Semiconductor manufacturing giant TSMC is revolutionizing the chip production process by incorporating Nvidia’s cuLitho platform. This platform aims to speed up the computational lithography process, which is a crucial step in semiconductor production.
Computational lithography involves complex calculations in electromagnetic physics, photochemistry, and computational geometry. Traditionally, these computations are carried out in data centers using CPU systems, consuming a significant amount of time, power, and space.
Nvidia’s cuLitho platform, powered by 350 Nvidia H100 Tensor Core GPU systems, promises to replace 40,000 CPU systems. This replacement results in reduced production time, power consumption, and data center space, leading to improved efficiency and cost savings.
TSMC CEO C.C. Wei praised the collaboration with Nvidia, highlighting the significant performance improvements, throughput enhancement, cycle time reduction, and power efficiency achieved through the integration of GPU-accelerated computing in their workflow.
Moreover, Nvidia has integrated generative AI capabilities into cuLitho, which could potentially double the speed of optical proximity correction (OPC). OPC is a critical process in semiconductor lithography that enhances the accuracy of transferring circuit patterns onto silicon wafers.
By accelerating computation lithography, chip manufacturers can produce masks faster, thereby speeding up the development of new chip capabilities and enabling the implementation of previously impractical techniques. One such technique is inverse lithography, which involves designing a circuit by starting with the desired final pattern on the wafer and computing the optimal photomask pattern while accounting for optical and physical distortions.
These advanced techniques have been confined to the research lab for years due to intensive computing requirements. However, with the introduction of cuLitho, these techniques could potentially be integrated into the manufacturing processes of next-generation semiconductors, ushering in a new era of innovation and efficiency in the industry.
The collaboration between TSMC and Nvidia represents a significant milestone in semiconductor manufacturing, showcasing the power of accelerated computing and AI in driving progress and pushing the boundaries of what is possible in chip production. As technology continues to advance, we can expect further breakthroughs and innovations that will shape the future of the semiconductor industry.