| Category | Difficulty |
|---|---|
| Papers | 4 |
| Paper Presentation | 5 |
| Term Paper | 6 |
| Term Project | 7 |
What sets this class apart from other computer architecture courses at CMU is that this one is particularly research focused. It covers topics such as parallel computing architectures, cache coherence and memory consistency, heterogeneous systems and accelerators, as well as emerging topics such as approximate and neuromorphic computing, intermittent computting, etc.
Throughout the semester, students are required to read and write short critiques (~200-300 words) on 2 research papers for every class session. Every class session is a discussion facilitated by a student's presentation on the papers that were assigned to that day. At the beginning of the course, students sign up to do paper presentations on a paper of their choice. The discussion and the presentation together are a 25% of the final grade.
The critiques as well as the term paper are worth another 25% of the final grade together. The term paper is just an extended paper critique. It is no more than 5 single-column, 1.5-spaced, 10pt font pages in length, where you analyze in more detail one of the papers from the first half of the list of papers.
The last 50% of the grade is the term project. This can be done in groups of 2-3 and is pretty much self guided (as in you guys can pick whatever topic you want, there's a helpful list of topics to brainstorm from). There are deadlines in place for the project proposal, infrastructure check-in, and experiment plan check-in.
- Prof. Lucia has a good rundown set of slides on how to effectively read research papers. Given the sheer amount of papers that you read in this class, it's worth it to figure out what works best for you (which sections to focus on, which ones are skimmable, etc.) to get a solid understanding on the concepts presented.
- It's very easy to get distracted during class discussions, but it's in your best interest to participate or at the very least pay attention (since every class is a discussion and so if you don't you won't get anything out of the class😭).
- If stuck anywhere in the project process (whether it be coming up with an idea, scoping it down, or in the actual experiments) ask the TA! They'll generally be happy to help and it's better to do this earlier on rather than all towards the end. It's very easy to deprioritize it since there aren't a lot of intermediate deadlines for it.