A library offering GPU parallel computation for array and matrix data types with CUDA.
- Linux/Unix
- g++ 4.2 or higher
- NVIDIA VGA card
- NVIDIA CUDA toolkit (>= CUDA 5.0)
- After installation, add the following 2 lines in ~/.bashrc
export LD_LIBRARY_PATH=/usr/local/cuda/lib
export PATH=$PATH:/usr/local/cuda/bin
- Use the header files in include/ folder directly.
- Refer to src/*.cu for use cases.
make
to create binaries in test/ for testing.