Acceleration For Every Workload

Featuring NVIDIA H100 Tensor Core GPU

The GPU For Accelerated Computing

The NVIDIA® H100 Tensor Core GPU, based on NVIDIA Hopper™ architecture, delivers unprecedented acceleration to power the world's highest-performing elastic data centers for AI, data analytics, and high-performance computing (HPC) applications. Accelerated servers with H100 deliver unparalleled compute power-along with 3 terabytes per second (TB/s) of memory bandwidth per GPU and scalability with NVLink and NVSwitch™- you can now tackle data analytics with high performance and scale to support massive datasets.

Transformer Engine

The Transformer Engine uses software and Hopper Tensor Core technology designed to accelerate training. Hopper Tensor Cores can apply mixed FP8 and FP16 precisions to dramatically accelerate AI calculations for transformers.

NVLink Switch System

The NVLink Switch System enables the scaling of multi-GPU input/output (IO) across multiple servers at 900 gigabytes per second (GB/s) bidirectional per GPU, over 7X the bandwidth of PCIe Gen5.

Multi-Instance GPU (MIG)

The Hopper architecture's second-gen MIG supports multi-tenant, multi-user configurations in virtualized environments, securely partitioning the GPU into isolated, right-size instances.

DPX Instructions

Hopper's DPX instructions accelerate dynamic programming algorithms by 40X compared to CPUs and 7X compared to NVIDIA Ampere architecture GPUs.

Featuring NVIDIA H200 Tensor Core GPU

Most powerful GPU for AI & HPC

The NVIDIA HGX H200 showcases NVIDIA H200 Tensor Core GPU based on NVIDIA Hopper^™ architecture, is equipped with cutting-edge memory capabilities designed to effortlessly manage extensive datasets for generative AI and high-performance computing tasks. It is the first GPU with HBM3e, the H200's larger and faster memory fuels the acceleration of generative AI and large language models (LLMs) while advancing scientific computing for HPC workloads.

Unlock Insights

An AI inference accelerator must deliver the highest throughput at the lowest TCO when deployed at scale for a massive user base. H200 doubles inference performance when handling LLMs such as Llama2 70B when compared to H100.

Faster AI Training

NVIDIA H200's Transformer Engine with FP8 precision and fourth-generation Tensor Cores speed up fine-tuning by 5.5X over A100 GPUs . This performance increase allows enterprises and AI practitioners to quickly optimize large language models such as GPT-3 175B.

Supercharge HPC

For memory-intensive HPC applications like simulations, scientific research, and artificial intelligence, H200's higher memory bandwidth ensures that data can be accessed and manipulated efficiently, leading to up to a 110X faster time to results.

Reduce Energy

Energy efficiency and TCO reach new levels with H200 GPU. For at-scale deployments, H200 systems provide 5X more energy savings and 4X better cost of ownership savings over the NVIDIA Ampere architecture generation.

Featuring NVIDIA Grace Hopper Superchip

Breakthrough GPU-CPU Integration for AI acceleration

The NVIDIA Grace Hopper^™ architecture integrates the groundbreaking performance of the NVIDIA Hopper^™ GPU with the versatility of the NVIDIA Grace^™ CPU, creating a singular superchip connected via the high-bandwidth, memory-coherent NVIDIA^® NVLink^® Chip-2-Chip (C2C) interconnect. This fusion offers unparalleled efficiency, with the NVIDIA Grace CPU delivering twice the performance per watt of conventional x86-64 platforms, making it the fastest Arm^® data center CPU, enabling scientists and researchers to reach unprecedented solutions for the world's most complex problems.

Power and Efficiency

The NVIDIA Grace CPU combines 72 Neoverse V2 Armv9 cores with up to 480GB of server-class LPDDR5X memory with ECC. This design strikes the optimal balance of bandwidth, energy efficiency, capacity, and cost.

Performance and Speed

A new Transformer Engine enables Hopper H100 Tensor Core GPU to deliver up to 9X faster AI training and up to 30X faster AI inference compared to the prior GPU generation.

Alleviate bottlenecks

The NVIDIA NVLink-C2C interconnect enables a high-bandwidth of 900 GB/s direct connection between two NVIDIA Grace CPUs, crucial for creating the powerful NVIDIA Grace CPU Superchip with up to 144 Arm Neoverse V2 cores.

Full Support

The NVIDIA Grace Hopper Superchip is supported by the full NVIDIA software stack, including the NVIDIA HPC, NVIDIA AI, and NVIDIA Omniverse^™ platforms.