2×RTX 3090 + EPYC box running qwen3.8-flash-next at ~38 tok/s
Summary
The user describes a hardware configuration with dual RTX 3090 GPUs and an AMD EPYC CPU for running the Qwen3-Flash-Next model using llama.cpp, achieving 38 tokens per second in single-stream inference, and seeks advice on whether to add a third GPU or upgrade the CPU to improve performance, especially for running multiple parallel agents.
Similar Articles
Qwen3.8-Flash-Next turns 4xR9700 into a local AI powerhouse! 120 t/s TG and 12k t/s PP single request with optimized vLLM
The article reports that the Qwen3.8-Flash-Next model achieves 120 tokens/second generation speed and 12k tokens/second prefill on a system with 4x AMD R9700 GPUs using optimized vLLM and a custom Docker image.
Qwen3.8-Flash-Next (UD-IQ4_XS) on 2x RTX 3060 + 7800X3D, from initial 36 tps prefill to 400 tps and other benchmarks (-sm tensor trap) + VRAM/RAM usage
The article benchmarks llama.cpp and ik_llama.cpp for running the Qwen3.8-Flash-Next model on dual RTX 3060 hardware, showing that optimizing tensor splitting and ubatch settings can dramatically improve prefill speed from 36 t/s to 400 t/s, with comparisons of RAM usage.
Qwen3.8-Flash-Next on 2x3090 + DDR4: 17 → 25-29 t/s decode with the expert cache PR
User benchmarks and details a GPU-resident expert cache PR in llama.cpp that boosts decode speed for Qwen3.8-Flash-Next on a dual RTX 3090 system from 17 to 25-29 tokens per second.
@Oluwaphilemon1: Qwen3.8-Flash-Next running at 15 tok/s on a 12GB RTX 5070 apparently wasn’t acceptable. So instead of buying a bigger G…
A user built an open-source inference engine called Strata to optimize running Qwen3.8-Flash-Next on modest hardware, achieving up to 65.1 tok/s from an initial 15 tok/s.
Qwen 3.8 Flash Next-GSQ-RCO-IQ2_XS at ~21 tok/s on just an RTX 3060 12GB + 16GB DDR4 RAM(No gate pruning, 100% bit-exact)
A developer showcases a custom `--moe-direct-io` feature for llama.cpp that achieves 10–15x speedup (20–24 tok/s) running the 125B MoE Qwen3.5-Flash Next (IQ2_XS, 68GB) on just an RTX 3060 12GB with 16GB RAM, while keeping output bit-exact to stock.