Presentation and Grading Details

[TANSixu]

Grading Overview

• Basic 70 (graded in presentation)
• Comprehensive 30 (graded after code submission)
• • simple 10 (partial score allowed)
• • medium 10 (partial score allowed)
• • complex 10 (partial score allowed)
• Bonus 20
• • Design 10 (graded in presentation)
• • Implementation 10 (graded after code submission)

Presentation

Each group will have 10min to present their work. The presentation mainly consists of two parts:

• Slides presentation
• Code inspection and Q&A

Hence, you will need to prepare simple slides and runnable code for your presentation. For most parts, you only need to show the screenshots of your testing results. But you may be required to run one or more tests on your presentation randomly. So Do Not Cheat with screenshots, misbehaviour may be considered plagiarism.

Slides content

For the sake of saving time, we will partially grade your work based on the slides (running every test and reading through code on presentation is time-consuming). We only grade your content, so please keep your slides simple and concise. You may include the following information in your slides:

1. Introduction to your design

Describe the working flow of your work, and any important components, for instance:

• classes you add
• data structure
• your timer
• your header (if you extend the standard header)
• ...

2. Handshaking check

• Show your test result screenshot from basic_handshaking_test
• If you modified the testing scripts, state changes and reasons.

3. Congestion Control check

• Show your test result screenshot from basic_transfer_test
• Show your cwnd record plot, and describe your plot (when are slow start, congestion avoidance, packet loss... or any other congestion control algorithms you implemented)
• If you modified the testing scripts, state changes and reasons.

4. Concurrency check

• Show your test result screenshot from basic_concurrency_test
• Show your concurrency visualization plot (automatically generated after running this test)
• If you modified the testing scripts, state changes and reasons.

5. Crash check

• Show your test result screenshot from basic_crash_test
• Generate and show a concurrency plot after running this test by

python3 test/concurrency_visualizer.py log/peer1.log 

• If you modified the testing scripts, state changes and reasons.

6. Comprehensive check (No points, but strongly recommended)

• Show one of your test result screenshots from advance_test, your network topo, tasks, and running time.
• This check helps us to ensure your code is runnable, and we may refer to your self-reported running time when running hidden comprehensive tests.

7. Bonus check

• Discuss what you have done to optimize the performance of your code
• Show (if applicable). For example, if you implement fast recovery, you should show us a different cwnd plot. If you implement SR, show us your corresponding peer log.

Code inspection and Q&A

You will be asked some questions regarding your presentation and code, and you may be required to run one or more tests.

Question examples

1. Which lines of code are used to maintain concurrency? How do they work?
2. Follow our instructions to change the lost packet in basic_transfer_test, and run the test, show us the cwnd plot again.

Contribution Rate

You can report the contribution rates(joint with Ass2) of each member of your team. By default, we will split your score equally.

We do understand that some students deserve more points as they undertook more work in their teams, or worse yet, their teammates became free riders because of the pandemic and abandoned them. With careful consideration of workload, fairness, and humanitarianism, we proposed the following directions:

1. Those who work more will get more, and vice versa.
2. Everyone should get at least some points unless he/she is completely an unfriendly free rider [discussed in Exception].
3. Only those who submit runnable code (cannot fail all comprehensive tests) are likely to be bumped up to 100 marks. This means, sacrificing one of your teammates to jump from 70 raw marks to 100 marks will NOT be possible. This is because 70/100 points will be graded in the presentation, but we need to guarantee your code is at least runnable to grant you 100 points.

Take all of the above into account, your individual score will be calculated by

$$s_i=min(S(1+1.08(w_i-\frac{1}{N})), 120)$$

where $s_i$ is the $ith$ member's score, $S$ is your group raw points, $w_i$ is the $ith$ member's contribution rate, $N$ is the numebr of students in your team. Note that
$$\sum{w_i}=1$$

With this, students in a 3-student team(most common) with 1 teammate lost need to get at least 85 raw points to be bumped up to 100 points. We mainly consider the situation of losing 1 teammate in the 3-student team because 88% of our teams are 3-student, and around one-third of students chose P/F.

Exception

Zero points can be given to a specific member of your team if you all agree that he/she contributes nothing to your work, even not emotionally. Note that this option may severely hurt one's feelings.

Comprehensive Test

• We will not crash your peer (but we cannot control if your peer crash itself)
• Running time does not matter (as long as they can finish within 1 hour, though typically just several minutes)
• We will run your code many times on the same task, and you will get the points as long as you can pass once.
• Partial score allowed. For example, if you finish downloading 4 fragments correctly but 3 incompletely, you can get a partial score.

Bonus Test

• Only those who implement at least one bonus feature can participate
• Those who can finish the task with less time will rank higher
• Final score will be given based on your ranking
• We will run many times to get your average running time.