📢 cuDF can now be used as a no-code-change accelerator for pandas! To learn more, see here!
cuDF (pronounced "KOO-dee-eff") is a GPU DataFrame library for loading, joining, aggregating, filtering, and otherwise manipulating data. cuDF leverages libcudf, a blazing-fast C++/CUDA dataframe library and the Apache Arrow columnar format to provide a GPU-accelerated pandas API.
You can import cudf
directly and use it like pandas
:
import cudf
tips_df = cudf.read_csv("https://github.com/plotly/datasets/raw/master/tips.csv")
tips_df["tip_percentage"] = tips_df["tip"] / tips_df["total_bill"] * 100
# display average tip by dining party size
print(tips_df.groupby("size").tip_percentage.mean())
Or, you can use cuDF as a no-code-change accelerator for pandas, using
cudf.pandas
.
cudf.pandas
supports 100% of the pandas API, utilizing cuDF for
supported operations and falling back to pandas when needed:
%load_ext cudf.pandas # pandas operations now use the GPU!
import pandas as pd
tips_df = pd.read_csv("https://github.com/plotly/datasets/raw/master/tips.csv")
tips_df["tip_percentage"] = tips_df["tip"] / tips_df["total_bill"] * 100
# display average tip by dining party size
print(tips_df.groupby("size").tip_percentage.mean())
- Try cudf.pandas now: Explore
cudf.pandas
on a free GPU enabled instance on Google Colab! - Install: Instructions for installing cuDF and other RAPIDS libraries.
- cudf (Python) documentation
- libcudf (C++/CUDA) documentation
- RAPIDS Community: Get help, contribute, and collaborate.
See the RAPIDS install page for the most up-to-date information and commands for installing cuDF and other RAPIDS packages.
- CUDA 11.2+
- NVIDIA driver 450.80.02+
- Volta architecture or better (Compute Capability >=7.0)
cuDF can be installed via pip
from the NVIDIA Python Package Index.
Be sure to select the appropriate cuDF package depending
on the major version of CUDA available in your environment:
For CUDA 11.x:
pip install --extra-index-url=https://pypi.nvidia.com cudf-cu11
For CUDA 12.x:
pip install --extra-index-url=https://pypi.nvidia.com cudf-cu12
cuDF can be installed with conda (via miniconda or the full Anaconda distribution from the rapidsai
channel:
conda install -c rapidsai -c conda-forge -c nvidia \
cudf=24.10 python=3.11 cuda-version=12.5
We also provide nightly Conda packages built from the HEAD of our latest development branch.
Note: cuDF is supported only on Linux, and with Python versions 3.10 and later.
See the RAPIDS installation guide for more OS and version info.
See build instructions.
Please see our guide for contributing to cuDF.