Hardware

Understanding & Designing Modern Storage Systems - L3: MQSim (Spring 2026)

Onur Mutlu Lectures
Onur Mutlu Lectures•Mar 13, 2026

Why It Matters

Accurate SSD simulation enables faster, lower‑cost development of storage solutions critical for data‑intensive AI workloads and next‑generation data‑center infrastructure.

Key Takeaways

  • •MQSim accurately models multi‑queue NVMe SSD behavior for researchers.
  • •Existing simulators miss steady‑state and full latency modeling.
  • •MQSim error rates 6‑18% versus 68‑85% for others.
  • •Modular C++ design enables easy extension for new memory tech.
  • •Open‑source MIT license encourages community research and adoption.

Summary

The video introduces MQSim, an open‑source C++ framework designed to simulate modern multi‑queue SSDs with NVMe protocols. It contrasts MQSim with legacy SATA‑oriented simulators, highlighting how the latter fail to capture multi‑queue behavior, steady‑state operation, and full end‑to‑end request latency, leading to large performance prediction errors. Key insights include the three major gaps in prior tools—lack of multi‑queue support, inability to model steady‑state garbage collection and cache warm‑up, and incomplete latency modeling across the host‑to‑flash path. MQSim addresses these by modularly implementing front‑end components (host‑interface layer, request fetch unit, data cache manager) and back‑end elements (FTL, transaction scheduler, flash channel controller), while supporting emerging non‑volatile memories. The presenter cites validation results: when benchmarked against four real data‑center SSDs (two MLC, two TLC) pre‑conditioned to 70 % capacity, MQSim’s predictions deviated by only 6‑18 %, far better than the 68‑85 % errors of existing simulators. Its architecture allows swapping modules such as scheduling policies or memory technology models, and the MIT‑licensed code invites community contributions. For researchers and engineers, MQSim offers a reliable, extensible platform to evaluate SSD designs, firmware algorithms, and emerging storage media under realistic, steady‑state workloads, potentially accelerating innovation and reducing reliance on costly hardware prototypes.

Original Description

Project and Seminars Course: Understanding and Designing Modern Storage Systems, ETH Zürich, Spring 2026
Lecture 3: MQSim
Lecturer: Rakesh Nadig and Dr. Mohammad Sadrosadati
Date: March 13, 2026
Slides (pptx):
Slides (pdf):
Recommended Reading:
====================
A Modern Primer on Processing in Memory
Memory-Centric Computing: Solving Computing's Memory Problem
Memory-Centric Computing: Recent Advances in Processing-in-DRAM
Intelligent Architectures for Intelligent Computing Systems
RowHammer: A Retrospective
Fundamentally Understanding and Solving RowHammer
Accelerating Genome Analysis via Algorithm-Architecture Co-Design
From Molecules to Genomic Variations: Accelerating Genome Analysis via Intelligent Algorithms and Architectures
RECOMMENDED LECTURE VIDEOS & PLAYLISTS:
========================================
Digital Design and Computer Architecture Spring 2025 Livestream Lectures Playlist:
Fundamentals of Computer Architecture Fall 2025 Livestream Lectures Playlist:
Seminar in Computer Architecture Spring 2025 Livestream Lectures Playlist:
Computer Architecture Fall 2024 Lectures Playlist:
Interview with Professor Onur Mutlu:
TCuARCH meets Prof. Onur Mutlu
Arch. Mentoring Workshop @ISCA'21 - Doing Impactful Research
The Story of RowHammer Lecture:
Accelerating Genome Analysis Lecture:
Memory-Centric Computing Systems Tutorial at IEDM 2021:
Intelligent Architectures for Intelligent Machines Lecture:
Featured Lectures:

Comments

Want to join the conversation?

Loading comments...