Code Ocean runs GPU-accelerated code. Certain Deep Learning frameworks (Caffe, TensorFlow, Theano, etc.) come pre-installed. Others can be installed on top of the base CUDA image using Code Ocean's package installation system. If you'd like to see built-in support for a new framework, please email us with your request. To run your code with GPU support:
- Create a new project. Select a language for which Code Ocean had provided a CUDA environment – C++, MATLAB, Python, or Lua.
- Click the Run Configuration button (right next to the Run button).
- Select one of the CUDA-enabled environments from the Base Environment dropdown.
- That's it! Your code will now automatically execute on one of our GPU machines. (Doing so will incur no special charges relative to a CPU-based run).