DD2451, Parallel and Distributed Computing, Pardis, 2011
Handin schedule
Handin 1: Out 4 Nov 16.00, return 11 Nov 16.00
Handin 2: Out 18 Nov 16.00, return 25 Nov 16.00
Handin 3: Out 2 Dec 16.00, return 9 Dec, 16.00
Exercises and Hand-in Problems
Lecture 1
- H&S ex 4, 5, 11, 13, 18, 19
Lecture 2
- H&S ex 24, 25, 26, 28, 31
Lecture 3
- H&S ex 35, 36, 40, 41, 43. Splitter exercises: Check lecture 1
Lecture 4
- H&S ex 47, 50 (so far)
- Also: H&S ex 53, 54, 68
Hand in 1
- Get it here (after 16.00)
Lecture 5
- H&S ex 85, 86, 91, 222 (in appendix B), 224 (also in appendix B)
Lecture 6
- H&S ex 102, 104 (can such a scenario arise in the lazy algorithm as wel1? In the lock-free algorithm? Why not?), 109, 112, 118
Lecture 7
- Exercises are in the slides
Hand in 2
- Get it here (a bit before 16.00 actually)
- In exercise 2 you may assume a) crash failures only and b) broadcasts are always completed. I.e. if A broadcasts m to B and C and neither B nor C fail then both B and C receive m. That is, a crash cannot cause A to send to B only and not to C (or vice versa).
Lecture 8
Lecture 9
Lecture 10
- Exercises in slides 49-51
Hand in 3
- Get it here (after 16.00)
- Hint for question 3: First part of the question may be a little cryptic. View the entwork graph as an ordinary graph with vertices (nodes) and edges (links). When running the algorithm the claim is that the labelling as determined by variable b eventually stabilizes in such a way that the graph with the labelling has a specific property. This could be, for instance: The labelling determines a Hamiltonian circuit, or the labelling determines a connected component (none of these apply, btw). Hope this helps.
|
