Qwen3.8-27B hits 31t/s on AMD Strix Halo
stereohype · reddit · 2026-08-20
On a device with Ryzen AI Max+ 395 and Radeon 8060S, the author achieved 31.4 t/s decode speed for Qwen3.8-27B (Q5KXL) at 80W using the DFlash2 speculative decoding and Vulkan backend, with 300 t/s prefill. Benchmarks show DFlash2 is 40% faster than the built-in MTP, and surprisingly, Q5 quantization is faster than Q4 in this setup due to higher draft acceptance rates. Full startup commands and optimization tips are provided.
More from Infra
- Google's Willow chip cuts quantum errors 40x, raising Q-Day concerns for Bitcoin — jamestagg · 2026-10-16
- Vibe Coding a Data Pipeline Is Easy; Running It at Billions of Rows Is Not — Zachly · 2026-10-16
- Large data centers employ only ~50-150 permanent on-site staff, Microsoft says — brezina · 2026-10-16
- Prediction: Opus 5.5-level models on laptops by 2027, backed by Qwen 27B data — cephaloform · 2026-10-16
- Oracle down 12% in a week, nearly 50% off June high as OpenAI exposure spooks market — alvelda · 2026-10-16
- US data center moratoriums top 400 — an AI bull explains why they're actually good for AI — kevinsxu · 2026-10-16