mJ/Frame 373fps 3D GS Processor Based on Shape-Aware Hybrid Architecture Using Earlier Computation Skipping and Gaussian Cache Scheduler

📋 论文概要

本文针对资源受限边缘设备上3D高斯散射（3D GS）渲染速度慢、能效低的问题，提出了一种基于形状感知混合架构的3D GS处理器。通过早期计算跳过和高斯分布优化技术，实现了每帧毫焦级能效和373fps的高帧率。

💡 主要创新点

• 提出形状感知混合架构，根据高斯分布形状动态调整计算资源分配，提升处理效率。
• 引入早期计算跳过机制，提前终止低贡献高斯点的渲染，减少冗余计算。
• 结合高斯分布特性优化数据流和光栅化流程，实现高性能低功耗边缘设备部署。

🏷 关键词

📄 原文摘要

applications like virtual reality and embodied AI. Unlike traditional Neural Radiance Fields (NeRF) [1], the novel 3D Gaussian Splatting approach (3D GS) [2] circumvents NeRF’s frequent sampling and intensive network inference. Hence, it demonstrates substantial accuracy and frame rate advantages on highperformance GPUs. However, the variability in Gaussian distribution shapes poses significant challenges on resource-constrained edge devices, as in Fig. 2.6.1. For example, 3D GS achieves only 6.4fps on edge Xavier NX. First, a complicated projection + rasterization dataflow is required to render the vast and diverse Gaussian distributions. Consequently, a 3D GS processor demands more kinds of operators and higher bitwidth than NeRF or networks. Additionally, the diversity in shapes leads to numerous input-dependent inefficient

👥 作者与机构

Xiaoyu Feng*, Hedi Wang*, Chen Tang, Tongda Wu, Huazhong Yang, Yongpan Liu

Tsinghua University, Beijing, China *Equally Credited Authors (ECAs) 3D rendering plays a crucial role in emerging