Xerxes: Extensive Exploration of Scalable Hardware Systems with CXL-Based Simulation Framework
Yuda An, Shushu Yi, Bo Mao, Qiao Li, Mingzhe Zhang, Diyu Zhou, Ke Zhou, Nong Xiao, Guangyu Sun, Yingwei Luo, Jie Zhang
Abstract
Compute Express Link (CXL) is an emerging industry standard that offers high-performance cache-coherent interconnects to heterogeneous devices, including host CPUs, computation accelerators, and memory devices. It aims to support high system scalability, peer-to-peer communication, and high-speed data transmission. To this end, the latest version of the CXL protocol introduces several new features, including port-based routing, device-managed coherence, and PCIe 6.0 support. However, the absence of CXL hardware and the methodological limitations of existing simulators hinder the exploration of these new architectures. To bridge this gap, we propose Xerxes, a novel simulation framework designed from the ground up to faithfully model the emerging features in the latest CXL protocol. It employs a dedicated interconnect layer to support interconnection within diverse system topologies. It also implements important components to conduct specific functions required by these features. Utilizing Xerxes, we comprehensively explore multiple aspects of CXL systems, including system topologies, device-managed coherences, and impacts of PCIe characteristics, and derive key observations that can inspire new designs of high-performance CXL systems. The codes of Xerxes are open-sourced and available at https://github.com/ChaseLab-PKU/Xerxes.
Ask about this paper
Your agent reads all of it.
Lune indexed this paper to the last equation, along with the top-tier papers that cite it. Ask a question and the answer quotes them.
Your agent calls
Luneget_paper_fulltext
Free to start. No credit card required.
Terminal
Install the CLIlune papers fulltext d0f7ea93-5448-49ed-be2d-c95569b7fa4aCited by top-tier papers1
Ask how each one uses itBuilds on14
- Language Models are Few-Shot LearnersTom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah et al.NeurIPS 2020 · 64,255 citations
- An Empirical Guide to the Behavior and Use of Scalable Persistent MemoryJian Yang, Juno Kim, Morteza Hoseinzadeh, Joseph Izraelevitz et al.FAST 2020 · 470 citations
- Pond: CXL-Based Memory Pooling Systems for Cloud PlatformsHuaicheng Li, Daniel S. Berger, Lisa Hsu, Daniel Ernst et al.ASPLOS 2023 · 328 citations
- TPP: Transparent Page Placement for CXL-Enabled Tiered-MemoryHasan Al Maruf, Hao Wang, Abhishek Dhanotia, Johannes Weiner et al.ASPLOS 2023 · 255 citations
- Demystifying CXL Memory with Genuine CXL-Ready Systems and DevicesYan Sun, Yifan Yuan, Zeduo Yu, Reese Kuper et al.MICRO 2023 · 133 citations
Related papers
- Cohet: A CXL-Driven Coherent Heterogeneous Computing Framework with Hardware-Calibrated Full-System SimulationYanjing Wang, Lizhou Wu, Sunfeng Gao, Yibo Tang et al.HPCA 2026 · 1 citation
- A Programming Model for Disaggregated Memory over CXLGal Assa, Moritz Lumme, Lucas Bürgi, Michal Friedman et al.ASPLOS 2026 · 4 citations
- vCXLGen: Automated Synthesis and Verification of CXL Bridges for Heterogeneous ArchitecturesAnatole Lefort, Julian Pritzi, Nicolò Carpentieri, David Schall et al.ASPLOS 2026 · 2 citations
- CXLMC: Model Checking CXL Shared Memory ProgramsSimon Guo, Conan Truong, Brian DemskyASPLOS 2026 · 1 citation
- C³: CXL Coherence Controllers for Heterogeneous ArchitecturesAnatole Lefort, David Schall, Nicolò Carpentieri, Julian Pritzi et al.HPCA 2026 · 1 citation
