Accelerating Scientific Workflows With the NVIDIA Grace Hopper Platform

NVIDIA Grace Hopper Superchips are a scale-up architecture ideal for scientific computing workflows involving CPUs and GPUs. Grace Hopper realizes NVIDIA NVLink C2C, which enables coherent memory between the Grace CPU and the Hopper H100 GPU with 7x the bandwidth of PCIe. This removes the conceptual CPU/GPU memory divide and lowers barriers when accelerating applications. With more code executing on GPUs, workload performance becomes increasingly susceptible to non-GPU limiters like data movement and CPU performance (Amdahl’s Law). Grace Hopper restores the required balance. This session presents HPC performance results with a technical deep-dive into the specific features of Grace Hopper that accelerate each workload. We discuss how Grace Hopper and the accompanying software stack create a platform that increases developer productivity, accelerates existing applications, and facilitates new standard programming models.