Quick Start

Download

Download the current files for Windows and Linux here.

Requirements

Windows

MacOS

Currently unsupported.

Using Hydra with MATLAB

  1. Download the MATLAB zip file from the download link above.
  2. Unzip the files.
  3. Place all folders with the + (plus) symbol in your MATLAB folder
    • Windows: C:\Users\username\Documents\MATLAB)
    • Linux: /home/username/Documents/MATLAB
  4. Open MATLAB
  5. Check if Hydra is loaded by running:
dev = HIP.Cuda.DeviceStatus()

Using Hydra with Python

  1. Download the source zip file from the download link above.
  2. Unzip the files.
  3. Navigate to the src/Python folder.
  4. Copy the HIP.lib or HIP.so to a location where you can import into python
  5. Check if Hydra is loaded by running:
import HIP
dev = HIP.Cuda.DeviceStats()
print(dev)

Setting up VSCode for Development

  1. Install python and conda

  2. Create virtual environment from environment.yml. conda env create -f src/python/environment.yml

  3. Activate environment conda activate hydra

  4. Configure the project with cmake

    • Windows - change C:\Users\EricWait\git\programming\hydra-image-processor with where you cloned the project to.
    C:\Program Files\CMake\bin\cmake.exe" --no-warn-unused-cli -DCMAKE_EXPORT_COMPILE_COMMANDS:BOOL=TRUE -Sc:/Users/EricWait/git/programming/hydra-image-processor -Bc:/Users/EricWait/git/programming/hydra-image-processor/build -G "Visual Studio 16 2019" -T host=x64 -A x64
    
    • Linux - change /home/ewait/git/programming/_core/hydra-image-processor with the path that you cloned the project to.
    /usr/bin/cmake --no-warn-unused-cli -DCMAKE_EXPORT_COMPILE_COMMANDS:BOOL=TRUE -DCMAKE_BUILD_TYPE:STRING=Release -DCMAKE_C_COMPILER:FILEPATH=/usr/bin/clang -DCMAKE_CXX_COMPILER:FILEPATH=/usr/bin/clang++ -S/home/ewait/git/programming/_core/hydra-image-processor -B/home/ewait/git/programming/_core/hydra-image-processor/build -G Ninja
    
  5. Use CMake Tools extension to build all projects. I recommended that you build in Release mode, so you don’t need the debug libraries for python.

  6. To test that the build worked, run python src/Python/test_nonchunk.py. You should get the following output (1, 1, 12, 128, 128) with no errors.

    1. In a windowing environment, you will also see

Happy coding!