Efficient Differentiable Hardware Rasterization for 3D Gaussian Splatting

TL;DR

This paper introduces a fast, memory-efficient differentiable hardware rasterizer for 3D Gaussian Splatting that significantly accelerates backward pass computations while maintaining low memory usage, suitable for resource-constrained devices.

Contribution

The authors develop a novel hardware rasterization method with programmable blending and hybrid gradient reduction, achieving over 10x faster backward rasterization and 3x overall speedup compared to traditional approaches.

Findings

Over 10x faster backward rasterization than naive atomic operations

3.07x acceleration in full pipeline execution on RTX4080 GPUs

Higher gradient accuracy with 16-bit render targets compared to float32

Abstract

Recent works demonstrate the advantages of hardware rasterization for 3D Gaussian Splatting (3DGS) in forward-pass rendering through fast GPU-optimized graphics and fixed memory footprint. However, extending these benefits to backward-pass gradient computation remains challenging due to graphics pipeline constraints. We present a differentiable hardware rasterizer for 3DGS that overcomes the memory and performance limitations of tile-based software rasterization. Our solution employs programmable blending for per-pixel gradient computation combined with a hybrid gradient reduction strategy (quad-level + subgroup) in fragment shaders, achieving over 10x faster backward rasterization versus naive atomic operations and 3x speedup over the canonical tile-based rasterizer. Systematic evaluation reveals 16-bit render targets (float16 and unorm16) as the optimal accuracy-efficiency trade-off, achieving higher gradient accuracy among mixed-precision rendering formats with execution speeds second only to unorm8, while float32 texture incurs severe forward pass performance degradation due to suboptimal hardware optimizations. Our method with float16 formats demonstrates 3.07x acceleration in full pipeline execution (forward + backward passes) on RTX4080 GPUs with the MipNeRF 360 dataset, outperforming the baseline tile-based renderer while preserving hardware rasterization's memory efficiency advantages -- incurring merely 2.67% of the memory overhead required for splat sorting operations. This work presents a unified differentiable hardware rasterization method that simultaneously optimizes runtime and memory usage for 3DGS, making it particularly suitable for resource-constrained devices with limited memory capacity.