Applications of Computing in Industry : Lecture

Processors won't become significantly faster: the clockrate can't be increased much further and the time it takes to execute individual operations also can't be tweaked much more. However, the processing speed can be increased by having more, although probably simpler, processors. Also, multiple pieces of data can be processed with each operation. The Cell processor as used for example in a Playstation 3 provides significant processing power by having multiple simple SIMD (single instruction, multiple data) enabled cores controlled by a more complex core. The individual processors don't immediately share memory although data can be transferred efficiently between the cores.

This presentation provides an overview of the involved technologies, the associated programming model, and shows how to program for the Cell processor in C++ using a simple example. Although the Cell processor is currently still relatively exotic it is expected that similar approaches will be used by future processors and thus similar programming techniques will become more common, too.

In particular, this presentation will:

• give a brief overview of the overall architecture of the Cell processor
• describe the tools used to program for the Cell processor
• show how processes are distributed to the various cores
• show how data is transferred between cores
• give an introduction on how to take advantage of SIMD in C++
• provide an outline how the technical details can be hidden in C++