isabella232 / dask_gdf

Dask support for distributed GDF object

DASK_GDF

A partitioned gpu-backed dataframe, using dask.

Setup

Setup from source repo:

1. Follow instructions in https://github.com/gpuopenanalytics/pygdf#setup to setup a conda environment for pygdf.
2. Activate pygdf environment: source activate pygdf_dev
3. Clone dask_gdf repo: git clone https://github.com/gpuopenanalytics/dask_gdf path/to/dask_gdf
4. cd path/to/dask_gdf
5. Install additional dependency: conda install dask distributed

Installing (Optional):

1. Run

   cd path/to/dask_gdf
   pip install .

Testing:

1. Ensure pytest is available: conda install pytest
2. cd path/to/dask_gdf
3. Run all tests with: pytest dask_gdf
4. Or, run individual tests with: pytest dask_gdf/tests/test_file.py