Saved in:
Bibliographic Details
Main Authors: Gui, Hao, Hu, Lin, Chen, Rui, Huang, Mingxiao, Yin, Yuxin, Yang, Jin, Wu, Yong, Liu, Chen, Sun, Zhongxu, Zhang, Xueyang, Zhan, Kun
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2412.17378
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909603945512960
author Gui, Hao
Hu, Lin
Chen, Rui
Huang, Mingxiao
Yin, Yuxin
Yang, Jin
Wu, Yong
Liu, Chen
Sun, Zhongxu
Zhang, Xueyang
Zhan, Kun
author_facet Gui, Hao
Hu, Lin
Chen, Rui
Huang, Mingxiao
Yin, Yuxin
Yang, Jin
Wu, Yong
Liu, Chen
Sun, Zhongxu
Zhang, Xueyang
Zhan, Kun
contents 3D Gaussian Splatting (3DGS) is increasingly attracting attention in both academia and industry owing to its superior visual quality and rendering speed. However, training a 3DGS model remains a time-intensive task, especially in load imbalance scenarios where workload diversity among pixels and Gaussian spheres causes poor renderCUDA kernel performance. We introduce Balanced 3DGS, a Gaussian-wise parallelism rendering with fine-grained tiling approach in 3DGS training process, perfectly solving load-imbalance issues. First, we innovatively introduce the inter-block dynamic workload distribution technique to map workloads to Streaming Multiprocessor(SM) resources within a single GPU dynamically, which constitutes the foundation of load balancing. Second, we are the first to propose the Gaussian-wise parallel rendering technique to significantly reduce workload divergence inside a warp, which serves as a critical component in addressing load imbalance. Based on the above two methods, we further creatively put forward the fine-grained combined load balancing technique to uniformly distribute workload across all SMs, which boosts the forward renderCUDA kernel performance by up to 7.52x. Besides, we present a self-adaptive render kernel selection strategy during the 3DGS training process based on different load-balance situations, which effectively improves training efficiency.
format Preprint
id arxiv_https___arxiv_org_abs_2412_17378
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Balanced 3DGS: Gaussian-wise Parallelism Rendering with Fine-Grained Tiling
Gui, Hao
Hu, Lin
Chen, Rui
Huang, Mingxiao
Yin, Yuxin
Yang, Jin
Wu, Yong
Liu, Chen
Sun, Zhongxu
Zhang, Xueyang
Zhan, Kun
Computer Vision and Pattern Recognition
3D Gaussian Splatting (3DGS) is increasingly attracting attention in both academia and industry owing to its superior visual quality and rendering speed. However, training a 3DGS model remains a time-intensive task, especially in load imbalance scenarios where workload diversity among pixels and Gaussian spheres causes poor renderCUDA kernel performance. We introduce Balanced 3DGS, a Gaussian-wise parallelism rendering with fine-grained tiling approach in 3DGS training process, perfectly solving load-imbalance issues. First, we innovatively introduce the inter-block dynamic workload distribution technique to map workloads to Streaming Multiprocessor(SM) resources within a single GPU dynamically, which constitutes the foundation of load balancing. Second, we are the first to propose the Gaussian-wise parallel rendering technique to significantly reduce workload divergence inside a warp, which serves as a critical component in addressing load imbalance. Based on the above two methods, we further creatively put forward the fine-grained combined load balancing technique to uniformly distribute workload across all SMs, which boosts the forward renderCUDA kernel performance by up to 7.52x. Besides, we present a self-adaptive render kernel selection strategy during the 3DGS training process based on different load-balance situations, which effectively improves training efficiency.
title Balanced 3DGS: Gaussian-wise Parallelism Rendering with Fine-Grained Tiling
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2412.17378