Mochi: Collision Detection for Spherical Particles using GPU Ray Tracing

2024-02-22

AI Summary

In discrete collision detection (DCD) for spherical particles, GPU ray-tracing (RT) hardware—specifically RT Cores—cannot be directly leveraged for BVH traversal due to their inaccessibility and non-programmability in conventional DCD pipelines. Method: This work pioneers the adaptation of RT Core acceleration to DCD by proposing an RT-aware broad-phase framework that exploits ray traversal to implicitly perform efficient BVH queries, coupled with a lightweight, custom sphere–sphere intersection test as the narrow-phase module. Contribution/Results: The approach bypasses traditional BVH construction and explicit traversal constraints, enabling end-to-end hardware-software co-optimization. Experiments on standard particle simulation benchmarks demonstrate 1.8×–3.2× speedup over state-of-the-art alternatives—including Taichi’s uniform grid, hash-based mapping, and optimized BVH implementations—thereby advancing the innovative use of GPU hardware primitives in physics simulation.

Technology Category

• Machine Learning: Hardware-aware ML
• Search and Optimization: Sampling/Simulation-based Search
• Computer Vision: Low Level & Physics-based Vision

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Responsible Web: Data and user privacy-enhancing technologies for the Web
• Systems and Infrastructure for Web, Mobile and WoT: Sustainability and carbon-aware systems for Web, mobile, and WoT

Abstract

Efficient Discrete Collision Detection (DCD) uses indexing structures for acceleration, and developing these structures demands meticulous programmer efforts to achieve performance. The Ray-Tracing (RT) architecture of GPUs builds and traverses an indexing structure called Bounding Volume Hierarchy (BVH) and performs geometric intersection tests, which are all the essential components of a DCD kernel. However, BVHs built by the RT architecture are neither accessible nor programmable; the only way to use this architecture is to launch rays and map DCD queries to ray traversal. Despite these challenges, we developed an RT-accelerated DCD framework, Mochi, for handling spherical objects. Mochi optimizes collision detection by utilizing hardware-accelerated BVH traversal in the broad phase and introducing a novel object-object intersection test in the narrow phase. We evaluate Mochi showing speedups on all of our end-to-end particle simulation benchmarks when compared to uniform grid and hash map implementations in Taichi, a high-performance framework targeting graphics applications, and the state-of-the-art BVH implementation.

Problem

• Develops GPU-accelerated collision detection for spherical particles
• Overcomes limitations of non-programmable BVH in ray-tracing architecture
• Optimizes broad-phase BVH traversal and narrow-phase intersection tests

Innovation

• Utilizes GPU ray tracing for collision detection
• Employs hardware-accelerated BVH traversal
• Introduces novel object-object intersection test