Though we have an academic pipeline set, skilled personnel and mature infrastructure then what is it that isn’t accelerating AI on a large scale? A panel of experts at EE Times’ recent AI Everywhere Forum was asked to pick the holdup for AI’s rollout — software or hardware?
However, the answer isn’t that simple
Need for Data Engineers
Data, the new oil of the 21st century lies at the core of making AI a success story. However, what the industry of code writers and chip designers can’t quite work their way around are the questions surrounding the data and a lack of experts who know data and AI systems, referred to as “data engineers.”
Some potential questions raised:
“What’s the Github of data?
How do you version data?
How do you incrementally add data?
How do you share data?
How do you as a community do it so you’re not doing everything in-house?
The solution seems to be we need innovate at every layer of the AI implementation stack. This can be done by having integrated systems that are optimized as such. Such an innovation will result in a vertically integrated solution thus creating a mesh of data aligned well to be used by AI. Apple products would be great example for “Go Vertical” philosophy, they innovate at every stack level including the chip design, hardware implementation and the software.
AI and EDA
Are AI models for EDA (electronic design automation) and chip design beginning to show real usefulness? How much potential is there for future algorithms to automate more parts of the chip design process?
Metcalfe said that “there is now very good proof that AI is going to be a very transformational technology from a chip design perspective. What we see today is small parts of the chip design flow adopting AI technology. Moving forward, there’s going to be a huge amount more work done in this area. So we can expand this to system-level optimization. Today, we’re looking very much more at the hardware implementation, but one layer above that is the whole system optimization. How do you meet the latency targets you need? How do you meet the processing targets? There are lots of different architectural decisions you can make very early on in the process that AI can certainly, certainly help with.”
That includes implementation, partitioning and optimization in 3DIC design, which is notoriously complicated, Metcalfe said.