ASPLOS 2026 Tutorial

MemSysExplorer - a community research framework for evaluating next-generation memory systems

Mar 23, 2026

MemSysExplorer is a comprehensive framework designed to promote memory-centric computing research across the different layers of the design stack via user-friendly characterization, simulation, and design space exploration tools. Our goal is to bring together experts from both industry and academia and enable cross-discipline collaborations between researchers in memory devices, circuits, computer architecture, computer systems, programming languages and compilers. This tutorial will showcase the up-to-date capabilities and main functionalities of MemSysExplorer. Domain experts will learn how to interface with specific parts of the framework and run full-system evaluations, as well as make contributions and feature requests to our ongoing community effort.

Tentative Half-Day Outline

  • Introduction and Framework Overview (30 min)
  • Technology-level Characterization Demo & Case Study (45 min)
    • Code structure overview and how-to-use
    • Technology database, tour of available configurations and features
    • Demo: adding a new memory technology
    • Demo: fault-injection simulation example
  • Application-level Profilers Demo (45 min)
    • Code structure overview and how-to-use
    • Application profilers and tour of available features, how to contribute
    • Demo: collecting and comparing cache statistics across tools
    • Demo: integrating a new benchmark
  • System-level Evaluation Examples (1 hour)
    • Case Study A: CPU cache technology case study (configure, evaluate, and explore data end-to-end)
    • Case Study B: GPU main memory characterization case study (configure, evaluate, explore data end-to-end)
    • Time for questions, additional data exploration / optimization according to audience feedback
  • Solicitation for more case studies, user data contributions (30 min)
    • As part of this, demo interactive web-based data visualizations with additional example studies, available datasets, and more
    • Optionally, configure and run additional case studies live according to user interests