Traditional EDA companies have approached the ESL market by presuming that RTL engineers will embrace higher levels of abstraction in design under the weight of complexity. On the other hand, two companies that have served system designers for some time, National Instruments and The Mathworks, have taken a different approach that is proving quite successful.
The problems with the traditional EDA approach are twofold: first RTL engineers are not system designers, and secondly a system is composed by more than just electronic parts.
The result has been a much slower growth of the market than expected. Some companies have tried a variation on the theme, targeting hardware/software co-development as a market. Strong of the argument that if hardware and software designers use the same language both design and verification would be more efficient, mostly startups have pushed SystemC as the vehicle for system design.
These companies have found some success, especially in Japan and Germany, but for the wrong reasons. The financial results are less than expected, because a significant portion of the development can be done using cheap and readily available C development environments, and using EDA licensed tools only when absolutely necessary. In the C environment it is also very easy to develop testbenches using only freely available software, another detriment to EDA revenues.
National Instruments (NI), which entered the EDA market only in the last two years with the purchase of the Canadian company Electronic Workbench, has traditionally served the electronic instrumentation market. Their approach is to enable technicians and system designers to design, develop, and debug, small to medium size analog designs without having to be SPICE experts. Their users build circuits by picking parts from a quite large inventory of parts supported by NI. The parts database offers schematics, SPICE models, and footprints of a varied inventory assembled and built in collaboration with semiconductor vendors like Analog Devices, Texas Instruments, and Linear Technology. Their environment allows developers to easily compare simulation results with actual results from breadboarding the design.
The Mathworks, a company that grew by offering solutions to mathematical problems associated with DSP design, instead has developed a large inventory of "toolboxes". These are application specific IP and mathematical solvers that allow an engineer specializing in a discipline different from electronics, to incorporate electronic solutions into the system design. In their environment, for example, an hydraulics engineer can incorporate control electronics into an hydraulic system.
Both companies are showing that in addressing a new market, one must not be afraid to abandon "the old ways" and study the possibilities with an open mind. The EDA industry is still blinded by silicon: it is time we realize that silicon is the mean to a solution, not the solution.