While the large EDA companies continue to view hardware and software as two distinct markets, The Mathworks, National Instruments and a few small companies and FPGA vendors are finally admitting that system level design must deal with both implementation technologies concurrently and with the same level of support.
Cadence, Magma, and Synopsys continue to have their heads firmly planted in the silicon that has provided the lion share of their revenues, and what they have been willing to share with me indicates that they do not see the shrinking number of companies able to undertake designs at the leading edge of the process curve as a problem. Just raise prices and all will be well. But all is not well, if the resignation of Moshe Gavrielov at Cadence is an indication.
Of the traditional four leading EDA companies Mentor has the largest investment in support of software development. But its efforts in System Level Design are mostly concentrated in the expansion of support for SystemVerilog in the Questa product line and the efforts to support C in the Catapult synthesis product. It is true that Mentor pioneered the support of MATLAB in its verification environment a few years ago, but I have not seen much expansion of the relationship since then.
When editors and analysts talk about EDA companies, National Instruments and The Mathworks are hardly ever mentioned, with the exception of Gary Smith who really appreciates what The Mathworks is doing in the ESL space, and myself. I have covered both companies on this site and The Mathworks in other publication before joining DesignLine. So you have to look at Catalytic to find an EDA company that offers the ability to generate firmware from MATLAB, or Xilinx after its acquisition of Accelchip and Synplicity to find an embedded software generation capability from MATLAB.
Of course none of these companies offers the type of system-level exploration and design that The Mathworks offers, so if you are an architect you have little choice.
Pundits continue to ask why the ESL market is not taking off. The answer: it is but not within EDA companies. EDA companies are so blinded by hardware, the root of their inceptions, that they have a difficult time understanding that a system is not just "electronics". The breakthrough will occur, as I said a number of times before, when we get rid of the "E" in ESL. It's the SYSTEM stupid, not the electronics part of it!
Being able to fully simulate a product in Simulink using MATLAB in a seamless way and then generate both embedded software code in C and synthesizable hardware in Verilog or VHDL is a powerful environment.