cugraph.dask.common.read_utils.get_chunksize#

cugraph.dask.common.read_utils.get_chunksize(input_path)[source]#

Calculate the appropriate chunksize for dask_cudf.read_csv to get a number of partitions equal to the number of GPUs.

Examples

>>> import cugraph.dask as dcg
>>> chunksize = dcg.get_chunksize(datasets_path / 'netscience.csv')