Many frameworks for scientific computing, visual computing or deep learning are heavily optimized for a small set of programming languages and hardware platforms. Funzel tries to deliver a solution for all situations in which none of those solutions can provide proper support because no sufficient hardware support exists or because bindings for the programming language of choice are missing.
The main objective is about providing all essential building blocks for scientific use cases, statistics, computer vision and deep learning in a portable and easy to use way. A set of language bindings for languages like Lua, Python, C# and Java should be delivered to provide easy access for those who need to use one of those languages.
Hardware support is currently focused on CPU and OpenCL backends for portability.
| This software is still in early development, breakage and missing features are common! |
The project uses CMake as its build system and is written mostly in C++. Initial language bindings use the SWIG interface generator.
git clone https://github.com/Sponk/funzel.git
cd funzel
git submodule update --init --recursive
mkdir build && cd build
cmake .. -DCMAKE_BUILD_TYPE=Debug # Debug, Release etc.
cmake --build .
VCPKG supports the major three platforms, Windows, macOS and Linux, thus the following set of packages can be used everywhere.
vcpkg install openblas opencl lua python3 gtest benchmark spdlog fltk
