Virtuoso Logo
✨ 2nd Edition

Infrastructure for Evaluating
Novel HW/OS Interfaces

A full-day workshop and tutorial on the principles, methodologies, and practical aspects of hardware/OS co-design for memory management

📅 Date: Monday, March 23, 2026 (Full Day)
📍 Location: Room Ft. Duquesne, ASPLOS 2026, Pittsburgh, PA
--Days
--Hours
--Minutes
--Seconds

More information on the ASPLOS 2026 website.

About the Workshop & Tutorial

Building on the success of our first workshop at MICRO 2025, we are excited to bring the Virtuoso Workshop to ASPLOS 2026! This second edition will continue to explore the frontier of hardware/OS co-design for memory management, featuring new talks, updated demos, and fresh perspectives from leading researchers.

Traditional computing systems face significant challenges due to rigid interfaces between hardware and operating systems (OS). These interfaces struggle to meet the performance, efficiency, and security demands of modern applications. For example, the growth in data requirements has turned virtual memory (VM) into a major performance bottleneck.

This has led to a paradigm shift towards hardware/OS co-design, where hardware components and OS mechanisms are designed in tandem to optimize the system. This tutorial and workshop will provide a comprehensive introduction to this area, focusing on memory management.

A core component will be a hands-on exploration of Virtuoso, a simulation framework that enables rapid prototyping and evaluation of HW/OS co-designs. Published at ASPLOS 2025, Virtuoso provides a practical environment for attendees to experiment with co-design strategies and gain practical skills. The workshop is designed for students, engineers, and researchers in computer architecture and operating systems.

Organizers

Photo of Konstantinos Kanellopoulos

Konstantinos Kanellopoulos

ETH Zürich

Konstantinos Kanellopoulos is a PhD candidate at ETH Zurich, advised by Prof. Onur Mutlu. His research interests are at the intersection of hardware, software, and operating systems, focusing on performance, programmability, and security. More info on his webpage.

Photo of Prof. Onur Mutlu

Prof. Onur Mutlu

ETH Zürich

Onur Mutlu is a Professor of Computer Science at ETH Zürich. His research focuses on designing fundamentally energy-efficient, high-performance, and robust computing systems, with an emphasis on computer architecture, hardware security, and memory systems. He is an ACM Fellow, IEEE Fellow, and has received numerous honors and awards. He is passionate about making research and education widely accessible. More info on his webpage.

Talks Schedule & Invited Speakers

Speaker lineup and schedule will be announced closer to the event.

Coming Soon
🎤

Speakers To Be Announced

We are currently finalizing our lineup of distinguished speakers from academia and industry.
Check back soon for updates!

Interested in Speaking?

We welcome proposals from researchers working on hardware/OS co-design, virtual memory systems, memory management, and related topics. If you would like to present your work at the workshop, please reach out to the organizers.

Call for Submissions

📢 Submit Your Abstract

We invite researchers, engineers, and students to submit abstracts describing their work related to hardware/OS co-design, virtual memory, memory management, simulation infrastructure, and related topics. Accepted abstracts will be presented as short talks during the workshop.

  • What to submit: An abstract describing your work, key contributions, and relevance to the workshop themes.
  • Topics of interest: Hardware/OS co-design, virtual memory optimization, memory management techniques, simulation frameworks, address translation, TLB design, page table management, memory security, and related areas.
  • Submission deadline: March 1, 2026
  • Notification: Within one week of submission

Please send your abstract via email to the organizers. We look forward to your contributions!

Hands-On Demonstration Plan

The goal of the demo is to provide attendees with a practical, hands-on experience using Virtuoso to prototype and evaluate hardware/OS co-design techniques for memory management.

Part 1: Measuring Memory Management Overheads with eBPF

In this hands-on demo, attendees will use eBPF (extended Berkeley Packet Filter) to instrument and measure the overheads of memory management operations at the OS level. This software-based approach provides a lightweight, production-safe way to observe page faults and other memory management events in real time, without modifying the kernel. Participants will learn how to write and deploy eBPF programs to collect fine-grained performance data and identify bottlenecks in virtual memory management.

Part 2: Accelerating Page Faults in Hardware with Virtuoso

Building on the insights from Part 1, this demo will explore a hardware-accelerated approach to page fault handling. Attendees will use the Virtuoso simulation framework to prototype and evaluate a hardware mechanism that offloads page fault processing from the OS, significantly reducing the overhead identified in the first part. This hands-on session will demonstrate Virtuoso's full workflow, from defining a hardware/OS co-design to running simulations and analyzing performance results.

What to Expect

Based on our MICRO 2025 experience, attendees can expect:

  • Environment setup and quickstart for both eBPF tools and Virtuoso
  • Guided, step-by-step implementation of each demo
  • A complete journey from measuring real-system overheads to prototyping hardware solutions
  • Practical skills in both software instrumentation and hardware/OS co-design simulation

Prerequisites for Attendees

Detailed prerequisites will be announced closer to the event. In general, attendees should have:

Basic Requirements

  • Laptop with admin/root access (macOS, Linux, or Windows with WSL2)
  • A terminal application and text editor (we recommend VSCode)
  • Basic familiarity with C/C++ and command-line tools
  • Interest in computer architecture and/or operating systems

Options for Participation

  • Cloud VMs: We will provide access to pre-configured VMs (details closer to the event)
  • Local Setup: At least 8GB RAM and 40GB free disk space, with Git and Docker installed
  • Remote Participation: Follow along via livestream if you cannot meet the requirements