Reflecting on the Future

Old, sequential EDA algorithms are running out of steam. But that’s actually good news.

popularity

Since returning from the Design Automation Conference, I’ve been reflecting on some very interesting discussions I had last week in Austin. The ones that are sticking with me concern the old, sequential algorithms that run EDA tools today. The fact is, given design complexity, they are running out of steam.

As a result, the industry is looking at possibly leveraging GPUs since they may be able to boost the sequential EDA algorithms. The question is how long that will work. My understanding is that to truly address design in the future, they need to be fundamentally rewritten. Is this a big challenge? Of course it is, but I have zero doubt the brilliant minds in this industry and in academia will develop what is needed…given the resources.

The resource issue leads to another interesting situation I wasn’t previously aware of. I had been under the impression that some of the revolutionary EDA technologies are developed in academia, which might be true in some cases, but a respected industry friend who holds a doctorate from Berkeley explained that funding at the universities happens in three-year projects, and during that time, the research team is expected to reach a certain result. However, this does not leave room for fundamental research for such things as…..yep, you guessed it: new fundamental EDA algorithms.

In a corollary vein, the same thing is happening in the area of so-called “low-power” design tools, which, another industry luminary pointed out, are just the same old performance-driven tools with low-power features kludged onto them. There too, a fundamentally new type of tool set needs to be created that focuses solely on power awareness.

I don’t mean this to sound negative in any way. In fact, these issues are the key drivers for the industry to not only grow beyond its traditionally moderate rate, but truly meet the design and verification needs of semiconductor developers, thereby raising the value of EDA tools overall.

~Ann Steffora Mutschler