A Programming Model for Disaggregated Memory over CXL
Open Access
- 22 March 2026
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 41-58
Abstract
CXL (Compute Express Link) is an emerging open industry-standard interconnect between processing and memory devices that is expected to revolutionize the way systems are designed. It enables cache-coherent, shared memory pools in a disaggregated fashion at unprecedented scales, allowing algorithms to interact with various storage devices using simple loads and stores. While CXL unleashes unique opportunities, it also introduces challenges of data management and crash consistency. For example, CXL currently lacks an adequate programming model, making it impossible to reason about the correctness and behavior of systems on top. In this work, we present CXL0, the first programming model for concurrent programs over CXL. We propose a high-level abstraction for memory accesses and formally define operational semantics. We demonstrate that CXL0 captures a wide range of current and future CXL setups and perform initial measurements on real hardware. To illustrate the usefulness of CXL0, we present a general transformation that enhances any linearizable concurrent algorithm with durability in a distributed partial-crash setting. We believe that this work will serve as a stepping stone for systems design and programming on top of CXL.Keywords
All Related Versions
Funding Information
- Israel Science Foundation (814\/22)
- European Research Council (851811)
This publication has 63 references indexed in Scilit:
- Demystifying a CXL Type-2 Device: A Heterogeneous Cooperative Computing Perspective2024
- Āpta: Fault-tolerant object-granular CXL disaggregated memory for accelerating FaaS2023
- Disaggregated Memory in the Datacenter: A SurveyIEEE Access, 2023
- Methodology for Evaluating the Potential of Disaggregated Memory Systems2022
- View-Based Owicki–Gries Reasoning for Persistent x86-TSO2022
- Linearizability of Persistent Memory Objects Under a Full-System-Crash Failure Model2016
- FDR3 — A Modern Refinement Checker for CSP2014
- System-level implications of disaggregated memory2012
- Power, Programmability, and Granularity: The Challenges of ExaScale Computing2011
- A Better x86 Memory Model: x86-TSO2009