CS 696: Group Study on Advanced Parallel Processing

Instructor: Sanjay Rajopadhye (email: svr@cs.colostate.edu)

Class meets: Th 15:00 - 17:00, USC 310C

Please note room & time change above. The schedule has also been modified: we have pushed things forward by one week. On Tu Feb 17, we will continue with Valinat's paper and further discussion of BSP. We also have an interesting paper tht Lakshmi just dug up.

In this course, we will work off a bibliography of research papers. We will use the following guidelines. Prerequisites are advanced graduate standing (instructor consent) and CS575 or its equivalent.

Each student will be responsible, in turn, for leading the discussion of the paper of the day.
All students are expected to have read the paper BEFORE each class, and will submit a 1 page summary plus a list of three questions before we start.
The discussion/presentation will last about 45 mins, the rest of the time will be devoted to a detailed critique of the paper, questions about specific points in the paper, and an analysis of its contributions, etc.
The student leading the discussion will submit by the following meeting, a 5-page critique of the paper, based on the discussion in class.
Students should register for 2 credits S/U grading per paper that they agree to present.

In case you are interested but not necessarily in coming regularly and/or do not desire to register, you are welcome if you have the appropriate background. Some papers below have not been assigned (and the list can always be extended), and there are some open slots in the semester. It would also be nice to get to some of the papers that Arnie Rosenberg discussed in his ISTeC lecture(s).

Tentative Schedule

Cls #	Date	Speaker	Reading
1	Feb 10	Keith Bush	[Valiant, CACM 1990]
2	Feb 17	All (Keith+Lakshmi)	Continue BSP [Valiant, CACM 1990] and [Gorlach, TOPLAS 2004]
2	Feb 24	Rama	[Culler et al, PPoPP 1993]
3	March 2	Tarun Banka	[Blumoffe & Leiserson, JACM 1999]
4	March 9	All	[Blumoffe & Leiserson, JACM 1999] contd.
5	March 16	Spring Break
5	March 23	Cancelled
6	March 30	Cancelled
7	April 6	Cancelled
8	April 13	Jason	[Blelloch, Gibbons & Matias, JACM 1999]
9	April 22	Sanjay	[Wolf & Lam, IEEE TPDS 1991]
10	April 29	Nate	[Lim & Lam, Parallel Computing, 1998]
10	May 6	Lakshmi	[Robert et al]

References

[Blelloch, Gibbons & Matias, JACM 1999] Provably efficient scheduling for languages with fine-grained parallelism (preliminary version presented at SPAA 1995).
[Blumoffe & Leiserson, JACM 1999] Scheduling multithreaded computations by work stealing (preliminary version presented at FOCS 1994?)
[Culler et al, PPoPP 1993] LogP: towards a realistic model of parallel computation San Diego CA (pp 1-12).
[Culler et al, CACM 1996] LogP: a practical model of parallel computation (Vol 39, No 11, November 1996, pp 78-85).
[Gorlach, TOPLAS 2004] Send-receive Considered Harmful: Myths and Realities of Message Passing (Vol 26, No 1 Jan 2004, pp 47-56).
[Greibl, Feautrier & Lengauer, IJPP 2000] Index Set Splitting (28(6): 607-631; Dec 2000)
[Hogstedt et al, TPDS 2003] On the parallel execution time of tiled loops (Vol 14, No. 3, March 2003, pp307-321)
[Juurlink & Wijshoff, ACM TOCS 1998] A quantitative comparison of parallel computation models
[Lee & Kedem, TOPLAS 2002] Automatic data and computation decomposition on distributed memory parallel computers
[Lim & Lam, Parallel Computing, 1998] Maximizing Parallelism and Minimizing Synchronization with Affine Partitions (Vol 24, No 3-4, May 1998). An unofficial copy is available here.
[Schreiber & Dongarra] Tiling Fundamentals
Some Paper(s) on Foundations of Tiling.
[Valiant, CACM 1990] A Bridging Model for Parallel Computation
[Wolf & Lam, IEEE TPDS 1991] A Loop Transformation Theory and an Algorithm to Maximize Parallelism