Synopsys has partnered with NVIDIA to enhance electronic design automation (EDA) efficiency, achieving up to 30x performance improvements for next-generation semiconductor design on the NVIDIA Grace Blackwell platform.
Performance Enhancements
At NVIDIA¡¯s GTC conference, Synopsys showcased its full-stack EDA solutions accelerated by NVIDIA GPUs and CUDA-X libraries. Key advancements include:
- PrimeSim SPICE simulation on the NVIDIA GB200 Grace Blackwell Superchip, expected to deliver 30x faster circuit simulation.
- Proteus computational lithography on the NVIDIA B200 Blackwell architecture, projected to achieve 20x faster simulation.
- Integration of NVIDIA NIM inference microservices with Synopsys.ai Copilot, doubling response speeds for generative AI-driven design assistance.
By 2025, over 15 Synopsys solutions will be optimized for the NVIDIA Grace CPU platform, expanding support for advanced EDA workflows.
Collaboration Details
Synopsys is leveraging NVIDIA¡¯s CUDA-X libraries and Grace Blackwell platform to optimize its EDA tools, significantly reducing computation times for critical tasks:
- Circuit Simulation: PrimeSim SPICE on Grace Blackwell is expected to achieve 30x acceleration, reducing simulation times from days to hours for complex circuits with SPICE-level signoff accuracy. The NVIDIA GH200 Superchip already provides up to 15x speedup.
- Computational Lithography: Proteus, used for optical proximity correction (OPC) and inverse lithography technology (ILT), integrates with NVIDIA¡¯s cuLitho library and H100 GPU for 15x OPC acceleration. Blackwell is expected to push this to 20x.
- TCAD Simulation: Sentaurus TCAD, under development, shows 10x faster process and device simulations using NVIDIA CUDA-X libraries and GPU support, with availability planned for later this year.
- Materials Engineering: QuantumATK for atomic-scale modeling achieves 100x faster computations on NVIDIA¡¯s Hopper architecture with CUDA-X, enhancing material simulation efficiency.
AI and Grace CPU Integration
Synopsys.ai Copilot, a generative AI knowledge assistant, doubles designer productivity. NVIDIA NIM microservices are expected to further double its response speed, accelerating design workflows.
Synopsys is optimizing over 15 EDA solutions for the NVIDIA Grace CPU, including circuit simulation, physical verification, static timing analysis, and functional verification, with expanded support planned for 2025.
