Qwen3.8-27B hits ~40 tok/s on dual RTX 3060, passes CUDA coding test
anderspitman · reddit · 2026-08-15
Reddit user anderspitman shares early performance report of Qwen3.8-27B on 2x RTX 3060 12GB. With specific llama.cpp config, 40 tok/s achieved. In a CUDA vector add smoke test, the model one-shot generated the program and installed compiler, but couldn't run due to VRAM limits. It attempted debugging, and eventually succeeded after shutting down llama-server.
Related event: Qwen3.8-27B Hits 50 tok/s on Dual RTX 3060s(2 posts)→
More from coding & agent
- Six-month prediction: hiring for generic agent harnesses built from scratch will cool — yi111 · 2026-10-16
- Omarchy ISO wouldn't boot on NVIDIA, so he had Claude patch it — indragie · 2026-10-16
- Dev uses Grok to clean up 3D printer firmware, catches it sneaking in nonsense code — yacineMTB · 2026-10-16
- A rude AI agent took 89% of a shared resource in 102-run multi-agent personality experiment — Brave-History-6502 · 2026-10-16
- Guardrails can make risky delegation work — but it takes a full-time team, says iandanforth — iandanforth · 2026-10-16
- What can you truly delegate? Only tasks with unambiguous, verifiable results — iandanforth · 2026-10-16