Motivation: an example

• We will begin by looking at matrix multiplication.
• We will study the performance of a $512 \times 512$ double-precision floating point matrix multiply.
• We will investigate the performance of various versions in order to determine what transformations a compiler should be applying.