Metal-Gauss trains 3D Gaussian splats on Mac in minutes, no CUDA needed, beats msplat on PSNR-per-minute
CSProfKGD · x · 2026-09-02
NandoMetzger released Metal-Gauss, an open-source project that trains 3D Gaussian Splatting natively on Apple Silicon using Metal kernels compiled at runtime — no CUDA, no Xcode, just Command Line Tools. It may be the fastest Mac-native Gaussian optimizer to date.
Benchmarked on all 8 NeRF-synthetic scenes with identical seed clouds on the official 200-view test split, it dominates on quality-per-minute: PSNR 21.5 at 30 seconds (vs 12.8-19.9 for msplat, Brush, spirula), 29.8 at 3 minutes, and converges to 31.9 within 15 minutes. Code is on GitHub and contributions are welcome.
More from Infra
- SGLang Summit sponsors grow from 11 to 36 in one week; Lip-Bu Tan, Lilian Weng to speak — BanghuaZ · 2026-10-18
- Yacine jokes his datacenter KPI is gigawatt-hours: get paged if not burning enough power — yacineMTB · 2026-10-18
- 1.68x inference speedup drops, directly accelerating GRPO post-training — yacineMTB · 2026-10-18
- OpenRouter is only a fraction of the total addressable market, poster argues — sloppenheimer · 2026-10-18
- Developer builds change alerts for LLM prices and retirements across Bedrock, Azure and Vertex — jsandhu0780 · 2026-10-18
- 12.3M tokens in 27 minutes for $0.14: agent session with 99.6% cache hit — teortaxesTex · 2026-10-18