NVIDIA RAPIDS 24.10 Enhances NetworkX and Polars with GPU Acceleration

The latest release of NVIDIA RAPIDS, version 24.10, brings significant enhancements to data science workflows by introducing GPU acceleration for NetworkX and Polars, among other features. This update emphasizes a seamless user experience for data scientists and developers, according to NVIDIA.

Zero Code Change NetworkX Acceleration

The RAPIDS cuGraph now offers GPU-accelerated NetworkX, which is generally available in this release starting with NetworkX 3.4. This upgrade enables end-to-end acceleration of graph workflows, significantly improving performance for large datasets. Users can activate this feature by setting the NX_CUGRAPH_AUTOCONFIG environment variable to True, allowing for substantial speedups in algorithms like betweenness centrality and PageRank.

Polars GPU Engine in Open Beta

The Polars GPU engine, powered by cuDF, is released in open beta, allowing users to experience up to 13x faster workflows with zero code change. This enhancement is integrated into the Polars Lazy API, enabling users to trigger GPU computation with the engine keyword.

UMAP for Larger Datasets

RAPIDS v24.10 extends the capability of cuML’s UMAP algorithm to handle datasets larger than GPU memory, preventing out-of-memory errors. This is achieved through a novel batched approximate nearest neighbor algorithm that processes data subsets on the GPU.

Improved cuDF-Pandas Compatibility

Enhancements in cuDF’s pandas accelerator mode now support true NumPy arrays, improving compatibility and eliminating previous workarounds. Additionally, cuDF now supports a wider range of PyArrow versions by utilizing the Arrow C Data Interface.

Guidelines for GPU Integration in CI Systems

NVIDIA has introduced new guidelines for integrating GPUs into GitHub-based continuous integration systems, leveraging GitHub Actions' support for hosted GPU runners. This facilitates easier integration and testing of RAPIDS libraries without local GPU hardware.

Platform Updates

The 24.10 release includes updates for compatibility with Python 3.12, NumPy 2.x, and other scientific computing software. However, it drops support for Python 3.9 and older versions of NCCL.

These updates in RAPIDS 24.10 continue to advance the accessibility of accelerated computing for data scientists and developers, offering enhanced performance and compatibility.