NVIDIA RAPIDS 24.10 Enhances NetworkX and Polars with GPU Acceleration

The most recent launch of NVIDIA RAPIDS, model 24.10, brings important enhancements to knowledge science workflows by introducing GPU acceleration for NetworkX and Polars, amongst different options. This replace emphasizes a seamless person expertise for knowledge scientists and builders, in response to NVIDIA.

Zero Code Change NetworkX Acceleration

The RAPIDS cuGraph now affords GPU-accelerated NetworkX, which is usually accessible on this launch beginning with NetworkX 3.4. This improve permits end-to-end acceleration of graph workflows, considerably enhancing efficiency for giant datasets. Customers can activate this characteristic by setting the NX_CUGRAPH_AUTOCONFIG atmosphere variable to True, permitting for substantial speedups in algorithms like betweenness centrality and PageRank.

Polars GPU Engine in Open Beta

The Polars GPU engine, powered by cuDF, is launched in open beta, permitting customers to expertise as much as 13x quicker workflows with zero code change. This enhancement is built-in into the Polars Lazy API, enabling customers to set off GPU computation with the engine key phrase.

UMAP for Bigger Datasets

RAPIDS v24.10 extends the aptitude of cuML’s UMAP algorithm to deal with datasets bigger than GPU reminiscence, stopping out-of-memory errors. That is achieved by means of a novel batched approximate nearest neighbor algorithm that processes knowledge subsets on the GPU.

Improved cuDF-Pandas Compatibility

Enhancements in cuDF’s pandas accelerator mode now help true NumPy arrays, enhancing compatibility and eliminating earlier workarounds. Moreover, cuDF now helps a wider vary of PyArrow variations by using the Arrow C Information Interface.

Pointers for GPU Integration in CI Methods

NVIDIA has launched new pointers for integrating GPUs into GitHub-based steady integration techniques, leveraging GitHub Actions’ help for hosted GPU runners. This facilitates simpler integration and testing of RAPIDS libraries with out native GPU {hardware}.

Platform Updates

The 24.10 launch consists of updates for compatibility with Python 3.12, NumPy 2.x, and different scientific computing software program. Nonetheless, it drops help for Python 3.9 and older variations of NCCL.

These updates in RAPIDS 24.10 proceed to advance the accessibility of accelerated computing for knowledge scientists and builders, providing enhanced efficiency and compatibility.

Picture supply: Shutterstock