What changed
NVIDIA has launched cuDSS in general availability, a CUDA-X library designed to shift critical parts of engineering and scientific simulations to GPUs. This tool accelerates solving large sparse systems, which are math-heavy steps that often limit the number of designs teams can test.
What this means in practice
The release of cuDSS enables applications to reduce time to results and increase simulation throughput. This allows end users to explore larger, more realistic models and more accurate representations at a reduced cost. Industry leaders such as Applied Materials, COMSOL, Cadence, Dassault Systèmes, Siemens, and Synopsys are already integrating cuDSS into their tools to transform workflows for their end customers. For example, Applied Materials reported a 10x speedup in sparse solves using cuDSS, while COMSOL achieved a 4x speedup in an EUV lithography plasma simulation benchmark.
Limitations and availability
cuDSS can scale from individual workstations to large computing clusters as simulation workloads grow in size and complexity. Developers can integrate cuDSS into existing engineering, scientific, and optimization software offerings using C, C++, Python, and Fortran interfaces. The supplied announcement does not specify exact hardware requirements or licensing conditions for cuDSS.
