As a fabless chip design company, NVIDIA’s semiconductor leadership has put the company in the driver’s seat for artificial intelligence (AI) and the company has partnered with lithography leader ASML, TSMC, and Synopsis on groundbreaking lithography breakthroughs and will help the computer industry make a new generation of powerful and energy efficient 2-nm semiconductors.
Only 1 nm smaller than the TSMC chips arriving a bit later this year, every step reduction in size now is getting exponentially more difficult to accomplish, creating a damper on progress with Moore’s Law.
NVIDIA cuLitho Library
The new NVIDIA cuLitho software library for computational lithography is being integrated by TSMC, the world’s leading foundry, as well as electronic design automation leader Synopsys into their software, manufacturing processes, and systems for next-generation NVIDIA Hopper™ architecture GPUs.
The chip industry is the foundation of nearly every other industry in the world.
Dutch semi-equipment maker, ASML, is working alongside NVIDIA on GPU design and cuLitho, and plans to integrate support for GPUs into all of their computational lithography software products. NVIDIA’s cuLitho software library and its integration into partner solutions and technology will help advance the industry into tinier transistors and wires than is now presently achievable while also boosting the time to market and the energy efficiency of the massive data centers that run 24/7 to drive the manufacturing process.
“The chip industry is the foundation of nearly every other industry in the world,” says Huang, founder and CEO of NVIDIA. “With lithography at the limits of physics, NVIDIA’s introduction of cuLitho and collaboration with our partners TSMC, ASML, and Synopsis allows fabs to increase throughput, reduce their carbon footprint and set the foundation for 2-nm and beyond.”
cuLitho is a software library that runs on GPUs involved in the lithography process and delivers a performance leap of up to 40x beyond current methods. This is the process of creating tiny patterns on a silicon wafer. The performance gain is massive, enabling just 500 NVIDIA DGX H100 systems to achieve the same work as 40,000 CPU systems, running all parts of the computational lithography process in parallel. In the immediate, cuLitho can help 3-5x the production of photomasks—the templates for a chips design—using 9x less power than current technology. And the longer-term impact of cuLitho from NVIDIA will boost higher density, greater chip yields from wafers, and AI-powered lithography.
Enabling Semiconductor Scaling
The cost of the computational time needed for the largest workloads in semiconductor manufacturing has been outpacing Moore’s Law. This is due to the fact that in the recent past chips have dramatically increased their transistor counts and must be manufactured under more demanding levels of precision.
Future nodes will require even more detailed calculations than today. Unfortunately, not all of the required calculations can fit within the currently available computational bandwidth of today’s current platforms, thereby slowing the pace of innovation in semiconductors.
NVIDIA cuLitho, as a software technology, will help accelerate and boost the computational bandwidth in today’s processes, leveraging newer technologies like H100, and removing bottlenecks in fab process changes that require OPC revision, in addition to making possible novel solutions and techniques like curvilinear masks, high NA EUV lithography, and sub-atomic photoresist modeling needed for the new technology nodes of the future.
Support From Industry Leaders
NVIDIA is working with its partners noted above on accelerating these technologies, and they are quite enthusiastic. Here are some notable comments on their support.
“The cuLitho team has made admirable progress on speeding up computational lithography by moving expensive operations to the GPU,” says C.C. Wie, CEO of Taiwan Semiconductor Manufacturing Company (TSMC). “This development opens up new possibilities for TSMC to deploy lithography solutions like inverse lithography technology and deep learning more broadly in chip manufacturing, making important contributions to the continuation of semiconductor scaling.”
“We are planning to integrate support for GPUs into all of our computational lithography software products,” says Peter Wennick, CEO of ASML. “Our collaboration with NVIDIA on GPUs and cuLitho should result in tremendous benefit to computational lithography and, therefore, to semiconductor scaling. This will be especially true in the era of high NA extreme ultraviolet lithography.”
“Computational lithography, specifically optical proximity correction, or OPC, is pushing the boundaries of compute workloads for the most advanced chips,” says Aart de Geus, chair and CEO of Synopsys. “By collaborating with our partner NVIDIA to run Synopsys OPC software on the cuLitho platform, we massively accelerate the performance from weeks to days! The team-up of our two leading companies continues to force amazing advances in the industry.”