This groundbreaking processor combines the high-performance NVIDIA Hopper™ GPU with the versatile NVIDIA Grace™ CPU. Designed to accelerate AI, high-performance computing (HPC), and data analytics workloads, this superchip delivers exceptional performance and energy efficiency for modern data centers.
Key Features
- Integrated Architecture: Combines a 72-core Arm Neoverse V2 Grace CPU with an H100 GPU, connected via a 900 GB/s NVLink-C2C interconnect for high-bandwidth, low-latency communication.
- High Memory Capacity: Supports up to 480 GB of LPDDR5X CPU memory and up to 96 GB of HBM3 GPU memory, providing ample capacity for large-scale AI models and datasets.
- Exceptional Bandwidth: Delivers up to 4 TB/s of GPU memory bandwidth, enabling rapid data processing and model training.
- Energy Efficiency: Designed for optimal power consumption, balancing performance and energy use in demanding workloads.
- Scalable Performance: Supports multi-node configurations with NVLink Switch System, allowing for seamless scaling across multiple superchips.
Target Applications
- Artificial Intelligence (AI): Accelerates training and inference of large language models, generative AI, and deep learning applications.
- High-Performance Computing (HPC): Enhances simulations and computations in scientific research, engineering, and climate modeling.
- Data Analytics: Speeds up data processing tasks, including real-time analytics and complex query execution.
- Enterprise Workloads: Optimizes performance for cloud services, virtualization, and large-scale enterprise applications.
Why Choose This Superchip?
This unified platform simplifies programming and accelerates a wide range of workloads. Its integrated architecture and high memory bandwidth provide the performance needed for today's most demanding applications, making it an ideal choice for organizations seeking to enhance their AI and HPC capabilities.
