Part of the  

Chip Design Magazine


About  |  Contact

Systems Beyond EDA’s ESL

The question often begins as follows: “Do you know of any ESL-like tools that would apply to a larger system that incorporates mechanical, electrical, HW, and SW.”

Great question! One that probably can not be answer by an EDA person, or anyone else that has a niche view. Likewise, it’s not a question that can be answered by the high-level, domain-independent tools, that is, System Engineering Tools like Slate, Core, Doors, etc. It’s analogous to the challenges – in our EDA world – of linking the worlds of algorithmic and RTL development. It’s a problem of too many layers of abstraction.

In the case of high level systems engineering tools like Slate, Core, and others, the problem is that all you can do at such a high level is establish an overall architecture then drill down with point tools for rqmts, hw-sw partitioning, etc. This was one of the problem that I faced many years ago while working for the DoE superfund clean-up site: How to go from a very high level problem discussions and perceptions from stakeholders, general public, lawyers, etc, to a traceable/verifiable solution implementation in hardware, software, peopleware, mechanical systems, etc. I should really talk more about that experience, since it’s very applicable here. But I don’t want to stray too far afield for this blog entry, that is , I want to focus on implementations that result in HW and SW.

To that end, earlier this year I directed my iDesign editor (Clive Maxfield) to focus exclusively on chip-package-board issues, more from an electronics viewpoint than a hw-sw division. Max wrote a great piece to initiate the new chip-package-board direction for iDesign.

This issue of subsystem views and tools (i.e., chip vs board vs module vs complete product) remains a challenge. EDA companies like Cadence, Synposys and Mentor are all trying to find ways to address chip-package-board level designs, but this is already the domain of big companies like Autodesk and Dassault Systems, among others:

“An acquisition of Cadence by AutoDesk – makers of AutoCAD – does make sense. Cadence makes several good point tools that would complement AutoCAD’s existing product engines, e.g., in the aircraft, automotive and multimedia markets. AutoCAD has all the 3D modeling, rendering and packaging tools that are coveted by the major EDA companies. AutoCAD is truly a big fish with around $4 ½ B in sales and a market cap of $9B. This makes AutoCAD roughly four times the size of Cadence. So an acquisition of Cadence makes both technical and financial sense.”

The problem of chip-package-board design is a big one – bigger than ESL. But the need for a solution is just as pressing. Any thoughts?

4 Responses to “Systems Beyond EDA’s ESL”

  1. Lou Covey Says:


    Welcome out of the EDA tunnel. You’ve stumbled on the future of EDA, and it’s acquisition.

    We should get together soon. I have some really big stuff in the works.

  2. John Blyler Says:

    Hi Lou. In some ways, I’ve approached the problem from the outside to inside, coming from a background of very large systems. Many of the EDA editors have come to the same conclusion looking from the inside to outside. Reaching the same conclusion from different starting points should validate the convergent findings of each perspective. We do live in interesting times.

    Would be great to chat again. I’ll be done for Intel Dev Forum in a couple of weeks, which will include a trip to San Jose. Let’s get together then.

  3. Jakob Engblom Says:

    A couple of comments on this.

    1. “The question often begins as follows: “Do you know of any ESL-like tools that would apply to a larger system that incorporates mechanical, electrical, HW, and SW.?” — I think that if you want to simulate a particular design to do the software, we have tools like MDA and large-scale virtual platforms that work fine for this. They are not tools that start from the top and design downwards, but they work fine to tune the software-driver-hardware-sensor/actuator-enviroment system without fiddling with physical prototypes…

    It is not designing the physical board, or car, or airplane. But it is certainly focused on the problems facing the electronics and in particular the embedded software and firmware that has to glue everything together.

    For the computer side of things, look at what companies like Virtutech are doing. Simulating truly complete systems from the software perspective.

    2. The economics you quote are fascinating. So Autocad itself is on par with the entire EDA industry. That is quite interesting, actually. It would seem that the “old fashion” of designing physical things from steel, plastics, and composites is a much bigger market and problem than designing the semiconductors that are necessary to create any modern product. The mechanical design of aircraft and cars seems to dominate the cost of designing the electronics that make them tick, from a software tools perspective.

    For some reason, computerization of everything has created a wealth of very powerful and previously unimaginable products… and huge markets for things like mobile phones, much better cars, audio and video electronics, etc… but that key role of computers and software in our modern world is not reflected in a huge market for tools to create them.

    The weirdest part is really beyond EDA in software. It seems that from a cost perspective in development, physical design is less than semiconductor design, and semiconductor design is swamped by embedded software design. But in tools, it is hte other way around. Embedded software is a pretty small market, EDA for chip design is bigger, and physical design is absolutely dominant.

    Guess us computer people are bad at inventing/accepting/paying for tools that make our lives easier, for some reason.

  4. John Blyler Says:

    EDA has more PhDs than any other industry. Yet the market itself is relatively small, compared to Autodesk and others. Concerning software, tho, the problem is one of perception. Hardware is tangible, SW is not. IMHO.

Leave a Reply