NVIDIA is working with TSMC to introduce a computational lithography platform called cuLitho, and significantly improves the efficiency of semiconductor manufacturing. cuLitho leverages NVIDIA's H100 Tensor Core GPU to increase the performance of computational lithography by up to 40x and contribute to reducing power consumption1[3] (https://www.linkedin.com/pulse/nvidia-culitho-gpu-powered-revolution-semiconductor-tunguz-ph-d--5xule) [5] (https://www.photonics.com/Articles/Semiconductor_Leaders_Deploy_NVIDIA_Computational/a68868) ).
- GPU Acceleration: cuLitho leverages GPU's parallel processing capabilities to perform lithographic tasks significantly faster than traditional CPU-based calculations, enabling 500 NVIDIA DGX H100 systems to replace tasks on 40,000 CPU systems[3](https://www.linkedin.com/pulse/nvidia-culitho-gpu-powered-revolution-semiconductor-tunguz-ph-d--5xule)[5](https://www.photonics.com/Articles/Semiconductor_Leaders_Deploy_NVIDIA_Computational/a68868). - AI INTEGRATION: NVIDIA applied generated AI to cuLitho, further doubling its Optical Proximity Correction (OPC) process, enabling more accurate mask generation in semiconductor manufacturing1[6] (https://convergedigest.com/tsmc-adopts-nvidia-culitho-for-faster-chip-production/) . - Industrial Application: cuLitho is being used by TSMC and Synopsys in production, which is contributing to reducing time and cost of semiconductor manufacturing[4](https://semiwiki.com/eda/343568-tsmc-and-synopsys-bring-breakthrough-nvidia-computational-lithography-platform-to-production/)[8](https://www.datacenterdynamics.com/en/news/tsmc-and-synopsys-use-nvidia-culitho-in-chip-production-explore-generative-ai-for-computational-lithography/).
- Cooperation with TSMC: TSMC CEO Dr. C.C. Wei noted, "Cooperation with NVIDIA has greatly helped improve performance, increase throughput, reduce cycle time, and reduce power consumption1[8] (https://www.datacenterdynamics.com/en/news/tsmc-and-synopsys-use-nvidia-culitho-in-chip-production-explore-generative-ai-for-computational-lithography/) . - Working with Synopsys: Synopsys integrates cuLitho into its software, significantly improving the throughput of its mask synthesis software [4] (https://semiwiki.com/eda/343568-tsmc-and-synopsys-bring-breakthrough-nvidia-computational-lithography-platform-to-production/) [8] (https://www.datacenterdynamics.com/en/news/tsmc-and-synopsys-use-nvidia-culitho-in-chip-production-explore-generative-ai-for-computational-lithography/) ).
CuLitho accelerates innovation in semiconductor manufacturing, enabling smaller transistors and higher density semiconductors. It will support the development of sub-2 nm technology nodes and promote the continuous development of the semiconductor industry[5](https://www.photonics.com/Articles/Semiconductor_Leaders_Deploy_NVIDIA_Computational/a68868) [7] (https://www.rs-online.com/designspark/tsmc-partners-with-nvidias-culitho-platform-to-revolutionize-semiconductor-manufacturing) . The collaboration between NVIDIA and TSMC will be instrumental in increasing the efficiency of semiconductor manufacturing and reducing its environmental impact. Source: https://blogs.nvidia.com/blog/tsmc-culitho-computational-lithography/
Share this post: