EE282: Computer Systems Architecture

Spring 2024, Mondays and Wednesdays, 9:30 - 10:50 AM, Thornton 102

Instructor: Caroline Trippel
Teaching Assistant: Yao Hsiao, Akshit Goel

Canvas: Class Homepage
Ed: Class Homepage
GradeScope: Class Homepage
Poll Everywhere (for in-class students): We will use Poll Everywhere to facilitate in-class discussions and quizzes. Please get set up at this link before the first lecture.
(April 8th) SCPD students have 1-week to finish the in-class exercises using assignments on Canvas within 1 week after the lecture recording is available.
Syllabus: PDF (Updated office hour on April 1st)

EE282 covers key hardware and software topics that are relevant for architecting advanced computer systems. Examples include: out-of-order processors, multi-level cache hierarchies, main memory systems, multithreading and vector techniques, GPUs and accelerators, virtualization, virtual memory, datacenter architecture, advanced IO systems, and non-volatile storage. Programming assignments include: simulation of multi-core processor designs to characterize performance, design and implementation of advanced processor optimizations, and formal verification of an open-source processor.
  EE282 is appropriate for undergraduate and graduate students specializing in the broad field of computer systems. It is also appropriate for other EE and CS students who want to understand, program, and make efficient use of modern computer systems of any scale in their day-to-day work.

Schedule

Required Textbook:
HP: J. Hennessy & D. Patterson, Computer Architecture: A Quantitative Approach, 6th edition.
MC: Morgan Claypool Synthesis Lectures (available through the library using your SUID).

Date	[Lecture #] Topic	Reading Assignment Additional Reading	Reference Problem/s [Due date]	Slides/ Notes
4/1	[1] Introduction, Review of Prerequisites	[Prerequisite material] HP: Appendices A.{1-8}, B.{1-3}, C.{1-2, 4, 6}, Sections 1.{1-13}		Slides (updated 04/01) Recording
4/3	[2] Advanced Caches I: Multi-level caches, Optimizations	HP: Sections 2.{1, 3, 6}, Appendix B.{3, 6}		Slides Recording
4/8	[3] Advanced Caches II: Optimizations (continued), Prefetching, Coherence problem	HP: Sections 2.5, 5.{1, 2, 4} A Hierarchical Neural Model of Data Prefetching Evolution of the Samsung Exynos CPU Microarchitecture	Problems 1,2 [4/16]	Slides Recording
4/10	[4] Cache coherence, Synchronization, and Memory consistency I	HP: Section 5.5 MC: A Primer on Memory Consistency and Cache Coherence, Second Edition: Chapters 2, 6 Protogen: automatically generating directory cache coherence protocols from atomic specifications HeteroGen: Automatic Synthesis of Heterogeneous Cache Coherence Protocols	Problem 3 [4/23] Prog. Assign. 1 [5/7]	Slides Recording
4/15	[5] Cache coherence, Synchronization, and Memory consistency II	HP: Section 5.{6-7} MC: A Primer on Memory Consistency and Cache Coherence, Second Edition: Chapters 3, 4, 5.{1-2}	Problem 4 [4/23]	Slides Recording
4/17	[6] Main Memory	HP: Section 2.2	Problem 5 [4/30]	Slides Recording
4/22	[7] Out-of-Order Processors I: Dynamic Scheduling and Branch Prediction	HP: Section 3.{1-4}	Problem 6 [4/30]	Slides Recording
4/24	OoO Processors (Cont.)		Problem 7 [4/30][5/03] (Apr 28 update) Prog. Assign. 2 [5/16] [5/23] (Apr 24 update)	Recording
4/29	[8] Out-of-Order Processors II: Speculation and Superscalar	HP: Section 3.{6-10}	Problem 8 [5/3][5/14] (Apr 28 update)	Slides Recording
5/1	[9] Multithreading	HP: Section 3.{11-12}		Slides Recordings
5/6	Midterm Exam (9:30 - 10:50 AM PST) Thornton 102	Lectures 1-10 1-9
5/8	[11] AI Processors Guest Lecturer: Ardavan Pedram	CACM: Domain-Specific Hardware Accelerators by Prof. Bill Dally et. al.	Problem 10 [5/14]	Slides Recordings
5/13	[12] GPUs Guest Lecturer: Michael Pellauer	HP: Section 4.{4,6-8}	Problem 11 [5/21]	Slides Recordings
5/15	[13] Accelerators Guest Lecterer: Sneha Goenka	HP: Section 7.{1-4} Roofline: an insightful visual performance model for multicore architectures	Problem 12 [5/21]	Slides Recording
5/20	[10] Vectors	HP: Section 4.{1-3,5}, and Appendix G	Problem 9 [5/28]	Slides Recording
5/22	[14] Virtualization	MC: Hardware and Software Support for Virtualization: Chapters 1,2,4,5; HP: 6.{1-4,7}	Problem 13 [5/28] Prog. Assign. 3 [6/6]	Slides Recording
5/27	Memorial Day (No Classes)
5/29	[15] Advanced Memory Management	MC: Hardware and Software Support for Virtualization: Chapters 1,2,4,5; HP: Appendix L	Problem 14 [6/4]	Slides Recording
6/3	[16] Advanced I/O & Networking	HP: Appendix F.{1-2} MC: Hardware and Software Support for Virtualization: Chapter 6	Problem 15 [6/8]	Slides
6/5	[17] Non-volatile memory	HP: Section 2.2, Appendix D.2 Flash storage memory by Adam Leventhal	Problem 16 [6/8]
6/11	Final Exam (8:30 - 10:30 AM PST) TBD	Lectures 1-17

