Overview
NVIDIA DGX H100 Deep Learning Console integrates 8x H100 SXM5 (80GB each, 640GB total) with NVLink fabric, InfiniBand 400Gb/s, 2TB DDR5 system RAM, and NVIDIA's DGX OS in NVIDIA's fully integrated and validated system. At $375,000, this is NVIDIA's reference configuration for organizations who want a validated end-to-end package with NVIDIA support rather than an OEM-assembled H100 cluster.
Key Features
- 8x NVIDIA H100 SXM5 — 640GB HBM2e with NVLink/NVSwitch
- InfiniBand 400 Gb/s (OSFP) network connectivity
- 2TB DDR5 ECC system memory
- NVIDIA DGX OS — pre-configured deep learning stack
- Full NVIDIA support + Base Command Manager
Ideal For
Enterprises and research institutions who want a fully validated, NVIDIA-supported H100 training system with end-to-end warranty and DGX OS stack, avoiding the integration complexity of OEM-assembled equivalent systems.
This vendor publishes no specification table. The following is derived from the product listing.
- CPU vendor
- Intel
- CPU series
- Xeon Scalable
- CPU count
- 2
- GPU / accelerator
- H100
- GPU count
- 8
- GPU memory
- 80GB
- System memory
- 2048GB
- Memory type
- DDR5 ECC
- Storage
- NVMe
- Form factor
- rack-server
Prices may vary. Verify on vendor site.