Table of Contents
Fetching ...

The ArborX library: version 2.0

Andrey Prokopenko, Daniel Arndt, Damien Lebrun-Grandié, Bruno Turcksin

TL;DR

ArborX 2.0 addresses the rigidity of earlier geometric search interfaces by introducing an API v2 that decouples data from the index via an IndexableGetter, enables per-match callbacks, and supports distributed MPI-based search, clustering, and ray tracing. The release expands geometry support, dimensionality, and precision, while modernizing the codebase and build system to leverage Kokkos 4.x and GPU backends. Key technical advances include a flexible BVH over arbitrary Value types, three query modes (including callback-based queries), and distributed execution with GPU-aware MPI. Performance gains come from architectural changes (64-bit Morton codes, Apetrei traversal, stackless BVH) and optimized sorting, yielding substantial speedups on large HPC datasets such as cosmology simulations, and enabling scalable, interactive spatial querying in multi-rank and accelerator-enabled environments.

Abstract

This paper provides an overview of the 2.0 release of the ArborX library, a performance portable geometric search library based on Kokkos. We describe the major changes in ArborX 2.0 including a new interface for the library to support a wider range of user problems, new search data structures (brute force, distributed), support for user functions to be executed on the results (callbacks), and an expanded set of the supported algorithms (ray tracing, clustering).

The ArborX library: version 2.0

TL;DR

ArborX 2.0 addresses the rigidity of earlier geometric search interfaces by introducing an API v2 that decouples data from the index via an IndexableGetter, enables per-match callbacks, and supports distributed MPI-based search, clustering, and ray tracing. The release expands geometry support, dimensionality, and precision, while modernizing the codebase and build system to leverage Kokkos 4.x and GPU backends. Key technical advances include a flexible BVH over arbitrary Value types, three query modes (including callback-based queries), and distributed execution with GPU-aware MPI. Performance gains come from architectural changes (64-bit Morton codes, Apetrei traversal, stackless BVH) and optimized sorting, yielding substantial speedups on large HPC datasets such as cosmology simulations, and enabling scalable, interactive spatial querying in multi-rank and accelerator-enabled environments.

Abstract

This paper provides an overview of the 2.0 release of the ArborX library, a performance portable geometric search library based on Kokkos. We describe the major changes in ArborX 2.0 including a new interface for the library to support a wider range of user problems, new search data structures (brute force, distributed), support for user functions to be executed on the results (callbacks), and an expanded set of the supported algorithms (ray tracing, clustering).

Paper Structure

This paper contains 16 sections, 1 table.