Lecture 15

Learning objectives

After this class, you should be able to:

1. Define the IEEE representation of floating point numbers.
2. Given an arithmetic operation, give the results in different rounding modes.
3. Explain why results on GPUs could differ from those on the CPU.
4. Use fast math functions to improve performance.

Reading assignment

1. UIUC Lecture 15.
2. Chapter 7 of text, sections 7.1 - 7.5.

Exercises and review questions

• Exercises and review questions on current lecture's material
  1. Compare the performance of the trignometric functions with and without the -use_fast_math option.

• Preparation for the next lecture
  1. None.

---

Last modified: 26 Feb 2013