cudf.pandas#
cuDF pandas accelerator mode (cudf.pandas) is built on cuDF and
accelerates pandas code on the GPU. It supports 100% of the
Pandas API, using the GPU for supported operations, and
automatically falling back to pandas for other operations.
%load_ext cudf.pandas
# pandas API is now GPU accelerated
import pandas as pd
df = pd.read_csv("filepath") # uses the GPU!
df.groupby("col").mean() # uses the GPU!
df.rolling(window=3).sum() # uses the GPU!
df.apply(set, axis=1) # uses the CPU (fallback)
Try it on Google Colab!#
Zero Code Change Acceleration Just |
Third-Party Library Compatible
|
Run the same code on CPU or GPU Nothing changes, not even your import statements, when going from CPU to GPU. |
100% of the Pandas API Combines the full flexibility of Pandas with blazing fast performance of cuDF |
Enable from the JupyterLab toolbar, no %load_ext needed
With the jupyterlab-nvdashboard extension, you can turn cudf.pandas
on for a notebook from the GPU Accel toolbar menu, without adding
%load_ext cudf.pandas or editing a single cell.
See JupyterLab Usage with jupyterlab-nvdashboard for setup instructions.
cudf.pandas is available as part of the cudf package. See the
installation and deployment guide
to get up-and-running with cuDF.
See Usage for other ways to enable cudf.pandas, including from the
command line and programmatically with cudf.pandas.install().