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.
@article{arxiv.2511.19740,
title = {CAMformer: Associative Memory is All You Need},
author = {Tergel Molom-Ochir and Benjamin F. Morris and Mark Horton and Chiyue Wei and Cong Guo and Brady Taylor and Peter Liu and Shan X. Wang and Deliang Fan and Hai Helen Li and Yiran Chen},
journal= {arXiv preprint arXiv:2511.19740},
year = {2025}
}