NVIDIA Blackwell and CUDA-X Boost Semiconductor Manufacturing Efficiency with Up to 25x Faster Lithography Simulations

The semiconductor industry is witnessing a significant leap in design and manufacturing capabilities through the adoption of NVIDIA's Blackwell GPUs, Grace CPUs, NVLink fabrics and switches, and domain-specific CUDA-X libraries. These technologies are driving improvements in computational lithography and device simulation, crucial for developing advanced chips. Key Players and Collaborations TSMC (Taiwan Semiconductor Manufacturing Company), the world's leading foundry, has partnered with NVIDIA to enhance its Technology Computer-Aided Design (TCAD) division. According to Jeff Wu, TSMC's Fellow and Director for the TCAD division, the computational acceleration provided by CUDA-X libraries and the NVIDIA Grace Blackwell platform will accelerate process development, reduce costs, and simulate complex manufacturing processes and device behaviors more efficiently. Cadence, an Electronic Design Automation (EDA) software and services provider, has launched the Millennium M2000 platform exclusively on NVIDIA Blackwell. This scalable solution integrates Grace Blackwell and CUDA-X libraries to accelerate Cadence's portfolio of design tools. Michael Jackson, Corporate Vice President and General Manager of the System Design and Analysis Group at Cadence, highlighted that the M2000 platform isn't just about faster simulation but redefines AI-driven infrastructure, enabling advanced silicon, system, and even drug design. Cadence has also adopted NVIDIA NVLink Fusion, a technology that facilitates custom silicon scaling for demanding workloads like model training and AI inference. Siemens EDA is integrating NVIDIA's CUDA-X libraries and Grace Blackwell to boost its Calibre platform, critical for semiconductor manufacturing. Mike Ellow, CEO of Siemens EDA, noted that this integration accelerates optical proximity correction without compromising accuracy, which is essential as chip complexity grows. The Calibre platform now performs tasks like physical verification, design for manufacturability analysis, reliability verification, and design-to-manufacturing flow automation more swiftly and accurately. Synopsys, another major EDA provider, is leveraging NVIDIA's CUDA-X libraries and Blackwell for various tools, including PrimeSim, Proteus, S-Litho, Sentaurus Device, and QuantumATK. By integrating with CUDA-X, Synopsys has seen remarkable performance gains, with PrimeSim and Proteus projected to run 30x and 20x faster, respectively, on Blackwell platforms. Sanjay Bali, Senior Vice President of Strategy and Product Management at Synopsys, emphasized the importance of this collaboration in redefining semiconductor manufacturing innovation through unprecedented speedups in simulation solvers. KLA, a semiconductor process control equipment manufacturer, has a long-standing relationship with NVIDIA. They have been optimizing high-performance computing solutions for AI-driven process control, which is becoming increasingly valuable due to complex designs, accelerated product cycles, and growing demand for advanced packaging. KLA's systems capture and process images using complex AI algorithms to detect critical defects quickly. KLA is currently evaluating the RTX PRO 6000 Blackwell Server Edition with CUDA-X libraries to further speed up inference workloads in chip manufacturing. Impact and Performance Gains The integration of NVIDIA's technologies is yielding notable performance improvements. For instance: NVIDIA cuLitho and Blackwell can speed up lithography by up to 25 times, allowing manufacturers to predict and correct issues before production. This is particularly beneficial for advanced nodes where accuracy and speed are paramount. Cadence's Millennium M2000 platform offers tailored options like the GB200 NVL72 system for large-scale designs and the RTX PRO 6000 Blackwell Server Edition for smaller projects, significantly enhancing simulation and validation efficiency. Siemens' Calibre platform sees a substantial boost in speed and accuracy for critical manufacturing steps, making it more adept at handling intricate processes. Synopsys' EDA tools demonstrate transformative speedups across different domains, with specific tools showing 12x, 15x, and 20x scale-up compared to existing CPU infrastructure. Industry Insights Industry experts are highly optimistic about the impact of these collaborative efforts. AI-driven innovations are redefining the landscape of semiconductor design and manufacturing, making it possible to create more complex and efficient chips in less time. The scalability and performance enhancements offered by NVIDIA's Blackwell and CUDA-X technologies are seen as game-changers, particularly as the demand for high-performance computing continues to rise. NVIDIA's comprehensive suite of hardware and software tools is not only accelerating traditional EDA workflows but also opening new avenues for AI and machine learning applications in the manufacturing process. By fostering these collaborations, NVIDIA is positioning itself as a pivotal player in the evolution of semiconductor technology, driving the industry towards more agile and innovative practices. Company Profiles NVIDIA is a leader in AI and graphics processing technology, known for its GPUs and CUDA ecosystem. The company's focus on high-performance computing and AI has positioned it at the forefront of semiconductor innovation. TSMC is the world's largest dedicated semiconductor foundry, providing advanced manufacturing services for a wide range of customers. Their adoption of cutting-edge technologies underscores their commitment to maintaining leadership in the industry. Cadence specializes in EDA software and services, offering solutions for system and chip design, verification, and analysis. Their collaboration with NVIDIA aims to push the boundaries of EDA, driving faster and more efficient design processes. Siemens EDA provides comprehensive tools for the design and manufacture of semiconductors. The integration of NVIDIA's technologies enhances their capabilities, ensuring that critical verification steps are completed quickly and accurately. Synopsys is a leading EDA tool provider, focusing on simulation and verification solutions. Their partnership with NVIDIA leverages GPU acceleration to achieve remarkable performance gains in their portfolio of EDA tools. KLA is a pioneer in process control and yield management for semiconductor manufacturing. Their use of NVIDIA's technologies optimizes defect detection and process control, critical for maintaining high-quality standards in chip production.