A new library: cuPotential
NVIDIA added cuPotential to its CUDA-X library family. With it, developers can build AI agents that simulate how atoms and molecules interact. In early-access testing, the library accelerated already GPU-optimized workflows by an additional 2-3x.
AI-defined fabs
An AI-defined fab is a semiconductor factory where AI helps engineers optimize processes, solve production issues and improve efficiency. According to NVIDIA, these fabs must expand production while managing increasingly complex manufacturing processes.
Who is using it?
Companies including Applied Materials, Cadence, KLA, Lam Research, Siemens and Synopsys are using NVIDIA's accelerated computing and AI stack across chip design, manufacturing and testing. Lam Research achieved 30x faster simulation with NanoVDB, an open source library for 3D data, while KLA accelerated inspection with NVIDIA NPP, a GPU image-processing library.
Early results
Synopsys is using cuPotential, cuEST and cuDSS. Its early QuantumATK results showed a potential 30x speedup with cuEST, and S-Device benchmarks showed a 12x speedup with cuDSS. NVIDIA aims for these tools to automate more engineering work and help fabs predict process behavior and improve yield.
