Lecture 21
Learning objectives
After this class, you should be able to:
- Given an application, identify its parallelism structure.
- Use loop transforms to enable further optimization of the code.
- Develop strategies to reduce the data movement overhead.
Reading assignment
- UIUC Lecture 22.
- Chapter 11 of the text book.
Exercises and review questions
- Exercises and review questions on current lecture's material
- Replace use of constant memory in the MRI code with the use of shared memory and evaluate its performance.
- Preparation for the next lecture
- None.
Last modified: 28 Mar 2013