NVIDIA expands NVIDIA Agent Toolkit with PhysicsNeMo and CUDA-X libraries

NVIDIA expanded NVIDIA Agent Toolkit for engineering by adding NVIDIA PhysicsNeMo and updating CUDA-X libraries. The update focused on agent-ready tools intended to connect physics, simulation, and performance analysis across engineering workflows that include complex design cycles.

The company said PhysicsNeMo was re-architected into agent-friendly libraries and that new and updated CUDA-X libraries were added to support complex engineering work. NVIDIA described PhysicsNeMo as providing AI physics skills for training and deploying models, while CUDA-X libraries bring accelerated solvers and quantum chemistry capabilities into agentic engineering workflows.

In the PhysicsNeMo and CUDA-X component set, NVIDIA named NVIDIA cuISS for iterative sparse solvers targeting large sparse linear systems, and NVIDIA cuDSS for direct sparse solvers for large, complex sparse linear systems used in EDA and scientific simulation. For quantum chemistry, NVIDIA pointed to NVIDIA cuEST (CUDA Electronic Structure Theory) for quantum chemistry simulations, including integration of density functional theory and post-DFT methods into production workflows.

NVIDIA also highlighted NVIDIA Nemotron 3 Ultra, stating it leads among open models in agentic register-transfer level coding on comprehensive verilog design problems benchmark tasks through ACE-RTL from NVIDIA Research. The company said developers could get started with Nemotron 3 Ultra using Cadence’s harness and using agents and toolkits associated with Synopsys and Siemens, as well as Hugging Face. NVIDIA added that Cadence, Synopsys, Siemens, Samsung, ChipAgents, Silvaco, Keysight, and others used components from the toolkit or related libraries in described workflows for design, verification, characterization, simulation, and solver acceleration.

“Engineering has reached an inflection point. AI can now work with tools of physics, simulation and design,” said Timothy Costa, vice president and general manager of computational engineering at NVIDIA. “With NVIDIA Agent Toolkit, developers can build agentic engineers that reason using physics, run complex simulations and generate high-fidelity data to become a new engine for innovation in chip and system design.”

Provided by Globe Newswire on behalf of Nvidia. Click to read original content.