Today was the last lecture in “Multicore Programming,” a course I teach for the students in our club. I’d like to thank the guys who, despite the fact that the course isn’t meant for first-year students, and despite the late hour of the classes, still kept coming right up to the end. Many teachers could envy me for the students’ having such strong internal motivation.
Over these 22 lectures and practical sessions we managed to cover quite a lot: we programmed with the OpenMP, WinAPI, MPI, and Intel Threading Building Blocks libraries, tried out Intel VTune Performance Analyzer, Thread Checker, and Thread Profiler, and even had enough time to get sidetracked and try optimizing sequential programs for the specifics of the Intel Core2 Duo microarchitecture.
I’d like to get feedback from you: what did you like about the course, and what didn’t you like? What can be improved, and what can be made worse? In particular, I have these questions:
-
should we study WinAPI, or should we start with OpenMP and focus more on Intel TBB?
-
should more attention be paid to how the microarchitecture is structured?
-
what was hardest to understand, and during which topics were you “yawning”?
-
what can be covered in more detail, and what can be thrown out?
-
what examples should be covered in the labs?
-
what should be done to make the course more interesting, more fun, and more useful?
You can answer in the comments, write to volodymyr.obrizan@gmail.com, or talk to me in person. Any help would be appreciated. Thanks!
Those who want to attend this course should wait until the fall. Most likely, I’ll teach it once more as part of our club sessions.