Nvidia announced on July 26, 2026, at the DAC 2026 conference, the expansion of its CUDA-X software suite with new solver libraries: cuISS for iterative sparse solvers, cuDSS for direct sparse solvers, and cuEST for quantum chemistry calculations. Performance gains are significant: up to 11x faster for computational engineering tools using cuDSS, and up to 20x for computational lithography among clients like Samsung. On August 20, 2026, Cloudera integrated Nvidia’s cuDF library to accelerate Apache Spark workloads, delivering up to 4x faster processing without requiring code changes. The CUDA-X suite now spans between 400 and 900 libraries covering AI, high-performance computing, physics, and engineering. These developments reinforce Nvidia’s strategy to embed GPU acceleration directly into existing engineering and AI workflows.
Source: Read the original article
