一句话：提出 Melody 框架，跨 265 个负载、4 种 CXL 设备、7 个延迟档与 5 个 CPU 平台系统刻画并分析 CXL 内存性能。
CXL（Compute Express Link）已成为内存扩展的关键互连技术，但其在不同设备、延迟区间、处理器与工作负载下的性能影响仍缺乏系统理解。本文提出 Melody——一个用于系统化表征与分析 CXL 内存性能的框架，并基于 265 个负载、4 种真实 CXL 设备、7 个延迟级别和 5 个 CPU 平台的大规模评测展开研究。该工作给出多项发现：工作负载对亚微秒级 CXL 延迟（140–410ns）的敏感性、首次公开的 CXL 尾延迟数据、CPU 对 CXL 延迟的容忍度、用于定位 CXL 瓶颈的新方法 SPA，以及 CXL 场景下 CPU 预取器的低效问题。
Compute Express Link (CXL) has emerged as a pivotal interconnect for memory expansion. Despite its potential, the performance implications of CXL across devices, latency regimes, processors, and workloads remain underexplored. We present Melody, a framework for systematic characterization and analysis of CXL memory performance. Melody builds on an extensive evaluation spanning 265 workloads, 4 real CXL devices, 7 latency levels, and 5 CPU platforms. Melody yields many insights: workload sensitivity to sub-μs CXL latencies (140-410ns), the first disclosure of CXL tail latencies, CPU tolerance to CXL latencies, a novel approach (SPA) for pinpointing CXL bottlenecks, and CPU prefetcher inefficiencies under CXL.
@inproceedings{LiuHWBNJNL25,
title = {Systematic CXL Memory Characterization and Performance Analysis at Scale},
author = {Jinshu Liu and Hamid Hadian and Yuyue Wang and Daniel S. Berger and Marie Nguyen and Xun Jian and Sam H. Noh and Huaicheng Li},
booktitle = {ASPLOS},
year = {2025},
doi = {10.1145/3676641.3715987}
}