The workflows use Nemotron, OpenShell and AgentEngineer for verification, AMS design and thermal analysis.<\/p>\n
Synopsys, Inc.<\/a>\u00a0announced advancements to agentic AI for engineering in collaboration with\u00a0NVIDIA<\/a>. Synopsys has developed fully autonomous, long-running agentic capabilities for chip design and electronics system design enabled with NVIDIA Nemotron on NVIDIA\u2019s accelerated computing platform and secured by the NVIDIA OpenShell runtime. Demonstrated at DAC for the first time, Synopsys\u2019 capabilities promise to be a force multiplier for R&D teams beyond task agents, transforming time-consuming chip verification and thermal simulation into automated insight delivery, engineering productivity, and system performance improvement engines.<\/p>\n Introducing Synopsys\u2019 fully autonomous design verification (DV) workflow<\/p>\n Coverage closure has been among the most significant bottlenecks in the DV process. Despite assistive tools, engineering teams spend significant labor and compute resources on incremental improvements. The companies are evolving the DV approach from tool-assisted to a goal-driven workflow that autonomously pursues coverage closure and traces root failure causes throughout development.<\/p>\n The solution, built on Synopsys\u2019 agentic AI platform and powered by\u00a0Synopsys AgentEngineer<\/a>\u00a0technology and NVIDIA\u2019s agentic AI infrastructure \u2014 including NVIDIA Agent Toolkit, NVIDIA Nemotron 3 Ultra open model, and OpenShell runtime \u2014 features a fully autonomous, long-running orchestrator agent. The orchestrator agent deconstructs DV goals from specification, design, test repository, and user inputs, and orchestrates specialized agents and tools in a closed loop workflow spanning the full chip verification lifecycle, from test plan generation to coverage closure and advanced debug. Demonstrated at DAC, the end-to-end fully autonomous verification closure agentic flow compresses weeks of manual labor into hours of agentic execution that achieves up to 50X faster time-to-validated RTL with an additional 20% improvement in coverage.1<\/p>\n Autonomous Analog & Mixed-Signal (AMS) workflows<\/p>\n AI-powered\u00a0Custom Compiler Layout Synthesis<\/a>\u00a0(CCLS) is laying\u00a0the foundation for autonomous analog and mixed-signal (AMS) design by automating layout generation, optimization, and design-layout convergence. Building on these capabilities, Synopsys AgentEngineer technology orchestrates multi-step analog flow spanning design creation, SPICE simulation, implementation, and verification. Engineers define design intent and performance goals in natural language, while autonomous agents execute and optimize the workflow, accelerating design closure and improving productivity by up to 3X.2<\/p>\n Delivering autonomous engineering agents across design and simulation<\/p>\n Synopsys developed a fully autonomous agentic CAE workflow for electronics thermal analysis using NVIDIA Agent Toolkit and NVIDIA CUDA-X libraries. Built with\u00a0Ansys Icepak<\/a>\u00a0electronics cooling simulation software \u2014 now part of the Synopsys portfolio \u2014 and open-source PyAEDT libraries, the agentic workflow autonomously executes simulation set-up, pre- and post-processing in a fraction of the time required for traditional approaches.<\/p>\n Extending the value of GPU acceleration to more engineering workflows<\/p>\n Synopsys continues to accelerate innovation with the industry\u2019s broadest portfolio of more than 20 GPU-enabled EDA and multiphysics products, unlocking deeper analysis and faster time-to-market across the design flow \u2014 from physical verification to photonics simulation. Recent developments include:<\/p>\n PrimeSim SPICE circuit simulations perform up to\u00a018X faster\u00a0leveraging NVIDIA GPUs.3<\/p>\n Synopsys QuantumATK<\/a>\u00a0accelerates next-generation semiconductor material innovation by up to 50X for Gaussian-basis quantum chemistry simulations enabled by cuEST and up to 200x faster machine-learned force field simulations using NVIDIA Blackwell GPU infrastructure.<\/p>\n Ansys Lumerical FDTD<\/a>\u00a03D electromagnetic simulation software achieved a\u00a010X speedup\u00a0on NVIDIA GPUs compared to CPUs when used within\u00a0Synopsys\u2019 Multiphysics Fusion<\/a>\u00a0solution for analog and photonic design.\u00a0<\/p>\n In addition, Synopsys continues its deep collaboration with NVIDIA\u00a0leveraging\u00a0CUDA-X libraries<\/a>\u00a0to accelerate its solvers, including\u00a0cuLitho<\/a>,\u00a0cuDSS<\/a>,\u00a0cuEST<\/a>, and\u00a0use cases in development<\/a>\u00a0with the newly announced\u00a0cuISS\u00a0library.<\/p>\n Availability<\/p>\n Customers are currently evaluating Synopsys\u2019 agentic EDA and CAE capabilities with availability planned for the second half of 2026.\u00a0<\/p>\n