Lecture 21

Learning objectives

After this class, you should be able to:

1. Use SSE instructions to optimize the performance of code on Intel x86 processors. In particular, optimize memory performance using cache pre-fetch and streaming stores.

Reading assignment

1. Refer to the Intel compiler intrinsics reference mentioned in the Lecture 21 slides.

Exercises and review questions

• Exercises and review questions on current lecture's material

  1. Use cache pre-fetches in a linked list traversal application, and compare the performance with code without the pre-fetch. Assume that you need to perform some computation on each item of the linked list. How does the performance difference vary with the time taken in the computation step?

  2. Use SSE caching intrinsics for searching a binary search tree, and compare the performance with code without these intrinsics.

• Preparation for the next lecture
  1. Prepare for the midterm.

---

Last modified: 12 Apr 2010