Content

• 

  Course Summary

  Parallel Architectures (PA) is a 10 credit course at Level 11, normally taken in Year 4. It runs in Semester 2. The exam is in April/May, and is worth 60% of the course mark. The University descriptor is here.
• 

  Course Outline

  In this course, you will learn about the principles that underpin parallel architectures of various kinds.

  	Unicore processor. We will briefly cover pipelining and superscalar processing.
  	Multicore processor. A significant chunk of the course will look at shared-memory multicores and techniques for exposing a  shared-memory interface to the programmer including: coherence protocols, consistency models, and synchronization operations.
  	GPUs. We will briefly look at Vector, SIMD and multithreaded processors and a modern-day incarnation of those ideas: GPUs.
  	AI Accelerators. We will briefly look at some of the recent efforts at specialising processors for machine-learning (e.g., TPUs).
  	Datacentre as a computer.  Time permitting, we will briefly look at how a datacentre can be understood as a massively parallel computer and how some of the ideas we learned (especially coherence and consistency) apply to datacentres.

  
  
• 

  Welcome to Parallel Architectures (Level 11) (2021-2022)[SEM2]

  Instructors: Vijay Nagarajan

  TA: TBD

  Piazza: Link TBA

  Course Contacts:

  • Please use Piazza for general technical and administrative questions about the course and coursework. 
  • For student-specific queries, Instructor email: vijay.nagarajan@ed.ac.uk
  • For student-specific queries relating to coursework 2, TA email: 

  Lecture times:

  • TBA

  Optional Textbooks

  • D.E. Culler, J.P. Singh, A. Gupta Parallel Computer Architecture: A Hardware/Software Approach, Morgan Kaufmann, 1998.
  •  J.L. Hennessey & D.A. Patterson Computer Architecture: a Quantitative Approach, 5th Edition, Morgan Kaufmann, 2011.
  • V. Nagarajan, D. J. Sorin, M. D. Hill, and D. A. Wood, A Primer on Memory Consistency and Cache Coherence, Second Edition, Morgan and Claypool, 2020 (free ebook)

  Coursework: There will be two courseworks associated with this course. 

  • First coursework (worth 15%, due: late Jan) will require you to read papers, analyse them, write reviews and make informed predictions. 
  • Second coursework (worth 25%, due: mid March) will involve programming.