As models explode in complexity, accelerated computing and energy efficiency are becoming critical to meet the demands of AI. The NVIDIA Grace™ CPU is a groundbreaking Arm® CPU with uncompromising performance and efficiency. It can be tightly coupled with a GPU to supercharge accelerated computing or deployed as a powerful, efficient standalone CPU. The NVIDIA Grace CPU is the foundation of next-generation data centers and can be used in diverse configurations for different data center needs.
NVIDIA GB200 NVL2
The NVIDIA GB200 NVL2 platform brings the new era of computing to every data center, delivering unparalleled performance for mainstream large language model (LLM) inference, vector database search, and data processing.
The NVIDIA GB200 Grace™ Blackwell Superchip combines two NVIDIA Blackwell Tensor Core GPUs and a Grace CPU. It can scale up to the GB200 NVL72—a massive 72-GPU system connected by NVIDIA® NVLink®—to deliver 30X faster real-time inference for large language models (LLMs).
The NVIDIA GB200 NVL2 platform brings the new era of computing to every data center, delivering unparalleled performance for mainstream large language model (LLM) inference, vector database search, and data processing.
The NVIDIA Grace Hopper™ Superchip combines the Grace and Hopper architectures using NVLink-C2C to deliver a CPU+GPU coherent memory model for accelerated AI and high-performance computing (HPC) applications.
The NVIDIA Grace CPU Superchip uses the NVLink-C2C technology to deliver 144 Arm Neoverse V2 cores and 1 terabyte per second (TB/s) of memory bandwidth.
Learn how Grace CPUs are powering the latest large-memory supercomputers.
Computer Industry Joins NVIDIA to Build AI Factories and Data Centers for the Next Industrial Revolution
At Computex 2024, the world’s top computer manufacturers joined NVIDIA to unveil the latest NVIDIA Blackwell-powered systems, including the GB200 NVL2, to lead the next industrial revolution.
NVIDIA GB200 NVL72 for Real-Time Trillion-Parameter Inference
The NVIDIA GB200 NVL72 Grace Blackwell Superchip is the architecture for the next generation of AI, supercharging real-time trillion-parameter language models.
Creating Accelerated Data Centers Faster With NVIDIA MGX
Learn how QCT and Supermicro are adopting modular designs to quickly and cost-effectively build multiple data center configurations for a wide range of AI, high-performance computing (HPC), and 5G applications.
NVIDIA Grace Hopper Ignites New Era of AI Supercomputing
From Climate and Weather to Scientific Exploration, Switzerland’s Alps Supercomputer, France’s EXA1-HE Supercomputer and Others to Deliver 200 Exaflops of AI for Groundbreaking Research Using Energy-Efficient Grace-Based Systems
Solving the largest AI and HPC problems requires high-capacity and high-bandwidth memory (HBM). The NVIDIA NVLink-C2C delivers 900GB/s of bidirectional bandwidth between the NVIDIA Grace CPU and NVIDIA GPUs. The connection provides a unified, cache-coherent memory address space that combines system and HBM GPU memory for simplified programmability. This coherent, high-bandwidth connection between CPU and GPUs is key to accelerating solutions to tomorrow’s most complex problems.
Tap Into High-Bandwidth CPU Memory With LPDDR5X
NVIDIA Grace is the first server CPU to use LPDDR5X memory with server-class reliability through mechanisms like error-correcting code (ECC). This enables it to meet the demands of the data center while delivering high-memory bandwidth and up to 10X better energy efficiency compared to today’s server memory. The LPDDR5X design strikes the optimal balance of bandwidth, energy efficiency, capacity, and cost for large-scale AI, HPC, and cloud workloads.
Boost Performance and Efficiency With Arm Neoverse V2 Cores and NVIDIA Scalable Coherency Fabric
As the parallel compute capabilities of GPUs continue to advance, workloads can still be gated by serial tasks that run on the CPU. A fast and efficient CPU is a critical component of system design to enable maximum workload acceleration. The NVIDIA Grace CPU integrates Arm Neoverse V2 cores with the NVIDIA Scalable Coherency Fabric to deliver high performance in a power-efficient design, making it easier for scientists and researchers to do their life’s work.
Supercharge Generative AI With HBM GPU Memory
Generative AI is memory- and compute-intensive. The NVIDIA GB200 Superchip uses 380GB of HBM memory, delivering over 4.5X the GPU memory bandwidth of the NVIDIA H100 Tensor Core GPU. The high-bandwidth memory in Grace Blackwell is connected to CPU memory over NVLink-C2C to provide almost 860GB of fast-access memory to the GPU. This delivers the memory capacity and bandwidth needed to handle the world’s most complex generative AI and accelerated computing workloads.