|
Guest Opinion
November 2000
Bridging the gaps
To achieve the promise of the embedded Internet, designers will need to exploit embedded software design automation.
Shaul Gal-Oz, Aisys
Embodied in a flood of smart products, ranging from personal information appliances to connected home products, the embedded Internet promises to seep into every facet of life. Unlike the familiar WinTel platform that defines the de facto standard in the PC sphere, this new environment will comprise diverse hardware platforms and software environments—creating an intensely interconnected, yet inevitably heterogeneous application environment. For companies looking to ride the embedded Internet to market leadership, however, the diversity of this new environment is at odds with growing time-to-market pressures—exposing some serious gaps in the road ahead.
Facing unparalleled demand for increasingly sophisticated products, product developers are already exploiting a new generation of high-integration processors. For the embedded Internet, designers are likely to consider a variety of processor architectures such as ARM, MIPS, and PowerPC, selecting the embedded processor that delivers the optimal mix of price, performance and integration of on-chip peripherals required for a particular design problem. Furthermore, SOC (system on chip) ICs carry this integration to the next level by combining these same processors with a broad array of peripherals and other intellectual property.
At first glance, this proliferation of diverse hardware seems of little concern, especially when software standards such as WAP (wireless application protocol) and XML (extensible markup language) seem to be settling in to stabilize key areas of interoperability. Yet a huge gap in function lies between the growing foundation of diverse high-integration embedded hardware and interconnected application-level software. Unless specialized low-level software code is available to connect the underlying hardware with application and RTOS (real-time operating system) software, there simply is no "system" in an embedded-system design.
Despite the importance of this layer, embedded developers continue using traditional techniques, which typically impose a cost—six months to a year to design, build, test, and debug hardware/software interfaces. This leads to further delays in downstream activities, including application development and RTOS integration. How much time will developers need as they move to the emerging generation of high-integration processors and SOCs? And how much time can manufacturers afford when they stand to lose not just market share but even entire windows of opportunity? Clearly, the answers to these two questions are currently badly out of step.
To achieve the promise of the embedded Internet, designers will need to exploit embedded software design-automation techniques, particularly as corporate expectations for higher productivity grow in anticipation of increasingly severe time-to-market requirements. Embedded design-automation tools that are already available can dramatically boost software-developer productivity for tasks such as hardware/software interface code generation. Most important, design-automation tools and methods let design teams focus more effectively on adding value at the application level, where it counts the most in the marketplace.
Author information
Shaul Gal-Oz is CEO of Aisys.
|
