The most needed EDA innovation

Written by Jan Decaluwe March 19, 2010 Hits: 872 Favorites

3.0 (1)

I have been unhappy with EDA lately, especially EDA for digital design. So unhappy, that I have been losing interest in technological innovations. In fact, I believe the most needed EDA innovation lies elsewhere.

In a recent blog post, I have explained my thoughts in detail. If you are interested, you can read it here.

User reviews

Average user rating from: 1 user(s)

To write a review please register or login.

Overall:

3.0

Missing the point

Overall:

3.0

Reviewed by Brian Bailey
March 24, 2010

Comments (1)
Report this review

In your blog you talk about wanting hardware design to be more like software design in terms of open source, common frameworks etc. You can never expect EDA companies to behave like software development companies because of the orders of magnitude difference in the number of users and the complexity of the preparation process. Lets start with numbers first. Software tools sell for a few $1000. There are perhaps 10,000 hardware designers out there and perhaps another 20,000 verification engineers. So is we sold every engineer a tool for $1000, we would have a $3M business. Now for that you just aren't going to get a lot of innovation - and it shows in the software industry - even though there are millions of them. They are still using the same antiquated design, development and verification strategies that have failed them for decades. Now think about the advances for the hardware engineers that have enabled them to achieve enormous productivity gains using technologies and geometries that were thought to be impossible just a few years ago. And the software guys still can't get their heads around concurrency!

If you think that starting from C or SystemC and creating complex concurrent hardware from it does not imply innovation and being able to produce hardware with few bugs because of the quality of the verification environments does not imply enormous creativity, then you really do have very dark glasses on.

Owner's reply

I wonder where your numbers come from? My own data suggests that the
number of HW designers is an order of magnitude higher. But anyway,
this wouldn't probably alter your analysis substantially, and it is
not my main point.

Elsewhere on my blog you'll find that my principal viewpoint is that
the division between EDA for digital design and software design tools
is artificial. From that angle, EDA would not be a different industry but
a specialized part of a much larger one. And, naturally, EDA would
directly benefit from many software design innovations.

Of course, this assumes that you believe that the software industry
has something to offer :-) You suggest that you don't (a little
surprizing because I have read different things from you in other
posts). However, I plead guilty here. After RTL synthesis, what has
helped me most in my hardware design career have been non-EDA specific
software development tools and techniques.

Note that my own assessment about EDA is nuanced. I have nowhere
asserted that there has not been any innovation recently. I have said
that I'm losing interest and that I don't know because access has
become so difficult (expensive).

Finally, a word about concurrency. There is a whole spectrum, from
fine-grained to coarse-grained, with centralized or distributed
control. Hardware designers have a very good understanding of one
type, at the RTL level: very fine-grained concurrency with central
control. Suggesting from this that we have a lot to teach about the
general problem is a bridge too far for me.

Written by :

Jan Decaluwe