CIS 5650 GPU Programming and Architecture
Fall 2024
Course Announcements
2023-05-11: Course Waitlist
Advance registration for CIS 5650 Fall 2024 is full. Please use the Course Waitlist to ensure that you are considered for the course and receive course communication ahead of the Fall semester. Additionally, complete the Student Survey linked below.
2023-05-11: Student Survey (Required)
All students who are interested in taking the course (registered, waitlisted, or otherwise) must complete the Student Survey as soon as possible. Failure to submit the survey before first day of class can jeopardize your registration for CIS 5650.
Course Description
A timeline section from the following topics:
- GPU Computing: GPU architecture, massively parallel programming, parallel algorithms, performance.
- Rendering: Graphics pipeline (rasterization), path tracing, deferred shading, forward+ rendering, VR.
- APIs: CUDA, WebGL, RTX, Vulkan.
This is a project-intensive course with significant coding, writing, and presenting. It is more work than any other course, but it is worth it.
For a course more focused on GPU architecture without graphics, see Joe Devietti’s CIS 601 (no longer offered at Penn).
Prerequisites
- Passion for computer graphics.
- At least one of:
- CIS 4600/5600: Introduction to Computer Graphics.
- CIS 4610/5610: Advance Rendering
- Preferably received an A. Knowledge of rasterization and ray tracing.
- Strong C or C++.
- If you have not completed these courses (or equivalents), it is strongly advised that you complete the courses first and take the CIS 5650 course the following year. In case you are unable to take the course in due time for your graduation plans, then it is strongly advised that you thoroughly complete the following projects before starting CIS 5650.
- Also useful:
- CIS 3800: Operating Systems
- CIS 5010: Computer Architecture
Student Survey (Required)
If you have registered as a student for the course, or plan to, please complete this required survey: CIS 5650 Fall 2024 Student Survey.
Github, Schedule, Class Forum, and LinkedIn
- Github: fork your repos from here
- LinkedIn Group: for networking with current and previous course students
Lecturers
Changes to office hour schedule will be made on the CIS 5650 Calendar and notified through Class Forum.
Shehzan Mohammed mza@seas.upenn.edu
Office Hours:
- 60 minutes prior to class and 30 minutes following the end of class; Held adjacent to Towne 337.
- By appointment (email me)
Teaching Assistants
Han “Andrew” Yang hanyoung@seas.upenn.edu
Office Hours:
- Monday - 2:00pm - 3:30pm
- Friday 2:00pm - 3:00pm
- Levine 057
Xiaoxiao “Crytal” Zou xizou@seas.upenn.edu
Office Hours:
- Thursday - 10:00am - 11:30am
- Tuesday - 12:30 - 2:00pm
- Levine 057
Aditya Gupta adityag1@seas.upenn.edu
Office Hours:
- Wednesday - 3:30pm - 5:00pm
- Thursday - 1:30pm - 3:00pm
- Levine 057
Course Advisor
Recommended Reading
- Moving Graphics Research into Development, Patrick Cozzi
- How to Make an Attractive GitHub Repository, Patrick Cozzi
No books are required, but course material comes from many sources including:
- Programming Massively Parallel Processors, Third Edition, 2016, David Kirk and Wen-mei Hwu.
- Real-Time Rendering, Fourth Edition, 2018, Tomas Akenine-Möller, Eric Haines, Naty Hoffman, Angelo Pesce, Michał Iwanicki, and Sébastien Hillaire.
- Ray Tracing Gems, First Edition, 2019, Eric Haines and Tomas Akenine-Möller. (Free PDF distributed under CC 4.0 License)
Other useful tools and material:
- Ray Tracing in One Weekend (Free PDF + Github)
- NVIDIA GTC On-Demand
- NVIDIA GDC 2019 Courses (Mostly Real Time Ray Tracing)
- Machine Learning by Andrew Ng (Coursera free)
- Deep Learning Book (Free to read in HTML)
- CUDA Programming Guide
- All previous CIS 5650 (https://cis565-fall-20XY.github.io)
Grading
- Projects: 50%
- Final Project: 45%
- Participation: 5%
Academic Integrity
An academic integrity violation will result in the student receiving an F in this course.
See Academic Integrity at the University of Pennsylvania: A Guide for Students.
Code submissions will be cross-checked for plaigarism against previous years’ submissions as well as submissions from your colleagues using automated software. Please do not copy code.
If you would like to use code not written by you for this class, please run it by the TAs using Class Forum for permission to use it. Examples of code you will need to ask permission to use:
- Public Github Repositories and other open source projects.
- Projects from other classes that will give you a non-trivial advantage for the project.
If you think you need to ask permission, you should ask. We will most likely approve all reasonable requests.
Penn Engineering COVID-19 Impact and Announcements
Please visit the COVID-19 for information on Penn Engineering’s announcements and information about COVID-19 impact and procedures.
COVID Protocols for CIS 5650
- Classrooms will be operating under normal, pre-COVID capacity during the Fall 2024 term.
- CIS 5650 will be in person and lecture recordings will be made available within 24 hours after the lecture.
- We will follow University policy for mask use.
- If a student (or staff) tests positive at any point during the semester, and there is probable risk of spread, a portion of the students, if not the entire class may be moved online-only temporarily. You will be notified through Class Forum.
Acknowledgments
Joe Kider, Gary Katz, and Suresh Venkatasubramanian taught this course before me.
All former TAs have helped shape this course:
Previous students have provided significant course feedback including:
Many passionate folks in our field have also provided course input:
- Eric Haines, (@pointinpolygon)
- Christophe Riccio, (@g_truc)
- Johan Andersson, (@repi)
- Quarup Barreirinhas, (@quarup)
- Wolfgang Engel, (@wolfgangengel)
- Mikkel Gjoel, (@pixelmager)
- Dominik Lazarek, (@Omme)
- Emil Persson, (@_Humus_)
- Sebastien Vandenberghe (@sebavanmicrosof)
- Chrome GPU Team, which includes many former CIS 5650 Alumni.