p src/dependencies
if [ ! -d "src/dependencies/FRNN" ]; then
git clone --recursive https://github.com/lxxue/FRNN.git src/dependencies/FRNN
fi

# Force the compiler to target known, stable GPU architectures and allow GCC 13
export TORCH_CUDA_ARCH_LIST="6.0;6.1;7.0;7.5;8.0;8.6;8.9;9.0"
export TORCH_NVCC_FLAGS="-allow-unsupported-compiler"
export RELAXED_CUDA_VERSION_CHECK=1

# Unset the flags that broke standard GCC just in case they are lingering
unset CFLAGS
unset CXXFLAGS

pushd src/dependencies/FRNN/external/prefix_sum
python setup.py install
popd
Expand Down