Engram conditional memory has emerged as a promising component for LLMs by decoupling static knowledge lookup from dynamic computation. Since Engram exhibits sparse access patterns and supports prefetching, its massive embedding tables are well-suited for offloading to lower-tier memory. In this paper, we propose using Compute Express Link (CXL) memory pool for Engram storage. Compared to RDMA, CXL provides fine-grained and low-latency access required by minimal and discrete retrieval patterns of Engram. We integrate the CXL-based Engram pool into SGLang, achieving near-DRAM end-to-end performance. This provides a scalable and cost-efficient storage solution for future Engram-integrated LLMs without compromising inference performance.
@article{arxiv.2603.10087,
title = {Pooling Engram Conditional Memory in Large Language Models using CXL},
author = {Ruiyang Ma and Teng Ma and Zhiyuan Su and Hantian Zha and Xinpeng Zhao and Xuchun Shang and Xingrui Yi and Zheng Liu and Zhu Cao and An Wu and Zhichong Dou and Ziqian Liu and Daikang Kuang and Guojie Luo},
journal= {arXiv preprint arXiv:2603.10087},
year = {2026}
}