The New Economics of Embedded Systems

Abstract

There is a fundamental shift coming in the economic model used to build successful embedded systems. Below 100nm fabrication geometries, the cost of designing and creating masks for new semiconductor devices is projected to rocket up, making it uneconomical to create a new device for every system, or even to do another manufacturing pass to fix bugs.

So why is this an interesting topic at a workshop on software and compilers? Because the increased cost of hardware, and the increasing demand for new capability in embedded systems mean that programmable products will explode as the solution of choice to cost-effective embedded systems. However, the need to keep power dissipation under control means that both programmability and configurability will be critical in new system design. This puts an increased burden on compilers to generate code for configurable processors, and requires additional capability in software to manage the complexity.

This talk will briefly review the reasons for the economic change, but focus primarily on a vision for a new type of embedded system in which software and compilers play a critical role. In some sense, the original RISC concepts have returned in that software will assume an increasing role in the design of new embedded systems.