Table of Contents

Advanced programming models & paradigms for multicore and distributed architecture (2012)


Asanovic et al. “A View of Parallel Computing Landscape” Communications of the ACM Volume 52 Issue 10, October 2009 Pages 56-67


Aim

The course aims at introducing and discussing different research themes generally related to the design and implementation of structured parallel programming frameworks for parallel and distributed architectures. The discussed themes will include:

Program

  1. Introduction (2h)
    • Advances in hardware and state of the art programming models.
    • Parallel/Distributed dicotomy revisited.
  2. Structured parallel programming (10h)
    • Algorithmic skeletons
    • Parallel design patterns
    • Components
  3. Hardware targeting (6h)
    • COW/NOW
    • Multicore
    • Co-processors
  4. Trends & perspectives (2h)

Exam

The exam will consist in

Exams given so far

Who When What Links
D. Buono & S. Lametti Oct. 2nd, 2012 Porting FastFlow on Tilera slides

Lessons

Mon 21 Tue 22 Wed 23 Thu 24 Fri 25
Sem Ovest 11-13 11-13 11-13 9-11 11-13
Mon 28 Tue 29 Wed 30 Thu 31 Fri 1
Sem Ovest 11-13 11-13 11-13 11-13 11-13

Support material

A large part of the arguments of the course is covered by the “SPM notes” (M. Danelutto, 2011) that are available on the WEB. These notes present arguments in a master degree course perspective, but host all the hints and references used to build the more advanced PhD course arguments. References to selected papers and other available material will be provided during the course.

Contents of the course

MindMap of the course contents (complete)

Lesson notes

Day Hours Arguments File(s)
Mon 21 2 Introduction. Motivations. Hw and sw progress. Blackboard dump
Tue 22 2 Parallel application design. Performance measures. Algorithmic skeletons: introduction, functional and parallel semantics. Blackboard dump
Wed 23 2 Skeleton classes. State. Template based and macro data flow based implementation of skeleton frameworks. Blackboard dump
Thu 24 2 Skeleton rewriting rules. Backus FP in skeleton perspective. Behavioural skeletons and non functional concern management. Blackboard dump
Fri 25 2 Hierarchical and multi-concern management in BS. Sample skeleton framework: SkePu. Blackboard dump
Mon 28 2 Sample skeleton framework: FastFlow. Skeleton and implementation template perspective. Blackboard dump
Tue 29 2 Parallel design patterns. Blackboard dump
Wed 30 2 Patterns in perspective: Berkley report, Exascale roadmap and Intel TBB Design patterns Blackboard dump
Thu 31 2 Industrial perspective: Google, mapreduce and FlumeJava. Component based frameworks (intro). Support techniques in COW/NOW. Blackboard dump ff_misdcorrect code
Fri 1 2 Support technique: multi/many core (M2C) and accerators Blackboard dump
End of the lessons

Links to material