TDDC78 mid-term evaluation 2018The course has been mid-term-evaluated by the muddy card method on 23/4/2018 during the 9th lecture.
34 students attended the lecture, and I received 29 cards. I summarize in the following the main issues and comment where appropriate. Concerns mentioned on only one card are taken into consideration but usually not listed below.
Overall, the course seems to run well. 4 cards had no dislike at all, and most criticism is not of a serious kind.
Topic and Organization:
- The topic of the course is considered interesting and very relevant (5).
The intro and visit to NSC was appreciated (2).
- Working on a real supercomputer is appreciated (1).
- Good schedule (1)
- Labs start too late into the lecture period (2)
Comment: This was caused by Easter being not so early this year and the Easter break thus cutting into the lecture period after 2 weeks. When we had the same situation in some earlier years, starting the labs early resulted in loosing too many students during Easter vacation, so we think it is better to keep it compact. This year, indeed a higher percentage of students are active in the course 2 weeks after the Easter vacation period, so it appears to be the right way.
Disliked 4-hour lecture/lesson sequence (2)
Comment: Me too. But given the faculty's central block schedule and the chain of dependences between lecture topics and labs, there is basically no choice in the first 3 weeks of the course; at least we avoided late evening hours this year.
- Lectures are appreciated as interesting, informative, clear (6).
- Very educational examples (2). Right tempo (1).
- Broad scope of presented topics in parallel computing (ranging from various architectures via different programming models to theory) is appreciated (3)
- Slide material is considered of high quality (2) / informative (3) /
(too) dense (2) /
good for repetition after the lecture (3).
One card suggested to have 2 separate slide sets, one for presentation and one for repetition.
Comment: Indeed, the slide material partly also serves the purpose of a compendium text, which might reduce the need to buy a textbook. I have also plans to write a compendium text for some of the lectures, which would allow me to reduce slide text in future years.
- Passionate and competent lecturer (3).
- The live coding example was very appreciated (1).
- Slides need be updated (1) /
some code examples are in Fortran or old-style C. (2)
Comment: I try to migrate them all to C as time permits.
Lecturer could be less monotonic / more interactive / less powerpoints (4).
Lecturer could speak more often towards the audience (2).
- Labs are considered very interesting/useful/instructive (12) and fun (6). Right level of difficulty (2)
- Helpful lab assistant (1)
- Sometimes long waiting for help/demo in the labs, could use another assistant
and/or more supervised time (2)
Comment: However, some lab sessions were not well attended last week...
Doing all measurements takes much time. Need more scheduled
(but non-supervised) time on the NSC resource (3)
Comment: It is possible to run programs on Triolith even outside reservations.
- Good lab compendium (1)
- Lab compendium misinformative about what to do for labs 1+2 (1)
- Some source files in the lab code are perceived as unnecessary (1)
- Hard to get started with the labs (1)
- Both demonstrating (with explanation) and writing a lab report is kind of a duplicate work (2)
A few further suggestions regarding labs were brought up, such as:
- Lab instructions could contain some additional basics/guides to get started faster (1)
- Lab 1+2 setup: Threshold filter should come first, because it is simpler (1)
- Doing the same again in pthreads (after MPI) seems unnecessary (1)
- Example code for lab 2 should be (also) in C (1)
- Lab templates should be in C++ (1)
- Soft deadlines for the labs? (1)
Comment: We will discuss these suggestions for next year.
Course information, web page:
- Easily accessible and extensive information (2)
- Suggestion: could upload more [commented example] codes on the web page (1)
- Restricted-page structure with slide material
should match the lecture list page structure (1)
Comment: There is no one-to-one matching between slide sets and lecture slots/days, some span several lecture slots and some only a half. However, I have revised the restricted-page to make this matching more clear.
Thanks for all comments and suggestions!
Christoph Kessler, course leader TDDC78
Page responsible: Webmaster
Last updated: 2018-04-27