Through the 1990s and into the first decade of this millennium, semiconductor devices have evolved from a single processor plus memory and logic to highly complex electronic systems with more than 100 separate on-chip functions. These functions include multiple general-purpose and application-specific processors as well as a multitude of interfaces for system level connectivity, including USB, Firewire, Ethernet, PCI Express and WiFi.
To support these massively complex architectures, new bus structures such as the Open Core Protocol have been introduced, requiring a whole new level of analysis to ensure that performance estimated at the system level is actually achieved in the hardware implementation.
New audio and video processing standards, as well as communication standards like WiMax and DVB-H, have added to software complexity. Support for Linux, Windows and other proprietary operating systems has placed additional burdens on software designers. Greater collaboration between the hardware and software developers through ESL methods is now required to ensure that tradeoffs between software and hardware implementation are well understood. Only through this understanding can designers achieve their goals of delivering solutions with the lowest power, highest performance and lowest cost.
Yet many electronics companies in North America are still employing RTL-level methodologies for hardware design, delivering no real reduction in time-to-market and development costs compared to previous product generations.
Where does that leave U.S. electronics companies? As a 20-year semiconductor veteran who experienced the RTL revolution, the prospects are concerning.
Engineering teams, driven by extreme time-to-market demands and project complexity, shift from one project to the next with little time in between. Engineering budgets only accommodate "must have" new tools and methods that can demonstrate immediate payback. Little resource is allocated for education and adoption of new design methods with longer term benefits. With little prospect for truly understanding the dramatic advantages of ESL design tools and methods, engineers in the U.S. are falling behind their counterparts in Asia and Europe in the ESL revolution.
The longer this continues, the harder it will be to catch up.
What needs to change? Priorities and allocation of talent. Both are required. Indeed, U.S. engineering companies need to jump onto the ESL train now. Influential and well-respected design managers need to be assigned the sole task of driving ESL methods into their respective companies. Corporate leadership needs to provide adequate financial support, assemble hardware/software engineering teams with well-respected engineering pioneers who are capable of establishing priorities for ESL deployment. They need to work with ESL and EDA partners to develop a plan for implementing an ESL methodology across product development organizations. ESL implementation plans should ensure that immediate, short-term benefits of ESL adoption can be demonstrated and should establish a long-term implementation vision that will result in widespread adoption of ESL across hardware and software design teams.
From an EDA perspective, tool vendors need to collaborate more closely to transform a tool chain into a mainstream ESL methodology. ESL tool vendors have been working with global industry leaders to create a strong base of knowledge and that expertise can—and should—be shared. They need to provide training and methodology consultation to facilitate adoption in the United States.
Additionally, closer collaboration between tool vendors and academia to ensure engineering curricula prepare new engineers for the shift to ESL design methods in the U.S. is key.
Innovation in technology has been the cornerstone of the United States's advantage over the rest of the world. If we recognize the opportunity for an ESL methodology shift and make the necessary changes, our innovation leadership can and will continue.
Tom Sandoval (email@example.com) is Chief Executive Officer of Calypto Design Systems.