Presentation
Complete tutorial slides
The full teaching deck, covering ray tracing fundamentals and four connected spatial database case studies.
Tutorial materials
Start with the slides for the complete tutorial narrative, or read the concise paper for the motivation, design space, and related systems.
PVLDB Paper
Tutorial paper
A compact overview of the tutorial scope, audience, case studies, learning goals, and supporting materials.
From graphics hardware to database systems
Modern database systems are increasingly shaped by domain-specific hardware, but effective use of that hardware requires new abstractions, programming methods, and algorithmic reformulations. This tutorial explains how ray tracing cores in commodity GPUs can be repurposed to accelerate spatial data operations by treating search and refinement tasks as ray-geometry intersection problems.
Four connected case studies create a path from hardware principles and ray tracing programming, through query operators and index design, to deployment in an industrial spatial database. The emphasis is not only on performance, but on the systems lessons behind making a specialized accelerator useful to database developers.
What the tutorial covers
The material moves from first principles to a production-oriented system, with each section building on the same ray-based view of spatial processing.
-
01
Ray tracing fundamentals
RT cores, BVH traversal, the rendering pipeline, and NVIDIA OptiX.
-
02
Spatial joins
Line-segment intersection, point-in-polygon, and limited-precision challenges.
-
03
Spatial indexing
Point and range queries, updates, mutability, and load balancing.
-
04
Spatial functions
Hausdorff distance as a case study in RT-accelerated spatial computation.
-
05
Apache SedonaDB
Engineering a GPU-friendly spatial engine and integrating acceleration end to end.
-
06
Lessons and opportunities
What generalizes, what remains difficult, and where the research can go next.
Explore the systems
The tutorial is grounded in open implementations. Use these repositories to reproduce results, inspect the engineering, or build on the ideas.
RayJoin
Ray tracing acceleration for spatial joins, including line-segment intersection and point-in-polygon.
LibRTS
A spatial indexing library built on ray tracing hardware.
X-HD
Fast Hausdorff distance computation with ray tracing cores.
SedonaDB GPU Spatial
The GPU spatial library and integration path used by Apache SedonaDB.
Presenters
The tutorial brings together the researchers behind the systems and the broader hardware-conscious data management research program.
Rubao Lee
Senior Research Scientist
The Ohio State University
Liang Geng
Ph.D. Candidate
The Ohio State University
Xiaodong Zhang
Professor
The Ohio State University
Suggested citation
Rubao Lee, Liang Geng, and Xiaodong Zhang. "Hardware Acceleration for Spatial Databases: Building Next-Generation Query Engines Using Ray Tracing Cores." PVLDB 19(12): 4862-4866, 2026.