Parallel computing
Simo Juvaste, University of Eastern Finland, Computer Science
(Rinnakkaislaskenta, 5 op)
Address: http://cs.joensuu.fi/pages/sjuva/parallel.html
Page contents
Official course description
Parallel computing (5 cr) 3621528
- laudatur -level elective course
- Introduction to parallel computing and parallel algorithms.
Models of parallel computing, parallel architectures, design principles
of parallel algorithms. Implementability of parallel algorithms in
different alrchitectures.
Literature: Jaja: Introduction to
Parallel Algorithms. Almasi, Gottlieb: Highly Parallel Computing, 2nd.
Prerequisites: Programming language C, Computer Systems, Data Structures and Algorithms.
Tentative contents
Parallel (and concurrent) computation.
Parallel computation is about achieving more computing power for a single task.
Parallel computing is becoming mainstream again at 2010!
Subjects
- An introduction to parallel computing
- What, Why, How
- Examples
- Benefits, drawbacks
- Parallel computers
- Difficulties, Design issues
- Models
- Theoretical models
- Standards
- Accepting physical and financial limitations
- Parallel computer architectures
- Parallel algorithms
- Simple algorithms using PRAM-model
- Sorting, etc.
- Parallel programming
- MPI, HPF, OpenMP, Shared memory (SMP), GPU computing (CUDA/OpenCL), ...
- Other topics
- Concurent programming (depending on time and interests)
- pthreads
- processes
- interprocess (&thread) syncronization & communication
- Distributed computing
Schedule at Fall 2010 (2nd period)
Lectures 30 h
- 27.10.2010 -- 15.12.2010
Wed 08-10, Thu 08-10 Joensuu: T/2D106, Kuopio: MT3
Exercises 14 h
- 1.11.2010 -- 13.12.2010
- Group 1: Combined to group 2.
- Group 2: Mon 14-16 T/B179 (Joensuu) Except 6.12. -> Tue 7.12. 12-14.
- Group 3: Mon 12-14 MT4 (Kuopio, Niina Päivinen) Except 6.12. -> Tue 7.12. 12-14.
Registering to the course
- Please register to the course through WebOodi.
- "Separate" courses at Joensuu and Kuopio because of Lukari. In reality, the same course.
- Register for the course and one exercise group.
Exam
- Thu 16.12. 08:00-10:00 Joensuu: T/2D106, Kuopio: MT3
- Detailed results (only uef.fi) Grades also at web-oodi.
Course evaluation form
Opintojakson arviointilomake
General exams
- X-exercises have to be made&passed before participating.
- Contact me to enroll.
Exercises
- Traditional
- Make beforehand, mark as done at the beginning, present when requested.
- It is not possible to get marks without participating the class.
- Part are traditional paper-whiteboard exercises, part has to be done using computer, compiled, will be presented using data
projector and ssh to cs.
- At least 1/3 has to be done.
- Obligatory X-exercises
- 4-5 (in total) on key subjects
- Will be graded (coarsely)
- Missing (not submitted) will be graded negatively
- Has to be submitted by email on previous day, lateness will reduce a point.
- A unique solution for every student (co-operation is allowed, but the same
answer from two or more students is not)
- Automated processing and comparison, strict submitting rules
- Not counted as traditional exercises when calculating the 1/3 rule.
Project work
Within the course the will be some project group works as an optinal extension (1-3cp)
to the course.
Grading
- X-exercises: ~40%
- Exams: ~60%
- Extra points for making more than 1/3 of standard exercises (up to 10% of maximum).
((n-N/3)/(2N/3))*(max/10) N is # of std ex, n is # of your marks).
- No roundings.
Workload
For an average student, for average grade.
|
7 |
8 |
|
Weekly |
Total |
| Lectures |
4 |
30 |
| Prepare |
2 |
16 |
|
|
|
| Exercises |
2 |
14 |
| Prepare |
9 |
63 |
|
|
|
| Exams |
|
2 |
| Prepare |
|
8 |
|
|
|
| Total(h) |
18.25 |
133 |
| Total(ECTS) |
|
4.99 |
| 26.67 |
|
|
Literature
- Lecture hand-outs (partial)
- 5 euros for the whole course, please have exact sum handy at the first lecture
- Part 1: Introduction, PRAM, algorithms, 27.10., pages 1-123.
- Part 2: Numerical problems, Real world, etc., 16.11., pages 132-221.
- Part 3: OpenMP, MPI, Fortran90, 24.11.
- Part 4:
- JaJa: An Introduction to Parallel Algorithms
- Akl: The Design and Analysis of Parallel Algorithms
- Penttonen Martti: Johdatus Algoritmien Suunnitteluun ja Analysointiin, osia
luvuista 17-25.
- Other
WWW-links
Tools
Exercises
Examples, templates for exercises, etc.
Here
Last modified
Wed Dec 22 10:08:03 EET 2010
SJ
