AI × AEC · CURATED INSIGHT
NVIDIA Launches cuDSS Library for GPU-Accelerated Simulations
NVIDIA announced the general availability of cuDSS, a CUDA-X library designed to solve large sparse linear systems on GPUs. It targets simulation workflows across engineering and science to cut compute times.
NVIDIA released the cuDSS library in general availability to accelerate large sparse linear system solvers on GPUs. The library scales from single workstations to computing clusters, offering C, C++, Python, and Fortran interfaces. Engineering software vendors including Dassault Systèmes, Synopsys, Cadence, and Siemens are integrating cuDSS into their simulation tools.
Key takeaways
- cuDSS shifts large sparse linear system calculations to GPUs, providing C, C++, Python, and Fortran interfaces that scale to compute clusters.
- NVIDIA reports partners like Dassault Systèmes, Applied Materials, and COMSOL have integrated cuDSS into simulation tools to accelerate structural and physics solvers.
Source：NVIDIA Blog Read the original ↗
Prepared, checked and automatically published by AI; not individually reviewed by a human.