DFlash加速Qwen3.6：英特尔AI PC上MoE模型解码提速2.2倍
原标题：Accelerating Qwen3.6 on Intel® Core™ Ultra Series 3 with DFlash
AI 摘要
英特尔与Hugging Face合作，在搭载Intel Core Ultra（Panther Lake）的AI PC上，利用DFlash投机解码和OpenVINO加速Qwen3.6-35B-A3B MoE模型，实现HumanEval 2.2倍、GSM8K 1.6倍、MT-Bench 1.3倍的解码加速。DFlash通过并行掩码草稿器替代自回归草稿模型，缓解了MoE模型在投机解码中的专家加载开销。此外，DFlash对密集模型Qwen3.6-27B和Qwen3.5-9B的加速效果更显著，最高达5.5倍，代码将在OpenVINO.GenAI中发布。
正文节选
- Qwen3.6-35B-A3B is a highly efficient Mixture-of-Experts (MoE) model that brings strong coding, reasoning, and agentic capabilities to the AI PC, making it a stepping stone toward local AI. - MoE models are more challenging to accelerate with Speculative Decoding (SD): verifying several drafted tokens can pull in many more experts, dramatically inflating the effective active parameter count. - Running Qwen3.6-35B-A3B with DFlash SD and OpenVINO™ on an Intel® Panther Lake laptop, we achieve an