arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2204.10319cs.LGcs.CVcs.PF
收藏

TorchSparse:高效的点云推理引擎

TorchSparse: Efficient Point Cloud Inference Engine

Massachusetts Institute of Technology(麻省理工学院);Cornell University(康奈尔大学)

Haotian Tang, Zhijian Liu, Xiuyu Li, Yujun Lin, Song Han

摘要:

由于点云上的深度学习在AR/VR和自动驾驶中的广泛应用,其受到了越来越多的关注。这些应用需要低延迟和高准确性,以提供实时用户体验并确保用户安全。与传统的密集工作负载不同,点云的稀疏和不规则性质对在通用硬件上有效运行稀疏CNN提出了严峻的挑战。此外,用于2D图像的现有稀疏加速技术不能转化为3D点云。在本文中,我们介绍了TorchSparse,这是一个高性能的点云推理引擎,可以加速GPU上的稀疏卷积计算。TorchSparse直接优化了稀疏卷积的两个瓶颈:不规则计算和数据移动。该算法采用自适应矩阵乘法分组技术,使矩阵乘法的计算效率提高到1.4- 1.5倍。它还通过采用矢量化,量化和融合的位置感知内存访问来优化数据移动,将内存移动成本降低了2.7倍。在三个基准数据集的七个代表性模型上进行评估,TorchSparse分别实现了最先进的MinkowskiEngine和SpConv的1.6倍和1.5倍测量端到端加速。

英文摘要:

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.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