raketenkater/ggrun

llama.cpp/ik_llama.cpp launcher: loads big MoE models across mismatched multi-GPU rigs by exact VRAM math.

3 best points · 0 comments · 2 Hacker News posts · 283 GitHub stars · View on GitHub

Hacker News discussions

• Ggrun v3 is 65% faster than Ollama (3 points, 0 comments, by raketenkater, 2026-06-23)
• Llama.cpp flags auto-tuning tool (3 points, 0 comments, by raketenkater, 2026-06-26)