C++ AMP

Chapter 4. Tiling

In this chapter:

Purpose and Benefit of Tiling

tile_static Memory

tiled_extent

tiled_index<N1, N2, N3>

Modifying a Simple Algorithm into a Tiled One

Effects of Tile Size

Choosing Tile Size

Summary

The time comparisons of the case study in Chapter 2, showed that C++ AMP can dramatically outperform single-threaded simple calculations on the CPU. If your only two choices when designing your application were simple C++ AMP and single-CPU, then the choice would be simple: you’d use C++ AMP. However, you have a number of other choices that can make your CPU solutions faster. One of these is clever caching of intermediate results to reduce the number of times each calculation must be performed. Another, done automatically for you on the CPU, ...