Lecture 23

Learning objectives

After this class, you should be able to:

1. Use the CUDA profiler to analyze your code.
2. Use the CUDA Occupancy Calculator to choose parameters that optimize use of hardware resources.
3. Figure out performance bottlenecks using PTX code.

Reading assignment

1. Slides on Performance Analysis.

Exercises and review questions

• Exercises and review questions on current lecture's material
  1. Use the tools discussed in class to compare your assignment 4 codes.

• Preparation for the next lecture
  1. None.

---

Last modified: 4 Apr 2013