用类脑存储替代传统注意力，实现更快更省电的Transformer加速。
CAMformer: Associative Memory is All You Need
- 将注意力机制重构成类脑存储查询，通过模拟电荷共享快速匹配相似项。
- 在BERT和ViT上实测：能效提升超10倍，吞吐量最高提升4倍，面积降低6-8倍。
- 适合需要低功耗高并发的AI芯片设计，尤其适合边缘端部署。
Transformer因注意力机制中查询与键的稠密相似度计算导致二次方复杂度，面临可扩展性挑战。本文提出CAMformer，将注意力重新诠释为关联记忆操作，利用电压域二进制注意力内容寻址存储器（BA-CAM）计算注意力分数。该方法通过模拟电荷共享实现常数时间相似度搜索，以物理相似性感知取代数字运算。CAMformer融合分层两阶段top-k过滤、流水线执行与高精度上下文建模，在BERT和Vision Transformer工作负载上实现超过10倍能效提升，吞吐量最高提升4倍，面积降低6-8倍，同时保持近无损精度。
原文摘要 · Abstract (English)
Transformers face scalability challenges due to the quadratic cost of attention, which involves dense similarity computations between queries and keys. We propose CAMformer, a novel accelerator that reinterprets attention as an associative memory operation and computes attention scores using a voltage-domain Binary Attention Content Addressable Memory (BA-CAM). This enables constant-time similarity search through analog charge sharing, replacing digital arithmetic with physical similarity sensing. CAMformer integrates hierarchical two-stage top-k filtering, pipelined execution, and high-precision contextualization to achieve both algorithmic accuracy and architectural efficiency. Evaluated on BERT and Vision Transformer workloads, CAMformer achieves over 10x energy efficiency, up to 4x higher throughput, and 6-8x lower area compared to state-of-the-art accelerators--while maintaining near-lossless accuracy.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。