"While Nvidia's GPUs have become synonymous with AI in recent years, they're by no means limited to machine learning workloads. Nvidia has developed numerous frameworks to speed up jobs that traditionally ran on CPUs - from semiconductor manufacturing and physics simulations to electronic design automation - using its chips. "Something that would take weeks could now happen in hours," Nvidia CEO Jensen Huang said during a press conference on Monday."
"Synopsys has been one of the most enthusiastic adopters of the tech. At GTC this spring, the company touted up to a 30x speedup for circuit sims running in its PrimeSIM suite and up to a 20x speedup across its Proteus computational lithography suite using Nvidia's latest generation of Blackwell accelerators. Synopsys also utilizes Nvidia's NeMo frameworks and NIMs - containerized models containing everything necessary to run them at scale - to power its AI knowledge assistant."
Nvidia is purchasing $2 billion of Synopsys common stock at $414.79 per share as part of a long-term partnership. Nvidia GPUs are applied beyond machine learning to accelerate semiconductor manufacturing, physics simulations, and electronic design automation. Synopsys reported up to 30x speedups in PrimeSIM circuit simulations and up to 20x speedups in Proteus computational lithography using Nvidia's Blackwell accelerators. Synopsys uses Nvidia NeMo frameworks and NIMs for an AI knowledge assistant. The companies will expand Nvidia hardware and CUDA-X library support to more applications and co-develop digital twins for industries including semiconductor design, robotics, aerospace, automotive, and energy.
Read at Theregister
Unable to calculate read time
Collection
[
|
...
]