NVIDIA expands simulation software stack with general release of cuDSS

Traders Union
All News
Business & Wealth
Companies
NVIDIA cuDSS general release

Highlights

NVIDIA made cuDSS generally available within the CUDA-X software stack, targeting sparse system solving in engineering and scientific simulations with GPU acceleration.
Applied Materials reported up to 10x faster sparse solves in Ginestra, Dassault Systèmes saw 8–10x speed increases in Abaqus, and COMSOL's ZEISS SMT plasma simulation benchmark ran in six days and 14 hours, 4x faster with COMSOL Multiphysics and cuDSS.
Industrial vendors including Siemens, Cadence, and Synopsys are integrating cuDSS to accelerate simulation workflows in multiphysics, chip, and semiconductor design processes.

General release targets sparse solver bottlenecks

According to NVIDIA Newsroom, the company is making cuDSS generally available as part of its CUDA-X software stack, positioning the library to move large sparse system solving from CPUs to GPUs in engineering and scientific simulations.

That stage of computation is a recurring and math-intensive element in simulation workloads used to test designs before physical production. NVIDIA says growing design complexity in areas such as semiconductors, aircraft and industrial systems is extending simulation times and raising compute requirements, which can limit how many design options teams are able to evaluate.

The company says cuDSS helps applications reduce time to results and increase throughput, allowing users to work with larger and more realistic models, more accurate representations and a wider design space at lower cost. Developers can integrate the library through C, C++, Python and Fortran interfaces, and NVIDIA says the software scales from single workstations to larger clusters out of the box.

Chip and industrial software vendors cite speed gains

Several software and industrial technology groups are already integrating cuDSS into commercial tools, with reported performance gains across semiconductor, structural and multiphysics workloads.

Applied Materials says it accelerated sparse solves in its Ginestra simulation platform by up to 10 times against CPU-only approaches. Dassault Systèmes reports an 8 to 10 times speedup in Abaqus structural and electromagnetic solvers, while Siemens says it uses cuDSS in Simcenter STAR-CCM+ for fluid-structure interaction workflows.

COMSOL says ZEISS SMT recorded a measured fourfold speedup in an EUV lithography plasma simulation benchmark using COMSOL Multiphysics with cuDSS, completing the full run in six days and 14 hours. Cadence is using the library in Celsius software for heat, structural stress and warpage analysis in advanced chip and 3D chip designs, and Synopsys is integrating it into Sentaurus Device to speed simulations for advanced logic, memory and power semiconductors.

In our earlier article on Micron Technology’s weakening price action, we noted that the stock had broken below a rising wedge pattern, a technical setup that can signal a fading rally and increased downside risk. We also highlighted deteriorating momentum indicators and the key support and resistance zones investors were watching to gauge whether the pullback could extend or be negated.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.