← All publications
MICRO, 20263DGS

DeGS: A Scalable 3DGS Architecture via Decoupled Workload Parsing and Reorganization

Minnan Pei, Gang Li, Zeyu Zhu, Siting Wang, Junwen Si, Zhuoran Song, Yu Feng, Fangxin Liu, Xiaoyao Liang and Jian Cheng

Overview

DeGS speeds up high-resolution Gaussian rendering by grouping useful pixel work before blending, so more processing elements can run in parallel. In 3D Gaussian Splatting (3DGS), irregular footprints and early pixel termination leave parallel processing elements idle. DeGS decouples workload parsing from blending and uses workload reorganization to pack the surviving pixel operations together. The paper measures scaling from 16 to 1,024 processing elements, at resolutions from 720p to 8K.

高分辨率新视角渲染时,增加处理单元并不总能提速。DeGS 在像素混合前重组有效工作,提高三维高斯溅射加速器的并行利用率。对三维重建后的高斯场景合成新视角时,高斯覆盖的区域不规则,每个像素结束混合的时刻也不同。DeGS 将覆盖和透射率检查与混合分开,评估了 16 至 1,024 个处理单元以及 720p 至 8K 分辨率下的扩展性。

  • 3D Gaussian Splatting
  • Computer architecture
  • Scalability

MICRO, 2026

Research by Minnan Pei (裴旻楠) and collaborators · Email