A Scalable PyTorch Abstraction for Multi-GPU Gaussian Splatting
This work addresses the scalability bottleneck in Gaussian splatting for neural reconstruction, enabling larger and higher-resolution scenes for practitioners in computer vision and graphics.
The paper presents a multi-GPU Gaussian splatting approach that scales reconstruction to higher resolutions and larger scenes, achieving city-scale reconstructions with over 1 billion Gaussian splats—more than 25 times the current state of the art—by distributing parameters and operators across GPUs via CUDA unified memory and NVLink.
Gaussian splatting methods have become increasingly popular for neural reconstruction of the real world. However, they are often limited in scale and resolution due to compute and memory constraints. We present a multi-GPU Gaussian splatting approach that scales reconstruction to higher resolutions and larger scenes while abstracting away the code complexity typically associated with distributing a model. To accomplish this, we propose a PyTorch backend that distributes the Gaussian parameters and splatting operators across GPUs via CUDA unified memory and NVLink. Because distribution occurs at the operator level, the model code requires no explicit cross-device communication. More broadly, the backend exposes multiple GPUs as an aggregate PyTorch device and supports other PyTorch operators. We demonstrate city-scale reconstructions with street-level detail consisting of over 1 billion Gaussian splats, more than 25 times as many as the current state of the art.