NVIDIA DGX™ Systems with Exeton

Datacenter

NVIDIA DGX H100

8x NVIDIA H100 GPUs, 32 PetaFLOPS, Dual Xeon Processors, and Robust Networking in One Powerhouse Server. Includes NVIDIA AI Software and 3-Year Warranty.

$495,000

Add to Cart

High-Performance Server with NVIDIA H100 GPUs

Powerful Computing, Robust Networking, and Pre-loaded AI Softwar

✓ Equipped with 8x NVIDIA H100 Tensor Core GPUs SXM5
✓ GPU memory totals 640GB
✓ Achieves 32 petaFLOPS FP8 performance
✓ Incorporates 4x NVIDIA® NVSwitch™
✓ System power usage peaks at ~10.2kW
✓ Employs Dual 56-core 4th Gen Intel® Xeon® Scalable processors
✓ Provides 2TB of system memory
✓ Offers robust networking, including 4x OSFP ports, NVIDIA ConnectX-7 VPI, and options for 400 Gb/s InfiniBand or 200 Gb/s Ethernet

The World’s Proven Choice for Enterprise AI

Drives Accelerated Performance

Maximize business innovation and optimization with NVIDIA DGX™ H100. As a vital component of the DGX platform and the newest iteration of NVIDIA's renowned DGX systems, DGX H100 serves as the cornerstone for NVIDIA DGX SuperPOD™, leveraging the revolutionary capabilities of the NVIDIA H100 Tensor Core GPU to drive accelerated performance.

DOWNLOAD DATASHEET

Explore DGX H100

Powering Accelerated Business Innovation

8 NVIDIA H100 GPUs, Providing 640 Gigabytes of Total GPU Memory
Each GPU Equipped with 18 NVIDIA® NVLink® Connections, Delivering 900 Gigabytes per Second of Bidirectional GPU-to-GPU Bandwidth
4 NVIDIA NVSwitches™, Delivering an Impressive 7.2 Terabytes per Second of Bidirectional GPU-to-GPU Bandwidth, a 1.5X Increase Compared to the Previous Generation
10 NVIDIA ConnectX®-7 Network Interfaces, Each Capable of 400 Gigabits-Per-Second, Providing a Peak Bidirectional Network Bandwidth of 1 Terabyte per Second
Dual Intel Xeon Platinum 8480C Processors, Boasting a Total of 112 Cores and Accompanied by 2 Terabytes of System Memory, Ensuring Exceptional Performance for Demanding AI Workloads

Gallery

Your cart is empty

Add item to it now