Design Automation Is More Than EDA
Gabe Moretti, Senior Editor
In a couple of months the EDA industry will hold its yearly flagship conference: the Design Automation Conference (DAC). And just a few days ago Vic Kulkarni, SVP & GM, RTL Power Business at ANSYS-Apache Business Unit, told me that the industry should be called Design Automation Industry, and not EDA. Actually both the focus of the EDA industry and the contents of DAC are mostly IC design. This activity is now only a portion of system design and thus the conference does not live up to its name and the industry does not support system design in its entirety.
For almost all its entire life the EDA industry has been focused only on hardware and it did it well. That was certainly enough when products were designed with the approach that electronic systems ‘s purpose was to execute application software with the help of an operating system and associated device drivers.
A few years ago the EDA industry realized that the role of hardware and software had changed. Software was used to substitute hardware to implement system tasks and to “personalize” systems that were not end-user programmable. It than became necessary to verify these systems and new markets, those of virtual prototyping and software/hardware verification, grew and continue to grow. Yet the name remains Electronic Design Automation and most of its popular descriptions, like in Wikipedia, deal only with hardware design and development.
The Internet of Things (IoT), where intelligent electronic products are interconnected to form a distributed and powerful data acquisition, analysis, and control system, is considered to be a major growth area for the electronics industry and thus for EDA. A small number of companies, like Ansys and Mathworks for example, have realized some time ago that a system is much more than “just” electronics and software and these companies now have an advantage in the IoT market segment.
By just developing and verifying the electronic portion of a product traditional EDA companies run the risk to fool designers into thinking that the product is as efficient and secure as it can be. In fact, even virtual prototyping tools cannot make an absolute statement regarding the robustness of the software portion of the design. All that can be said using this tools is that the interface between hardware and software has been sufficiently verified and that the software appears to work correctly when used as intended. But a number of organizations and individuals have pointed to critical security issues in existing real time systems that are the precursors of IoT. The latest attention to security in automotive applications is an example.
The use of MEMS in IoT applications should push EDA leaders to ask why MCAD is still considered a discipline separate from EDA. The interface and concurrent execution of the electro-mechanical system is as vital as the EDA portion. Finally the EDA industry has another weakness when it comes to providing total support for IoT products. Nature is analog, yet the analog portion of the EDA industry lags significantly from the development achieved in the digital portion. Digital tools only offer approximations that have been, and still are, good enough for simpler systems. Leading edge processes are clearly showing that such approximation yields results that are no longer acceptable, and product miniaturization has resulted in a new set of issues that are almost entirely in the analog domain.
The good news is that the industry has always managed to catch up in response to customers’ demand. The hope offered by IoT growth is not simply revenue growth due to new licenses, but a maturing of the industry to finally provide support for true system design and verification.