Advanced multiprocessor architecture

Summary

Multiprocessors are basic building blocks for all computer systems. This course covers the architecture and organization of modern multiprocessors, prevalent accelerators (e.g., GPU, TPU), and datacenters. It includes a research project on multiprocessors and post-Moore era datacenters.

Content

Learning Prerequisites

Recommended courses

• Advanced computer architecture
• Systems for data management and data science

Important concepts to start the course

Performance evaluation, Programming models, Processors, Memory hierarchies, Accelerators, Datacenters

Learning Outcomes

By the end of the course, the student must be able to:

• Design and evaluate parallel computer organizations and benchmark parallel software.
• Analyze the performance characteristics of foundational cloud workloads.
• Quantify and measure performance metrics of parallel systems and architectures.
• Explore the components of modern parallel systems: processors, cache hierarchies, memory systems, interconnects, and accelerators. 
• Contextualise how these components are adapted for handling cloud and emerging workloads. 
• Investigate system, architecture, and workload trends in post-Moore computing systems and datacenters.
• Interpret and critique research papers to find insights and define open research questions.
• Plan , propose, and conduct an empirical research project related to advanced parallel systems.
• Present and communicate original research contributions.

Teaching methods

Lecture, research paper discussions, and research projects

Expected student activities

• Self-motivated participation in course and discussion
• Critical thinking and examination over research papers
• Group discussion and collaboration on course projects

Assessment methods

• Homework (paper reviewing and write-up): 20%
• Research project (in group): 30%
• Midterm exam: 20%
• Final exam: 30%

Resources

Websites

• https://parsa.epfl.ch/course-info/cs471/