16GB GPU runs Qwen3.8-27B at 110k context and 55 tok/s via iGPU display trick
jayecin · reddit · 2026-10-31
A Reddit user shares a complete recipe for running Qwen3.8-27B on a 16GB RTX 5070 Ti at 110k context and 55 tok/s. The key move: enable Hybrid Graphics so the 9800X3D's iGPU drives the display, freeing the full 16GB VRAM (Xorg alone ate 600-800MB) for llama.cpp with a 4.2BPW GGUF quant, q80/q40 KV cache quantization and flash attention. Games still run fine, and the author says it's fast enough to drop ChatGPT.
More from Infra
- Northflank hacked: users told to rotate all secrets as AI keys confirmed abused — basedjensen · 2026-10-31
- Arm exec: AI datacenters need a third 'agentic CPU' class as DPUs become security table stakes — BenBajarin · 2026-10-31
- Seroter's 900th reading list: hill-climbing agents and prefix caching costs — rseroter · 2026-10-31
- Post-training API ships end-to-end custom losses: define locally, run on hundreds of GPUs — ibab · 2026-10-31
- Qwen3.5-4B variant with RISC-V compiled weights trends on Hugging Face — divagr1925 · 2026-10-31
- Cerebras breaks down how it deploys multi-trillion parameter models on wafer-scale chips — pritisinghhhh · 2026-10-31