TorchSparse: Efficient Point Cloud Inference Engine

TL;DR

TorchSparse is a GPU-accelerated inference engine that significantly improves the efficiency of sparse convolution computations on point clouds, enabling real-time processing for AR/VR and autonomous driving applications.

Contribution

It introduces novel optimization techniques for irregular computation and data movement, achieving substantial speedups over existing sparse CNN frameworks.

Findings

Achieves 1.6x and 1.5x end-to-end speedup over MinkowskiEngine and SpConv.

Reduces memory movement cost by 2.7x through optimized data access.

Provides consistent performance improvements across multiple models and datasets.

Abstract

Deep learning on point clouds has received increased attention thanks to its wide applications in AR/VR and autonomous driving. These applications require low latency and high accuracy to provide real-time user experience and ensure user safety. Unlike conventional dense workloads, the sparse and irregular nature of point clouds poses severe challenges to running sparse CNNs efficiently on the general-purpose hardware. Furthermore, existing sparse acceleration techniques for 2D images do not translate to 3D point clouds. In this paper, we introduce TorchSparse, a high-performance point cloud inference engine that accelerates the sparse convolution computation on GPUs. TorchSparse directly optimizes the two bottlenecks of sparse convolution: irregular computation and data movement. It applies adaptive matrix multiplication grouping to trade computation for better regularity, achieving 1.4-1.5x speedup for matrix multiplication. It also optimizes the data movement by adopting vectorized, quantized and fused locality-aware memory access, reducing the memory movement cost by 2.7x. Evaluated on seven representative models across three benchmark datasets, TorchSparse achieves 1.6x and 1.5x measured end-to-end speedup over the state-of-the-art MinkowskiEngine and SpConv, respectively.