FlashQLA是基于TileLang开发的高性能线性注意力内核，专为提升个人设备上智能体AI性能而设计。它实现了2-3倍的前向传播加速和2倍的反向传播加速。其核心技术包括门控驱动的片上自动计算与通信重叠、硬件友好的代数重构，以及TileLang融合的Warp专用内核。该设计通过自动片上通信重叠显著提升了流处理器利用率，在张量并行、小模型和长上下文任务中效果突出。尽管在大批量处理时，其将GDN流程拆分为两个内核的策略会带来额外内存开销，但在边缘设备和长上下文实际场景中性能更优。反向传播部分通过构建16级、严格片上内存限制下的Warp专用流水线，实现了超过2倍的内核级加速。
Qwen 把线性注意力的推理效率压到了新台阶，2-3 倍加速对想做本地 Agent 的开发者是实打实的，不是论文灌水，是能跑在设备上的代码。
🚀 Introducing FlashQLA: high-performance linear attention kernels built on TileLang.
⚡ 2–3× forward speedup. 2× backward speedup.
💻 Purpose-built for agentic AI on your personal devices.
💡Key insights:
1. Gate-driven automatic intra-card CP.
2. Hardware-friendly algebraic reformulation.
3. TileLang fused warp-specialized kernels.
FlashQLA boosts SM utilization via automatic intra-device CP. The gains are especially pronounced for TP setups, small models, and long-context workloads.
Instead of fusing the entire GDN flow into a single kernel, we split it into two kernels optimized for CP and backward efficiency. At large batch sizes this incurs extra memory I/O overhead vs. a fully fused approach, but it delivers better real-world performance on edge devices and long-context workloads.
The backward pass was the hardest part: we built a 16-stage warp-specialized pipeline under extremely tight on-chip memory constraints, ultimately achieving 2×+ kernel-level speedups.
We hope this is useful to the community!🫶🫶
Learn more:
📖 Blog: https://qwen.ai/blog?id=flashqla
💻 Code: https://github.com/QwenLM/FlashQLA
来源：Qwen · x.com