Syllabus

Course Information

Meeting Time: MWF 2:40 - 3:45 PM (165 Baskin Engineering)

Dates: Winter Quarter 2019 (1/7/19 - 3/15/19, holidays on 1/21/19 & 2/18/19)

Instructor: Scott Beamer (sbeamer@ucsc.edu)

Office Hours: MW 4:00 - 5:00 PM (229 Engineering 2) or by appointment

Homepage: https://cmpe293-winter19-01.courses.soe.ucsc.edu

Canvas (for submitting): https://canvas.ucsc.edu/courses/19203

Piazza (for discussion & help): https://piazza.com/ucsc/winter2019/cmpe293

Course Description

This graduate course will explore the current research and industrial landscape of programmable hardware accelerators. By specializing for a target domain, hardware accelerators can deliver outstanding efficiency improvements which makes them a promising way to cope with Moore's Law slowing down. This course will cover not only accelerators from a wide range of domains, but also foundational concepts to understand how to evaluate acceleration opportunities. Example accelerators include: deep learning (e.g. TPU), reconfigurable logic (FPGA and CGRA), bioinformatics, and graph processing. After completing this course, students will be able to analyze a given a target application and recommend an accelerator for it. The course will consist primarily of paper readings, paper discussions, student presentations, and course projects.

Student Learning Outcomes

After completing the course, the student will be able to ...

Characterize a workload - Recognize what computational patterns make up the workload. Determine which features are most relevant. Measure how well it is using a given platform. Identify its bottlenecks.

Understand how hardware accelerators provide benefit - Summarize technology trends and implementation options. Explain how architectural features can exploit workload characteristics to improve efficiency.

Suggest an accelerator for a target workload - Recall what broad classes of accelerators currently exist. Predict how well an accelerator will handle the target workload. Assess tradeoffs and select best accelerator given application needs.

In addition to the skills above, the course activities will also be synergistic in developing students' research skills including: reading papers critically, providing reviewer feedback, discussing contributions, directing projects, and presenting results.

Prerequisites

The only course prerequisite is knowledge of undergraduate computer architecture topics such as what is an ISA, pipelining, and caches. We especially encourage students from beyond the Computer Science and Engineering Department to take this course since their domain expertise will be a valuable contribution.

Grading

50% Project

20% Reading Summaries

10% Presenting Paper & Leading Discussion

 5% Scribing Paper Discussion

15% Participation & Attendance

Assignment Load

Project - Pairs of students will complete one research project over quarter, including: a proposal, two reports, one peer review, and one presentation.

Reading Summaries - Students will individually prepare and submit reading summaries by 9AM the day of a paper discussion. The top 20 scores will be kept, and the bottom few will be dropped.

Presenting Paper & Leading Discussion - Once per quarter, each student will individually present a paper from the reading list and lead a discussion about it.

Scribing Paper Discussion - Once per quarter, each student will individually take notes on a paper discussion and submit them to the instructor within one week of the discussion.

Notice of Accomodation

UC Santa Cruz is committed to creating an academic environment that supports its diverse student body. If you are a student with a disability who requires accommodations to achieve equal access in this course, please submit your Accommodation Authorization Letter from the Disability Resource Center (DRC) to me privately during my office hours or by appointment, preferably within the first two weeks of the quarter. At this time, I would also like us to discuss ways we can ensure your full participation in the course. I encourage all students who may benefit from learning more about DRC services to contact DRC by phone at 831-459-2089, or by email at drc@ucsc.edu.

Academic Integrity

As a researcher, you will be learning from and using prior work, but you must not plagiarize.

All work on the project (proposal, research, reports, peer review, and presentation) must be done by only you and your partner. All other work (reading summaries, leading paper discussion, scribing) must be done by the student individually.

The project must be original work, and if any prior work is used, it must be cited.

Frequently Asked Questions (FAQ)

1) How do I access papers off campus?

https://guides.library.ucsc.edu/offcampusaccess

2) What are slack days?

On the schedule, you will see some days listed as "slack days." Class will not be canceled on those days, but they are currently free to accommodate changes to the course schedule for things such as guest speakers. Expect the class activity to be specified at least 1 week in advance.

3) Will there be a final exam?

There will be no final exam, but please keep the exam slot open (3/19/18 4-7 PM) as it may be needed for presentations.