Advanced multiprocessor architecture

CS-471 / 8 credits

Teacher: Falsafi Babak

Language: English

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

  • Introduction
  • Metrics and methodologies
  • Parallel programming models
  • Communication models
  • Applications and Workloads
  • Cache hierarchies & memory models
  • Memory & storage hierarchies
  • Interconnects
  • GPUs
  • AI/ML/Analytic accelerators
  • Near-memory computing
  • Datacenters & cloud Computing
  • Cloud-native CPU
  • Cloud-native memory hierarchies
  • Sustainable architecture

Learning Prerequisites

Recommended courses

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

Learning Outcomes

  • Explore the development trend of computation systems and datacenters
  • Establish the basic model to analyze the performance and characteristics of foundational workloads operating in cloud environments.
  • Classify and describe the components of modern parallel systems, including multiple processors, cache hierarchies, memory systems, interconnects, and accelerators, and their roles in handling emerging workloads
  • Define and clarify research questions and opportunities
  • Interpret and critique research papers and extract insights for research questions
  • Plan and conduct a research project
  • Present research contributions

Teaching methods

Lecture, research paper retrieval, and a research project

Assessment methods

  • Homework: 10%
  • Research project (in group): 40%
  • Midterm exam: 20%
  • Final exam: 30%

 

Resources

Websites

In the programs

  • Semester: Fall
  • Exam form: During the semester (winter session)
  • Subject examined: Advanced multiprocessor architecture
  • Lecture: 4 Hour(s) per week x 14 weeks
  • Project: 8 Hour(s) per week x 14 weeks
  • Type: optional
  • Semester: Fall
  • Exam form: During the semester (winter session)
  • Subject examined: Advanced multiprocessor architecture
  • Lecture: 4 Hour(s) per week x 14 weeks
  • Project: 8 Hour(s) per week x 14 weeks
  • Type: optional
  • Semester: Fall
  • Exam form: During the semester (winter session)
  • Subject examined: Advanced multiprocessor architecture
  • Lecture: 4 Hour(s) per week x 14 weeks
  • Project: 8 Hour(s) per week x 14 weeks
  • Type: optional
  • Semester: Fall
  • Exam form: During the semester (winter session)
  • Subject examined: Advanced multiprocessor architecture
  • Lecture: 4 Hour(s) per week x 14 weeks
  • Project: 8 Hour(s) per week x 14 weeks
  • Type: optional
  • Semester: Fall
  • Exam form: During the semester (winter session)
  • Subject examined: Advanced multiprocessor architecture
  • Lecture: 4 Hour(s) per week x 14 weeks
  • Project: 8 Hour(s) per week x 14 weeks
  • Type: optional
  • Semester: Fall
  • Exam form: During the semester (winter session)
  • Subject examined: Advanced multiprocessor architecture
  • Lecture: 4 Hour(s) per week x 14 weeks
  • Project: 8 Hour(s) per week x 14 weeks
  • Type: optional
  • Semester: Fall
  • Exam form: During the semester (winter session)
  • Subject examined: Advanced multiprocessor architecture
  • Lecture: 4 Hour(s) per week x 14 weeks
  • Project: 8 Hour(s) per week x 14 weeks
  • Type: optional
  • Semester: Fall
  • Exam form: During the semester (winter session)
  • Subject examined: Advanced multiprocessor architecture
  • Lecture: 4 Hour(s) per week x 14 weeks
  • Project: 8 Hour(s) per week x 14 weeks
  • Type: optional
  • Semester: Fall
  • Exam form: During the semester (winter session)
  • Subject examined: Advanced multiprocessor architecture
  • Lecture: 4 Hour(s) per week x 14 weeks
  • Project: 8 Hour(s) per week x 14 weeks
  • Type: optional
  • Semester: Fall
  • Exam form: During the semester (winter session)
  • Subject examined: Advanced multiprocessor architecture
  • Lecture: 4 Hour(s) per week x 14 weeks
  • Project: 8 Hour(s) per week x 14 weeks
  • Type: optional
  • Semester: Fall
  • Exam form: During the semester (winter session)
  • Subject examined: Advanced multiprocessor architecture
  • Lecture: 4 Hour(s) per week x 14 weeks
  • Project: 8 Hour(s) per week x 14 weeks
  • Type: optional
  • Exam form: During the semester (winter session)
  • Subject examined: Advanced multiprocessor architecture
  • Lecture: 4 Hour(s) per week x 14 weeks
  • Project: 8 Hour(s) per week x 14 weeks
  • Type: optional

Reference week

Tuesday, 10h - 12h: Lecture INF019

Thursday, 10h - 12h: Lecture INJ218

Related courses

Results from graphsearch.epfl.ch.