GEMM-GS: Accelerating 3D Gaussian Splatting on Tensor Cores with GEMM-Compatible Blending

TL;DR

GEMM-GS accelerates 3D Gaussian Splatting by reformulating blending into GEMM-compatible operations to leverage GPU tensor cores, achieving significant speedups for real-time 3D scene rendering.

Contribution

It introduces a novel GEMM-compatible blending transformation and a high-performance CUDA kernel to accelerate 3D Gaussian Splatting on modern GPUs.

Findings

GEMM-GS achieves 1.42x speedup over vanilla 3DGS.

Combining GEMM-GS with existing methods yields an additional 1.47x speedup.

Code is publicly available at the provided GitHub link.

Abstract

Neural Radiance Fields (NeRF) enables 3D scene reconstruction from several 2D images but incurs high rendering latency via its point-sampling design. 3D Gaussian Splatting (3DGS) improves on NeRF with explicit scene representation and an optimized pipeline yet still fails to meet practical real-time demands. Existing acceleration works overlook the evolving Tensor Cores of modern GPUs because 3DGS pipeline lacks General Matrix Multiplication (GEMM) operations. This paper proposes GEMM-GS, an acceleration approach utilizing tensor cores on GPUs via GEMM-friendly blending transformation. It equivalently reformulates the 3DGS blending process into a GEMM-compatible form to utilize Tensor Cores. A high-performance CUDA kernel is designed, integrating a three-stage double-buffered pipeline that overlaps computation and memory access. Extensive experiments show that GEMM-GS achieves $1.42\times$ speedup over vanilla 3DGS and provides an additional $1.47\times$ speedup on average when combining with existing acceleration approaches. Code is released at https://github.com/shieldforever/GEMM-GS.