18-740 Computer Architecture – Fall 2015

Lectures of 740 will meet at HH 1107 (in Pittsburgh) and B23 118 (in Silicon Valley) on Mondays and Wednesdays

Computer architecture is the science and art of designing, selecting and interconnecting hardware components and co-designing the hardware/software interface to create a computer that meets functional, performance, energy consumption, cost, and other specific goals. This course qualitatively and quantitatively examines fundamental computer design trade-offs, with the goal of developing an understanding that will enable students to perform cutting-edge research in computer architecture. We will learn, for example, how uniprocessors execute many instructions concurrently, how state-of-the-art memory systems deliver data into the processor and why they are so complex, and how/why multiple processors are interconnected to execute portions of a program or multiple programs in parallel, as done in modern multi-core processors. Examining trade-offs requires that you already know how to correctly design a computer, as is taught in the important prerequisite 18-447. This course will involve an at least two-month long research/implementation project in which students work in groups of 2-3.

• Number of Units: 12
• Prerequisites: 18-447 or equivalent (review background material)
• Undergraduate Course Designation: Breadth, Depth, Coverage
• Undergraduate Course Area: Computer Hardware Engineering

Subject to change.

• Syllabus

• Lectures (Pittsburgh): M & W 7:30pm to 9:20pm , Hammerschlag Hall 1107
• Recitation (Pittsburgh): T 7:30pm to 9:20pm, DH A302

• Lectures (SV): M & W 4:30pm to 6:20pm, B23 118
• Recitation (SV): T 4:30pm to 6:20pm, B19 1065

Arguably, there is no perfect textbook for computer architecture. I encourage you to do your own research, consult multiple sources, question assumptions and statements, and talk with me and TAs whenever you have questions.

Lecture notes and videos will serve as the main source of information and they will provide the required references to textbooks or other reading material (such as research articles).

A good source of information on all covered topics is the research articles that introduced or built upon the covered topic. These articles are usually published in top conferences (such as ISCA, MICRO, ASPLOS, HPCA) or journals (such as IEEE or ACM Transactions). I strongly encourage you to dig out the original source of the covered topics as well as the research that builds upon it. This will help you become a successful and well-read researcher in computer architecture/systems. When in doubt, ask questions. The following textbooks could be very useful as supplements to lectures. Some lecture readings will be assigned from them:

• Parallel Computer Organization and Design, by Dubois, Annavaram, and Stenstrom, Cambridge University Press
• Modern Processor Design: Fundamentals of Superscalar Processors, by Shen and Lipasti

Other useful readings could include the following:

• Computer Architecture: A Quantitative Approach, Fifth Edition by Hennessy and Patterson, Morgan Kaufmann/Elsevier
• Computer Architecture and Implementation by Harvey Cragon, Cambridge University Press
• Computer Organization by Hamacher, Vranesic, Zaky, McGraw-Hill
• Structured Computer Organization by Andrew Tanenbaum, Prentice Hall
• High Performance Computer Architecture by Harold Stone, Prentice Hall (dated but good reference)
• Readings in Computer Architecture by Hill, Jouppi, Sohi, Morgan Kaufmann (dated but good reference with reprints of papers)
• Parallel Computer Architecture: A Hardware/Software Approach, by Gupta, Culler, Singh, Morgan Kaufmann

• Mailing List: 740-official@ece.cmu.edu (sent to instructors and TAs)