RDNA4 local inference hits ~100 tok/s running Qwen3.8 Flash on dual R9700
Public_Umpire_1099 · reddit · 2026-09-14
A developer shipped an R9V update pushing local RDNA4 inference to 100 tok/s text generation with Qwen3.8 Flash Next IQ4XS on dual R9700 GPUs plus 128GB RAM, and added Q4KXL support at 50 tok/s.
Key fixes and notes:
- Crashes resolved via n-gram SSD streaming, with improved diagnostics and pinned image support
- Torture-tested for 12 hours with no instability found; Q4KXL still needs more tuning
- The author is also building an inference engine and a 6-month-old deep research/site-builder app, and will return to R9V after shipping those
- Community members in the Launch80 Discord report even better numbers with other quants, suggesting these non-fully-VRAM-resident configs are nearing their ceiling
More from Infra
- Your AI agent would cost 260% more without prompt caching, Arena data shows — arena · 2026-10-30
- Nebius runs auction-driven GPU pricing, lifting rates ~20% amid tight supply to 2028 — BenBajarin · 2026-10-30
- macOS update silently re-adds eMudhra root certs users had removed — steipete · 2026-10-30
- Bittensor's Terza Inference routes one API across miners' models on subnet 103 — const_reborn · 2026-10-30
- SGLang Summit 2026 sells out sponsor booths and Full Passes ahead of Nov 12–13 event — ying11231 · 2026-10-30
- OpenRouter rebuilds provider routing: 46% fewer capacity errors, 16.9% faster TTFT — steipete · 2026-10-30