Image to help understand the article
Nvidia on Dec. 3 highlighted how artificial intelligence and graphics processors can accelerate the complex simulations used to design and manufacture semiconductors, cutting some predictions that once took days to milliseconds.
The company, best known for pioneering GPUs and supplying computing technology for AI, gaming, robotics and autonomous vehicles, described the approach as an “AI-defined semiconductor factory.” It combines AI and accelerated computing across tasks ranging from analyzing physical processes to operating equipment and optimizing production.
Such simulations become increasingly demanding as chip components shrink. Manufacturers must simultaneously model factors including heat transfer, electrical properties and mechanical stress to predict what will happen during production.
Simulations Accelerated by Up to 35 Times
Applied Materials incorporated Nvidia’s cuDSS GPU-accelerated numerical library into Ginestra, a platform that analyzes semiconductor materials and device physics at the atomic level. Nvidia said the integration delivered processing speeds up to 10 times faster than a CPU-only approach.
Applied Materials also adopted Nvidia’s PhysicsNeMo framework in its ACE+ multiphysics simulation platform. That application accelerated simulations predicting changes in microscopic structures during manufacturing by up to 35 times. Multiphysics software calculates how several physical forces and processes interact.
Comsol, another multiphysics simulation software provider, is using cuDSS to accelerate semiconductor manufacturing simulations by as much as 10 times. The technology allows manufacturers to optimize equipment and process settings and evaluate alternatives before testing them on an actual production line. Nvidia’s announcement also included applications involving Siemens and Lam Research.
U.S. Chip Industry Connections
The examples link Nvidia’s AI computing technology with major U.S. semiconductor equipment and software companies, including Applied Materials and Lam Research. The effort reflects how American chip-industry suppliers are applying AI not only to semiconductor products but also to the costly and technically demanding processes used to manufacture them.
Interest surrounding the announcement also extended to searches for Nvidia’s DGX Spark system and stock price, as well as practical queries about Nvidia drivers, its graphics control panel and software downloads and installation.
Comments
Post a Comment