NVIDIA said it is adding cuPotential, a new CUDA-X library that developers can use to build AI agents that simulate how atoms and molecules interact, as semiconductor leaders adopt its libraries for what it calls AI-defined fabs.[1] AI-defined fabs are semiconductor factories where AI helps engineers optimize processes, solve production issues and improve production efficiency, NVIDIA wrote.[1] NVIDIA said cuPotential is available in early access.[1]
Modeling atomic interactions is slow even in GPU-optimized software, NVIDIA said, and cuPotential adds GPU-accelerated functions that developers can add to existing software.[1] In early-access testing, the library accelerated already GPU-optimized workflows by an additional 2-3x, the company said.[1] Synopsys is among the early adopters using cuPotential to accelerate materials simulations.[1]
NVIDIA said Applied Materials, Cadence, KLA, Lam Research, Siemens, Synopsys and other semiconductor leaders are using its accelerated computing and AI stack across chip design, manufacturing and testing.[1] Lam Research intends to integrate NanoVDB into its SEMulator3D process modeling platform, and NVIDIA said the addition is expected to accelerate select wafer-etch modeling by up to 30x.[1] KLA applied insights from its production kernels to a new NPP kernel that achieved performance levels 50x faster than earlier CPU versions, NVIDIA said.[1] Applied Materials is using Omniverse and CUDA-X libraries across materials discovery and fab optimization, enabling up to 35x faster chamber simulations, according to NVIDIA.[1]
Teradyne is using NVIDIA NPP and accelerated computing to accelerate image-sensor testing, and NVIDIA's newsroom says it delivers approximately 85x faster probe-maintenance inference.[1] NVIDIA reported that examples include Cadence AuraStack AI Super Agent optimizing advanced packages for heat and stress 8x faster using cuDSS with Cadence Celsius.[1]
