Seminar in Comp. Arch. - L6: Machine Learning-Driven Memory and Storage System Design (Spring 2026)

Onur Mutlu Lectures
Onur Mutlu LecturesMar 23, 2026

Why It Matters

Integrating machine learning into memory system design promises substantial performance gains and energy savings for data‑intensive workloads, reshaping the competitive landscape of cloud and high‑performance computing.

Key Takeaways

  • ML optimizes memory hierarchy for workload-specific patterns.
  • Processing-in-memory reduces data movement bottlenecks.
  • RowHammer mitigation benefits from predictive ML models.
  • Genome analysis accelerates via architecture-algorithm co-design.
  • Memory-centric computing reshapes future data-intensive applications.

Pulse Analysis

Machine learning is rapidly becoming a cornerstone of next‑generation memory architecture, moving beyond traditional heuristic‑based tuning. By leveraging predictive models, designers can dynamically allocate cache levels, prefetch data, and adjust refresh rates, aligning hardware behavior with real‑time workload characteristics. This adaptability not only trims latency but also curtails power consumption, a critical advantage as data centers grapple with escalating energy costs.

Processing‑in‑memory (PIM) and memory‑centric computing represent a paradigm shift where computation migrates closer to the data. Recent research, such as the "Modern Primer on Processing in Memory" and the "Memory‑Centric Computing" series, demonstrates that embedding AI accelerators within DRAM can slash data movement by orders of magnitude. For domains like genomics, where terabytes of raw sequences must be parsed, architecture‑algorithm co‑design—highlighted in Mutlu’s genome‑analysis papers—delivers speedups that traditional CPU‑only pipelines cannot match.

The practical implications extend to reliability and security. RowHammer, a well‑known DRAM disturbance issue, can be mitigated using machine‑learning classifiers that anticipate vulnerable rows before failure occurs. As memory densities climb, such proactive safeguards become indispensable. Collectively, these advances signal a future where intelligent memory systems drive both performance and resilience, positioning firms that adopt ML‑enhanced storage solutions at the forefront of the data‑driven economy.

Original Description

Seminar in Computer Architecture, ETH Zürich, Spring 2026 (https://safari.ethz.ch/architecture_seminar/spring2026/doku.php?id=schedule)
Lecture 6: Machine Learning-Driven Memory and Storage System Design
Lecturer: Rahul Bera, Rakesh Nadig, and Prof. Onur Mutlu
Date: 26 March 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...