The videos were very well done. The exercises were a bit too easy in my opinion. But the exercises did focus on the issue at hand, unlike some other courses that get you stuck on complicated but unrelated issues used only as an example problem.

T​he course is interesting and well explained, but Coursera Grader is a pain: fails in nearly all miniprojects. The last one don't work at all.

Provides background material for the course on concurrency in the same specialization. Assignments are trivial.

I would prefer see examples based on standard Java code, not in a library developed for some small group.

A very good introductory course only rating 3 star because the projects did not feel greatly involving.

good coverage of basics.. would have been great if there were few more coding practice papers.

Heavy theory but poor practical. Need more enhancement for the next time.

Focusing more on theoretical parts

Good overview, but less details

The covered material is useful, but the presentation is... well, not so nice. The professor writes with a pen on a glass board, and his handwriting is a catastrophe (it's not about individual letters, but the way he clutters up space in a completely unorganized way), and on top of that it slows down the delivery. So I ended up just reading the summaries and watching the demonstration videos and only referring to the videos when needed.

The mini projects are way too easy in my opinion, though useful.

But my main criticism is that the course's title is "Parallel Programming in Java." However, the course doesn't teach you the Java API, which it definitely should. It recommends using RecursiveTask from the Fork/Join framework to implement a future, which is kinda stupid. It certainly works, but if you do it in real life, it immediately exposes you as ignorant of Java's API. No mentioning of Executors whatsoever. Doesn't teach you how you can distinguish the number of logical and physical cores, and indeed the provided unit tests are all such that you can get 100% from the Coursera grader, which uses 4 tasks, but will let you fail if you run it on an octacore AMD Ryzen 7 with 16 threads, because it unrealistically (for CPU-bound tasks) expects almost 16-fold speedup from your parallelized solutions. In the end, you're left wondering if it is better to construct a ForkJoinPool yourself or to use the commonPool(), or whether you should call submitAll() or fork(), or whether you should construct your Phaser with 0, 1 or the number of tasks, and what the difference between awaitAdvance and awaitPhase might be, instead of leading by example and showing you a clean use of the API. On top of that, it introduces you to Rice's HPDP (or something like that) library... That's cute (and to be fair, the libary seems quite useful and elegant), but it's not really what people come here for.

I can say that I learned something from this course, but I would not have wanted to pay for the course if it wasn't included in Coursera Plus.

Good part is that I have learnt quite a lot of things from this course and I enjoyed it. Not so good part is that I don't get answers to my questions I post on the forums, graders are not working correctly and I don't know the correct answers to the quiz (once I passed it, it would be nice to know the answers to the ones I answered incorrectly). I feel cheated because I paid for the course, but it makes me feel like I wasted my money a bit.

I spent 6 days on this course, and I feel it is not worth the time. One of the problems is that the lectures are overly basic, and the mini projects are not challenging, at least not in a helpful way (quizzes are sometimes challenging due to some concepts not covered in the lectures at all). I guess the instructor did not really spent much time preparing the materials and assignments.

T​he lectures and the projects often did not match up. The autograder was also very capricious and you can pass and fail with the same submission.

Please make the homework better.