Homework and Projects

Assignments include a problem set and 3 programming assignments. The problem set consists of 16 questions. Each problem within the set has a reference lecture and the corresponding due date is put up on the lecture schedule. You will work on all assignments in groups of 2 students. All assignments are due by 11:59pm PT on the dates indicated on the assignment in the schedule. Solutions to problem sets will be available online shortly thereafter. All assignments should be submitted through GradeScope. You will be granted a total of 3 “late days” to use however you see fit for eligible coursework, except that you may use up to 1 late day for Problem 7, 15, and 16 (04/29 update) (to ensure timely release of homework solutions before the Midterm). Beyond the allotted late days, you will be docked on the following system without exception:

1 day late: 25% of the maximum allowable marks

2 days late: 50% of the maximum allowable marks

3+ days late: 75% of the maximum allowable marks

Logistics

Announcements: Visit this web page regularly to access all the handouts, solutions, and announcements. Please check your email regularly as well for announcements from Ed!

Office Hours:

Caroline Trippel: Wednesdays (10:50am - 11:50am, Gates 470) (starting 04/03/2024) or by appointment.

Yao Hsiao: Following slots (starting 04/02/2024) or by appointment

• Tuesdays 9:00am - 10:00am, Gates 259
• Mondays 10:50am - 11:50am on Zoom

Akshit Goel: Thursdays 1:30pm - 2:30pm, Packard 107

You may request virtual appointments if needed for COVID-related reasons.

Review Sessions:

Fridays, 9:00AM - 10:00AM, over zoom and it is recorded.

Exams: There will be two in-person exams: one Midterm and one Final.

• Midterm on Monday, May 6th (9:30AM - 10:50AM PST) in Thornton 102, covers lectures 1-10.
• Final on Tuesday, June 11th (9:30AM - 10:30AM PST) in TBD, covers lectures 1-18.

Grading:

Category	Grade %	Notes
Attendance and Participation	5%	3 conditional absence “passes” (see Course Policies); Arrival/departure > 10 min late/early counts as an absence Is not eligible for late days
Problem Set	15%	Is eligible for late days
3 Programming Assignments	30%	Is eligible for late days
Midterm	25%	Is not eligible for late days Can be revised for extra credit
Final	25%	Is not eligible for late days Cannot be revised for extra credit

Participation:
You are expected to attend the lectures in person. This is your opportunity to ask questions, contribute answers and insights, and affect the nature of the class. Also note that the lecture notes and the textbook are not guaranteed to capture 100% of every topic discussed aloud during lecture.
The lecture notes will be made available before class. During the lectures, there will also be in-class exercises which you will be given time to work on and discuss with your neighbors. You will submit answers via Poll Everywhere. You will only be graded based on your participation. SCPD students are expected to complete the in-class exercises within 1 week after the lecture recording is available.

Collaboration: See: honor code and collaboration for some general guidelines, which apply to both project assignments and problem sets. In general, collaboration is encouraged subject to the following guidelines:

• No more than 2 people can collaborate on a homework or project assignment.
• Students working together should submit a single assignment for the pair.
• Any assistance received for homework or programming assignment solutions should be acknowledged in writing with specific details.
• No sharing of code, or partial or complete solutions among groups is permitted.

SCPD Video Recording Disclaimer: Video cameras located in the back of the room will capture the instructor presentations in this course. For your convenience, you can access these recordings by logging into the course Canvas site. These recordings might be reused in other Stanford courses, viewed by other Stanford students, faculty, or staff, or used for other education and research purposes. Note that while the cameras are positioned with the intention of recording only the instructor, occasionally a part of your image or voice might be incidentally captured. If you have questions, please contact a member of the teaching team.

Adapted from a template by Andreas Viklund.