articleschannelstagsspacestoolkit
NVIDIA Technical Blog

NVIDIA RAPIDS 24.10 Introduces Accelerated NetworkX with Zero Code Change, Updates for UMAP and cuDF-Pandas

View Original
thumbnail

Table of Contents

  • Zero code change accelerated NetworkX
  • Polars GPU engine in open beta
  • Bringing UMAP to larger-than-GPU-memory datasets
  • Improved cuDF pandas compatibility with NumPy and PyArrow
  • Guidelines for incorporating GPUs into GitHub-based CI systems
  • RAPIDS-wide support for Python 3.12 and NumPy 2.x

Zero code change accelerated NetworkX

NetworkX accelerated by RAPIDS cuGraph is now generally available in the 24.10 release, providing GPU-accelerated graph creation, a new user experience, and expanded documentation for accelerated graph workflows with large graphs. The seamless transition between CPU and GPU offers improved performance. Users can explore the benchmarks and learn more about this feature in the documentation.

Polars GPU engine in open beta

The Polars GPU engine is now available in open beta, allowing users to configure Polars to utilize the GPU for computations with the keyword. This integration enhances performance and enables smooth compatibility with code relying on the NumPy C API.

Bringing UMAP to larger-than-GPU-memory datasets

RAPIDS v24.10 introduces enhanced support for UMAP on datasets larger than GPU memory, expanding the capabilities of data processing and analysis for large-scale datasets.

Improved cuDF pandas compatibility with NumPy and PyArrow

cuDF now supports a range of PyArrow versions, providing improved arrow compatibility and flexibility for users working with different versions. The update also ensures compatibility with NumPy 1.x and 2.x.

Guidelines for incorporating GPUs into GitHub-based CI systems

GitHub Actions now support hosted GPU runners, enabling projects to leverage NVIDIA GPUs for CI workloads. The RAPIDS Deployment documentation offers detailed guidance on setting up GPU-powered workflows, optimizing performance, and best practices for incorporating GPU CI into projects.

RAPIDS-wide support for Python 3.12 and NumPy 2.x

The latest release of RAPIDS extends support to Python 3.12, NumPy 2.x, fmt 11, and spdlog 1.14, allowing users to leverage the latest features and optimizations in their data science workflows. New users can refer to the provided resources to get started with RAPIDS.