Vinci pushes the boundaries of chip design with physics AI

Vinci pushes the boundaries of chip design with physics AI

Physics AI is bringing a new approach to semiconductor engineering, where increasingly complex designs make physical behavior harder to predict. Advanced accelerators combine multiple chips, memory and packaging components, creating tightly coupled engineering challenges.

Vinci4D Inc. uses physics AI to bring simulation into engineers’ everyday design work. Its approach addresses a familiar engineering tradeoff, according to Dinesh Panneerselvam (pictured), head of technical solutions at Vinci.

“Oftentimes, what I see in the field is engineers have to choose between accuracy or fast physics,” he said. “They don’t get to do both. What Vinci does is bring them both at the speed at which they’re designing.”

Panneerselvam spoke with theCUBE Research’s Christophe Bertrand at SEMICON West, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed physics-based simulation, earlier design exploration and protecting customer data. (* Disclosure below.)

Physics AI moves analysis upstream

Vinci combines a physics foundation model with a GPU-native physics kernel in an approach it calls Continuous Physics Reasoning. The combination produces repeatable results with solver accuracy, according to Panneerselvam.

“In a product development world, the team that understands the physics — or the behavior of their product early on — they win the race,” he said. “That’s exactly what Vinci is providing. That means you design at your architectural level. We provide what the performance is.”

The model is trained on Vinci’s own data rather than customer data. Deployment behind the customer’s firewall keeps proprietary design information with the customer, Panneerselvam explained.

“The way we have designed our platform, it works right into your workflow,” he said. “You don’t have to change any of your current workflows. We ingest your design files as is and start building the model. There is so much automation built in that there’s very little manual input.”

Extending physics AI beyond chips

Physics AI could also support the development of physical AI, including robots and humanoid systems. Vinci’s simulation approach could help engineers understand how those products would behave before committing to a particular design, according to Panneerselvam.

“Customers are able to open doors for new innovation — things that they haven’t imagined before,” he said. “For example, you have an engineering intuition for a product, and it stops there. What Vinci and products like Vinci allow is basically [to] explore beyond that. We call it the art of the impossible, and you can move into that space.”

Vinci is exploring applications in automotive and aerospace engineering as it expands beyond its semiconductor base. Its $250 million Series B funding round will support development across additional physics disciplines and industries, including planned capabilities in electromagnetics and fluid dynamics, Panneerselvam explained.

“The bottom line is physics doesn’t change,” he said. “It’s just the applications that we have to build. The physics remains the same.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of SEMICON West:

(* Disclosure: TheCUBE is a paid media partner for the SEMICON West event. Neither Vinci, the sponsor of theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE