?s on EDA Technology, other applications

[Arthur Hanson]

Knowing very little about EDA, I have a few questions as to other applications of the knowledge base. Is it possible much of the same knowledge, skill sets and templates could be used to automate platforms and the interface of AI systems and platforms for everyday use? If this is possible, being a financial type, it should be possible to lower the cost of EDA by spreading the cost over a much larger base. The use of logic in EDA should be applicable to many other areas, especially in user interfaces for complex platforms. If this created a compounding machine, this has the capability of dramatic changes in the world in education, design and interface with systems, bringing a fundamental change to almost everything we come in contact with. I feel this could create a whole new wave of applications in interfaces, education and design through the use of platforms. Any thoughts or comments are appreciated. Also are any projects of this type going on now.

 

[Daniel Payne]

I've been in EDA since the 1980's and have yet to see any EDA vendor find and apply their domain-specific technology to an everyday use. Yes, the computer programmers in EDA can leave the industry and continue to apply their coding skills to everyday problems because computer science skills are in high demand.

 

[Daniel Nenni]

Currently systems companies are the target EDA customers versus the traditional chip companies. EDA companies are moving up the electronics supply chain, mostly by acquisition. Synopsys is the best example:

https://www.synopsys.com/company/acquisitions.html

Software security and quality is big of course. Synopsys just today acquired an analytics company:

Synopsys Acquires Semiconductor Analytics Innovator Qualtera

As a result EDA software has had a nice multiplier over the past few years. SNPS used to trade at $20+ 10 years ago and now it is approaching $200. Great company, great leadership and great strategy, absolutely.

Knowing very little about EDA, I have a few questions as to other applications of the knowledge base. Is it possible much of the same knowledge, skill sets and templates could be used to automate platforms and the interface of AI systems and platforms for everyday use? If this is possible, being a financial type, it should be possible to lower the cost of EDA by spreading the cost over a much larger base. The use of logic in EDA should be applicable to many other areas, especially in user interfaces for complex platforms. If this created a compounding machine, this has the capability of dramatic changes in the world in education, design and interface with systems, bringing a fundamental change to almost everything we come in contact with. I feel this could create a whole new wave of applications in interfaces, education and design through the use of platforms. Any thoughts or comments are appreciated. Also are any projects of this type going on now.

 

[Arthur Hanson]

Another observation I have made is that doing using smart/AI programs in the cloud over a larger and larger base with ever more diversity of uses would also gain in intelligence and diversity as volume increases and time passes. Someone will set up an Amazon style market of automated and AI platforms for numerous areas and fields. This could radically increase the versatility and penetration of AI platforms and continuously decrease their costs. It will be interesting to see what companies take first mover advantage. This will present a large competitive threat to all other companies that not only design automation, but any business or research organization that relies on AI and/or automated platforms. This could also radically upset the currently vastly overpriced educational/training structures that currently exist by creating a vast amount of leverage of any product of this type over an ever growing base. Amazon, Alphabet and Microsoft are all very, very well capitalized and have the skill sets in place to do this. I hope the US becomes the leader in this, for if we loose this race to China, which I feel is a definite possibility, it could put China into a position to be the world leader. A very interesting possibility would be for TSM to merge EDA directly into its ecosystem as a way to improve the depth, speed and quality of their ecosystem, increasing their lead to a point they would have no real competitorn. The possibilities and options for combining AI and platforms are almost infinite in scope, power and economics in ways most haven't even imagined maybe even creating a universal design automation for everything. Any comments on this also appreciated.

 

[MarcL]

Check out Aditazz (www.aditazz.com). This company is trying to apply EDA methodology and formalisms to the construction industry. For example, the layout of a hospital with all its conduits and constraints on patient transport can be tackled as a layout optimization problem. I know the founders include some with extensive EDA experience.

 

[Bernard Murphy]

Check out Aditazz (www.aditazz.com). This company is trying to apply EDA methodology and formalisms to the construction industry. For example, the layout of a hospital with all its conduits and constraints on patient transport can be tackled as a layout optimization problem. I know the founders include some with extensive EDA experience.

