Taichi is an open-source computer graphics library that aims to provide easy-to-use infrastructures for computer graphics R&D. It's written in C++14 and wrapped friendly with Python.

This is a fork of Taichi legacy to use Taichi-MPM with some modifications.

To install Taichi and Taichi-MPM follow these instructions (Tested on Ubuntu 20.04 with an Intel Skylake).

1. Clone the repository

git clone https://github.com/dblanm/taichi.git

2. Go into the repository, clone the external dependencies and start the submodules

cd taichi
git clone https://github.com/yuanming-hu/taichi_runtime external/lib -b linux --depth 1
git submodule update --init --recursive

3. Install all the required libraries

sudo apt-get update
sudo apt-get install python3-dev git build-essential cmake make g++ libx11-dev
sudo apt-get install libtinfo-dev clang-7 libc++-dev
pip install --user psutil
pip install --user colorama numpy Pillow flask scipy pybind11 flask_cors GitPython yapf distro requests PyQt5

4. Create the build directory and the build directives

mkdir build && cd build
cmake .. -DCMAKE_CXX_COMPILER=/usr/bin/clang++-7

For new CPU architectures the architecture might not be found. In that case modify the compiler directive to your architecture (the available architectures can be found in the file taichi/cmake/OptimizeForArchitecture.cmake). You can replace skylake with your CPU architecture.

cmake .. -DCMAKE_CXX_COMPILER=/usr/bin/clang++-7 -DTARGET_ARCHITECTURE=skylake

To find your architecture in linux you can use the command:

gcc -march=native -Q --help=target|grep march

5. Now build the library

make -j5

6. Finally add the environment variables to your .bashrc file:

export TAICHI_NUM_THREADS=8
export TAICHI_REPO_DIR=/your/path/to/taichi
export PYTHONPATH=$TAICHI_REPO_DIR/python/:$PYTHONPATH
export PATH=$TAICHI_REPO_DIR/bin/:$PATH

where you should set TAICHI_REPO_DIR to your own directory.

And that's it, you are ready to go!