System Level and IP-based Design

Abstract

This two part lecture covers several fundamental issues in the design of embedded real-time systems. Embedded real-time systems are ubiquitous in today's world, their numbers tend to increase exponentially. Yet the design methodology used is still based on older tools and practices designed with very different goals than those for embedded systems. The most important characteristic of these systems is the massive use of programmable components to achieve the design goals. Hence their design requires the use and optimization of both hardware and software.

In this lecture, we begin by outlining the revolution that is taking place in the electronic industry due to the advent of deep submicron and the continuing pressure on time-to-market. We present the notion of System Level Design, and IP-based design, and underline the challenges we will have to face. Then we focus on a top-down, constraint-driven design methodology that emphasizes the analysis and optimization of the top part of the design where all the important algorithmic and architectural decisions are taken, followed by design exploration, evaluation, iteration, and then final implementation.

The framework for functional design, architecture selection, hardware-software co-design, software optimization, and real-time operating system design will be POLIS. We will use this environment developed at UC Berkeley to illustrate some of the key points of the design methodology.

The lecture is aimed at two basic goals:

1. providing system designers with a practical perspective on new methodologies for hardware/software co-design of embedded controllers;

2. exposing CAD engineers to the problems to be overcome in system design and presenting new algorithms and approaches to the solution of these problems.

This talk constitutes the first part of the lecture.