Track 4 - Computer Architecture and Memory Systems

This series of lectures in Computer Architecture and Memory Systems will focus on building fundamentally better, data-​centric, data-​driven, and data-​aware architectures in computer systems.

You will learn:

• An introduction to intelligent architectures that are NOT overwhelmed by data
• How to architect fundamentally secure/reliable/safe systems
• How to architect fundamentally energy-​efficient systems
• How to architect fundamentally low-​latency and predictable architectures
• How to design accelerators for modern data-​intensive applications, e.g., AI/ML (including LLMs), Genomics, Medical & Health Systems, Graph Analytics, Data Analytics

The course will be in English with support from the SAFARI Research Group at ETH Zurich.

Nine 2-hour modules with lectures from the SAFARI Research Group and international collaborators.

Limited to 50 participants

Prerequisites: General knowledge on computer architectures

Agenda:

We will have 18 hours of lectures focused on computer architecture and memory systems. Below, you can find the detailed agenda, together with some related sample lectures from Prof. Mutlu’s courses. You can find the complete list of Prof. Mutlu’s courses on this link: https://safari.ethz.ch/courses/

• Basics of Modern Memory Systems: Caching, Memory, Storage
  • Duration: ~4 hours
  • Room: ETF E1
    
  • Sample lectures
    • Lectures 21-23: Digital Design and Computer Architecture Spring 2023
    • Lecture 27: Computer Architecture Fall 2022
    • FPGA-based Exploration of DRAM and RowHammer
    • Understanding and Designing Modern Storage Systems

• Memory reliability, security, safety
  • Duration: ~4 hours
  • Room: ETF E1
  • Sample lectures
    • Lectures 6-7: Computer Architecture Fall 2022
    • FPGA-based Exploration of DRAM and RowHammer

• Accelerating ML, Genomics, and data-intensive workloads: SIMD/GPU/Systolic Arrays
  • Duration: ~4 hours
  • Room: ETF E1
  • Sample lectures
    • Lectures 25, 26, and 28: Computer Architecture Fall 2022
    • Programming Heterogeneous Computing Systems with GPUs and other Accelerators
    • Accelerating Genome Analysis with FPGAs, GPUs, and New Execution Paradigms

• Memory-Centric Computing and Storage-Centric Computing
  • Duration: ~4 hours
  • Room: ETF E1
  • Sample lectures
    • Lectures 3-5: Computer Architecture Fall 2023
    • Lecture 6: Seminar in Computer Architecture Spring 2024
    • Data-Centric Architectures: Fundamentally Improving Performance and Energy

• Memory System for ML and ML for Memory System
  • Duration: ~2 hours (one lecture)
  • Room: ETF E1
  • Sample lectures
    • Lecture 10: Seminar in Computer Architecture Spring 2024

Prof. Mutlu has taught in several summer schools. Here are two examples where you can find the lectures and talks:

• Memory Systems and Memory-Centric Computing Systems — HiPEAC ACACES Summer School 2018
• Memory Systems and Memory-Centric Computing Systems: Fundamentals and Recent Research, TU Wien 2019
  

More info:

Program on SAFARI webpage