NVIDIA’s HPC SDK v25.7 Enhances Ocean Modeling with Unified Memory
NVIDIA's latest HPC SDK v25.7 introduces unified memory programming, automating data movement between CPUs and GPUs to streamline high-performance computing workflows. The update significantly benefits scientific applications, particularly ocean modeling, by reducing manual data management and boosting developer productivity.
The Barcelona Supercomputing Center has Leveraged this technology for the Nucleus for European Modelling of the Ocean (NEMO), demonstrating tangible performance gains. NVIDIA's GH200 Grace Hopper Superchip and GB200 NVL72 systems, already deployed at major supercomputing centers, further amplify these advantages through high-bandwidth NVLink-C2C interconnects.