NVIDIA announced that ASML, TSMC and Synopsys are adopting the newly introduced cuLitho library. The tool is integrated into the software for designing photomasks used in the production of chips, and greatly accelerates the preparation of lithographic photomasks. Weeks of heaviest compute workloads run in 8 hours on a cluster of NVIDIA GPUs. This is the way to the future of semiconductor lithography, the company is sure.
An example of a modern photomask (photomask). Image Source: Intel
NVIDIA claims that running cuLitho on the GPU delivers up to a 40x performance boost over current software. If we imagine that many factories and design centers design photomasks all over the world, then this is tens of billions of hours of processor load per year. These are megawatts and megawatts of energy that NVIDIA wants to save the semiconductor industry from burning.
The company estimates that 500 NVIDIA DGX H100 units will do the same job as 40,000 CPUs. This will not only reduce your carbon footprint, but will also help you produce 3-5 times more photomasks every day using 9 times less energy compared to current configurations. A photomask that took two weeks to make can now be processed in one night.
“The chip industry is the backbone of almost every other industry in the world,” said Jensen Huang, founder and CEO of NVIDIA. “With lithography at its physical limits, the introduction of NVIDIA cuLitho and collaboration with our partners TSMC, ASML and Synopsys allows fabs to increase throughput, lower their carbon footprint, and lay the foundation for 2nm and beyond.”
Vivek Singh, Vice President of NVIDIA Advanced Technology Group, explained at the press conference that cuLitho will run not only on the latest NVIDIA Hopper architecture, but also on the company’s older GPUs, starting with the 2017 Volta architecture.
Designing photomasks has long ceased to be a trivial task. Today, each layer of a chip may require several dozen photomasks, and this is not just a geometrically verified pattern. To make the part on silicon as small as possible and with sharply defined edges, various tricks with light and projection are used, when not only light comes into play, but also shadow, and distortion, and cross-shading, and much more. The photomask has become an art that has ceased to be on the shoulder of a person.
According to NVIDIA, ASML and Synopsys are busy integrating the cuLitho library into their software. TSMC, as explained by Jensen Huang, will begin testing the software on production equipment in June this year.
As far as you can tell from Huang’s talk, the cuLitho library doesn’t use AI algorithms yet. But there is no doubt that, over time, full-length AI will be used in the preparation of chips for production. Machine learning is already being used in chip design, as Google talked about some time ago. Offering the cuLitho library to the industry is undoubtedly doing a good deed. At the same time, it deeply embeds itself in the chain of production and supply of chips. And the further, the stronger this dependence will be, as today the world of advanced chips depends on only one company – the Dutch ASML
If you notice an error, select it with the mouse and press CTRL + ENTER.