Interesting concept

 

[Arthur Hanson]

EDA from my perspective is the one of the most powerful near AI tools in existence with very widespread use. To not leverage this resource is to give an advantage to other industries. The first EDA firm to implement strategies effectively that broaden the customer base outside their traditional customer base could easily develop a competitive threat to other EDA firms. Also, what must be considered is some other industry/business may have the ability to take away market share by adapting their software to EDA for semis from their traditional area. Who ever develops the best practices for developing AI and its applications may take all. For this reason EDA firms must take first mover advantage before someone else does. It isn't to difficult to imagine a smart AI/ML firm could come to dominate not only EDA, but much of AI which will become key in many areas including advanced uses of augmented reality. This is an exciting race between all forms of EDA already in use in many fields and applications. The opportunities are simply to large to pass up.

 

[count]

EDA from my perspective is the one of the most powerful near AI tools in existence with very widespread use

What makes you say this? I'm not an expert in EDA, but I know a fair bit about AI/optimization, and my understanding is EDA is for the most part a (fairly advanced) application of constrained optimization techniques that already fairly widely used in many different industries (although perhaps less intensely). Are there specific techniques in EDA that you know of that are more broadly applicable? Where do you think they apply?

 

[Arthur Hanson]

What makes you say this? I'm not an expert in EDA, but I know a fair bit about AI/optimization, and my understanding is EDA is for the most part a (fairly advanced) application of constrained optimization techniques that already fairly widely used in many different industries (although perhaps less intensely). Are there specific techniques in EDA that you know of that are more broadly applicable? Where do you think they apply?

I said specifically it is one of the most powerful tools, but there are many. The advantage of EDA is the amount of resources and money put into semis which in themselves cover a very, very broad area. EDA has to continuously evolve to keep up the rapid advance of semis in many, many areas with a very broad diversity of applications. Data science is about the manipulation of information and science and with EDA and semi equipment of all types being monitored continuously for performance and results real time, it is constantly evolving as new designs, materials, architectures and equipment work together to advance at a rate almost unmatched in other areas. The only other area that I can think of that coverage this broad is sophisticated finance programs that are mostly very proprietary with just as much secrecy as the semi sector, if not significantly more. This is all about massive data and its applications that are changed and advance in real time. Count, I deeply appreciate your observations and comments and value them. Data science is the ultimate application of AI/ML which is going to take man kind into more frontiers virtual and real than most can even imagine, let alone comprehend. Comments, thoughts and additions welcome since I'm just scratching the surface.

 

[count]

I see, so you're interested in other industries that use AI/ML, advanced optimization at a similar intensity as EDA.

Finance is one as you've pointed out, AI is absolutely transforming the industry right now. You are seeing a lot of interest in using different types of data in insurance underwriting (your phone has an enormous amount of info on you that's useful for insurance companies, like where you spend your time, how fast you drive, ect ect). Health is another big one, although constrained by regulation. Natural resources is an overlooked sector, big data has played a pretty big role in the shale revolution, but generally speaking they are using more Baysian methods vs deep learning. That's been changing and there has been more of a push there in AI/ML, but I'd say that sector is trailing a bit in adoption. One interesting area is media, where you can use generative neural nets to automatically create faces, and you can deepfake an actors face onto a double. Another interesting area is design of engineered parts, where you have spec for a part with a bunch of constraints on it, and you can generate a part that fits the constraints. In the future this may extend beyond engineered parts into more complex engineered systems. In AgTech/vertical farming, this is another really interesting area where historically we've bred plants for environmental resistance and that artificial selection process has taken hundreds of years in some cases, but if we change the constraints and we have total control of the environment, a vertical farm might want to identify plant genetics that thrive in a controlled environment, and they don't want to spend 50-100 years through trial and error artificial selection.

 

[Arthur Hanson]

Count, I feel in the future education/training will be about learning an AI/ML platform that continuously advances that one subscribes to and your task will be to be more creative and effective in applications. Some of these platforms will advance with each use by subscribers and others for IP reasons will be closed. Any thoughts on this will be appreciated. I feel standard education as we know it is a thing of the past for it will never achieve the leverage of an active real time platform system.

 

[Arthur Hanson]

I've been in EDA since the 1980's and have yet to see any EDA vendor find and apply their domain-specific technology to an everyday use. Yes, the computer programmers in EDA can leave the industry and continue to apply their coding skills to everyday problems because computer science skills are in high demand.

Daniel, as you can see from the timeliness of my reply, I have given much contemplation to the fact that EDA companies have not given serious thought to broadening the use of their knowledge base. I feel this would not only expand their market base, but also their thinking and even improve their base EDA products. Any additional thoughts on this are appreciated.