| Class | Topic, lecture notes |
Reading list | Assignments/Milestones |
14.02.11 | Introduction: Multi-core processor technology,
concurrent computational patterns (slides: 1up,
2up, 4up) |
A View of the Parallel Computing Landscape,Communications of the ACM, 52(10):56–67, October
2009. | Form project groups |
| 16.02.11 | Roofline model (slides: 1up,
2up, 4up) |
Roofline: An Insightful Visual Performance Model for Multicore Architectures |
Fill out and submit account forms
Read paper for discussion on Monday 21.02: Stencil Computation Optimization
and Autotuning on State-of-the-Art Multicore Architectures |
| 21.02.11 | Paper discussion | Stencil Computation Optimization
and Autotuning on State-of-the-Art Multicore Architectures | First Class Project |
| 23.02.11 | POSIX threads (slides: 1up,
2up, 4up) | Pthreads Programming: A POSIX Standard
for Better Multiprocessing, Bradford Nichols,
Dick Buttlar, Jacqueline Proulx Farrell, O'Reilly &
Associates, | Familiarize with project programming models (OpenMP tutorial, OpenMP specs, Cilk, SMPSs overview,
SMPSs programming model). Read specs and simple examples
Read paper for discussion on Monday 28.02: Threads Cannot be Implemented
as a Library |
| 28.02.11 | Paper discussion | Threads Cannot be Implemented
as a Library | |
| 02.03.11 | Cilk (slides: 1up, 2up, 4up) | Cilk 5.4.6 reference manual, | First project should be running on Cilk by end
of the week. Paper assignment for next week: Lazy
Binary-Splitting: A Run-Time Adaptive Work-Stealing Scheduler |
| 11.03.11 | Cilk | Lecture notes of 02.03.11 | First project should be running on
Cilk and OpenMP by now |
| 14.03.11 | Paper discussion | Lazy
Binary-Splitting: A Run-Time Adaptive Work-Stealing Scheduler | |
| 16.03.11 | OpenMP (slides: 1up, 2up, 4up) | OpenMP ARB Specifications, | First project
should be running on all three programming models. Paper assignment for next week: The Design of OpenMP Tasks |
| 18.03.11 | OpenMP | Lecture notes of 16.03 | |
| 23.03.11 | Race Detection (slides) | LockSmith: A tool for finding races in C programs | Second class project |
| 28.03.11 | Paper discussion | The Design of OpenMP Tasks | Study MapReduce ReverseIndex benchmark. Run benchmark on an SMP |
| 30.03.11 | Task-Based Dataflow Programming Models (slides) | SMP SuperScalar | Unoptimized ReverseIndex should be running
in OpenMP, Cilk, SMPSs |
| 11.04.11 | Programming Multiprocessors with Explicitly Managed Memory Hierarchies – Cell (slides: 1up,
2up, 4up) |
Cell Broadband Engine Processor:
Design and Implementation, | Paper assignment for next week:
Thread criticality predictors for dynamic performance, power, and resource management in chip multiprocessors
|
| 13.04.11 | Paper discussion |
Thread criticality predictors for dynamic performance, power, and resource management in chip multiprocessors
| |
| 04.05.11 | Parallelizing with compiler assistance | (slides) | |
| 06.05.11 | | | Third class project |
| 11.05.11 | General-Purpose Graphics Processing Units, CUDA | (slides: 1up,
2up, 4up) | Design of parallel
web server should be ready |
| 13.05.11 | General-Purpose Graphics Processing Units, CUDA | slides of previous lecture |
Paper assignment for 20.05: An asymmetric distributed
shared memory model for heterogeneous parallel systems |
| 16.05.11 | Optimizing CUDA code | (slides: 1up,
2up, 4up) | |
| 20.05.11 | Paper discussion | An asymmetric distributed
shared memory model for heterogeneous parallel systems | |
