Guardado en:
Detalles Bibliográficos
Autores principales: Fu, Yuzhe, Ye, Hancheng, Guo, Cong, Zhang, Junyao, Wang, Qinsi, Lin, Yueqian, Zhou, Changchun, Hai, Li, Chen, Yiran
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:https://arxiv.org/abs/2604.17720
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866918455137009664
author Fu, Yuzhe
Ye, Hancheng
Guo, Cong
Zhang, Junyao
Wang, Qinsi
Lin, Yueqian
Zhou, Changchun
Hai
Li
Chen, Yiran
author_facet Fu, Yuzhe
Ye, Hancheng
Guo, Cong
Zhang, Junyao
Wang, Qinsi
Lin, Yueqian
Zhou, Changchun
Hai
Li
Chen, Yiran
contents Point-based Neural Networks (PNNs) have become a key approach for point cloud processing. However, a core operation in these models, Farthest Point Sampling (FPS), often introduces significant inference latency, especially for large-scale processing. Despite existing CUDA- and hardware-level optimizations, FPS remains a major bottleneck due to exhaustive computations across multiple network layers in PNNs, which hinders scalability. Through systematic analysis, we identify three substantial redundancies in FPS, including unnecessary full-cloud computations, redundant late-stage iterations, and predictable inter-layer outputs that make later FPS computations avoidable. To address these, we propose \textbf{\textit{FlashFPS}}, a hardware-agnostic, plug-and-play framework for FPS acceleration, composed of \textit{FPS-Prune} and \textit{FPS-Cache}. \textit{FPS-Prune} introduces candidate pruning and iteration pruning to reduce redundant computations in FPS while preserving sampling quality, and \textit{FPS-Cache} eliminates layer-wise redundancy via cache-and-reuse. Integrated into existing CUDA libraries and state-of-the-art PNN accelerators, \textit{FlashFPS} achieves 5.16$\times$ speedup over the standard CUDA baseline on GPU and 2.69$\times$ on PNN accelerators, with negligible accuracy loss, enabling efficient and scalable PNN inference. Codes are released at https://github.com/Yuzhe-Fu/FlashFPS.
format Preprint
id arxiv_https___arxiv_org_abs_2604_17720
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle FlashFPS: Efficient Farthest Point Sampling for Large-Scale Point Clouds via Pruning and Caching
Fu, Yuzhe
Ye, Hancheng
Guo, Cong
Zhang, Junyao
Wang, Qinsi
Lin, Yueqian
Zhou, Changchun
Hai
Li
Chen, Yiran
Machine Learning
Computer Vision and Pattern Recognition
Point-based Neural Networks (PNNs) have become a key approach for point cloud processing. However, a core operation in these models, Farthest Point Sampling (FPS), often introduces significant inference latency, especially for large-scale processing. Despite existing CUDA- and hardware-level optimizations, FPS remains a major bottleneck due to exhaustive computations across multiple network layers in PNNs, which hinders scalability. Through systematic analysis, we identify three substantial redundancies in FPS, including unnecessary full-cloud computations, redundant late-stage iterations, and predictable inter-layer outputs that make later FPS computations avoidable. To address these, we propose \textbf{\textit{FlashFPS}}, a hardware-agnostic, plug-and-play framework for FPS acceleration, composed of \textit{FPS-Prune} and \textit{FPS-Cache}. \textit{FPS-Prune} introduces candidate pruning and iteration pruning to reduce redundant computations in FPS while preserving sampling quality, and \textit{FPS-Cache} eliminates layer-wise redundancy via cache-and-reuse. Integrated into existing CUDA libraries and state-of-the-art PNN accelerators, \textit{FlashFPS} achieves 5.16$\times$ speedup over the standard CUDA baseline on GPU and 2.69$\times$ on PNN accelerators, with negligible accuracy loss, enabling efficient and scalable PNN inference. Codes are released at https://github.com/Yuzhe-Fu/FlashFPS.
title FlashFPS: Efficient Farthest Point Sampling for Large-Scale Point Clouds via Pruning and Caching
topic Machine Learning
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2604.17720