Requirements
- python 2.5 or greater
- Numpy
- CUDA 4.0 toolkit
Installing arrayfire
linux, OSX
- Check setup
Open setup.py and make sure CUDIR (on line 12) points to your CUDA directory
- Install arrayfire
- Run the following command in terminal
sudo python setup.py install && sudo rm -rf build
Testing arrayfire
- Run the following command in terminal
python examples/info.py
- You should see information about the GPU available!
- You can also run other examples in examples folder
Windows
- Check setup
Find out where python is installed (usually C:\Python26)
- Install arrayfire
- Open cmd.exe
- Navigate to arrayfire/python directory and run the following
C:\Python26\python.exe setup.py install
Testing arrayfire
- Run the following command in cmd.exe
C:\Python26\python.exe examples\info.py
- You should see information about the GPU available!
Notes
- On Windows numpy is not offcially supported for the 64 bit architecture.
- You will need to install the 32 bit version of python, numpy and arrayfire.