Hide menu

TDDC78 mid-term evaluation 2023

The course has been mid-term-evaluated by the muddy card method on 19/4/2023 during the 10th lecture.
22 students attended the lecture, and 21 cards were received. I summarize in the following the general outcome and main issues, and comment where appropriate.

Overall, the course seems to run quite well. Nevertheless, some of the cards pointed out some issues, which are discussed below.

Topic and Organization:
Several cards state that the topic of the course is considered interesting and very relevant. The intro and visit to NSC was clearly appreciated. Working on a real supercomputer is appreciated on many cards. It is also explicitly appreciated how the different topics in the course build upon each other.

The compendium is very appreciated.
Some cards ask for more information about Pthreads, or maybe a separate lecture. I understand that this is a consequence of a heterogeneous student group attending the course, where most - but not everyone - have already taken a course on concurrent programming and operating systems before, where Pthreads is taken up. Instead, recommended Pthreads tutorials are linked from the course web page and included in the recommended course literature for self-studies where required. In addition, we might consider adding some pthreads slides in next year's lab introduction lesson.

Several cards ask for ready solutions of the demo exams. For pedagogical reasons, I generally do not do that. (When I once did it in another course, I later got too many verbatim identical answers in exams even where the demo questions were not asked at all. This is not the learning effect that I would like to achieve.)
Instead, we will, at the end of the lecture period, have a separate lesson where we will go through some old exam questions; we will give the ideas and instructions for solving, but no complete solutions. The final step is still yours.
Please also have a look into the compendium, where each chapter has a section with exercises.

Several cards complained that there was an unplanned Sigma maintenance during a scheduled lab session. This is unfortunate, but not always avoidable even with a reservation (e.g. for urgent security patches) and not under our control. Fortunately, such events are very rare.

Lectures are appreciated as interesting, well presented, clear and detailed. Code examples and figures are generally appreciated, a few cards ask for even more code examples.
A few cards find the slides too dense / too much text and recommend various ways of changing this, which I will consider for next year.
One card criticizes that lectures sometimes go over time. Note that the wall clock in A36 is 2 minutes early - I actually try to finish on time. In rare cases it might happen that I forget the time over some complicated explanation; if that should happen, please let me know.

The labs are appreciated on many cards (well designed, fun, right level of difficulty).

The lab documentation is also appreciated on some cards, while a few cards ask for even more lab instructions e.g. on how to compile and execute on Sigma. Please note that we do have detailed instructions for that, given both in the guest lecture, in the introductory lab lesson, and in the lab instruction sheet posted on the course web page. Not all of this is also printed in the lab compendium, but overall there should be comprehensive documentation available. If you still find any concrete unclear points in the available documentation, please let us know.
Somewhat related, several cards find it hard to read the instructions and get started on the labs. Please note that this is an advanced-level course: It is normal that it takes some effort and time to fully understand the problem and get familiar with the lab skeleton before the first line of own code can be written / added. Take the necessary time.
One card mentions that it was unclear that both filters have to be written in both Pthreads and MPI. Sorry, but just that was explicitly explained in the introductory lab lesson, so it should have been really clear from the beginning.
One card writes that the soft lab deadlines are too strict (w.r.t. access to the lab assistant for demonstrations), while another reports on proceeding at a faster pace. The soft deadlines have been chosen based on previous years' experience with average student progress; some will need more time and some less. Please note that only the dates of the last lab session are hard deadlines for the entire lab series.
One card mentions that a single lab demonstration consumed 40min of the lab assistant's time, leading to long waiting times for the others. Such long demonstrations are, of course, not sustainable. From now on, we will enforce a strict time limit of at most 15 minutes for lab demonstrations. Demonstrations that are not accepted after 15 minutes, for whatever reason, will be preempted and moved to the end of the queue.
The few further comments and suggestions about the labs have been discussed with the lab assistant; we will take these into consideration when planning for next year's instance of the course.

Thanks for all comments and suggestions!

Christoph Kessler, course leader TDDC78

Page responsible: Webmaster
Last updated: 2023-04-24