NVIDIA 718-DG7050+P2CMI60
DGX H100 AI Server – 8x 80GB GPUs | AI Supercomputing Server System
The NVIDIA 718-DG7050+P2CMI60 DGX H100 is an elite AI Server platform, purpose-built to deliver unmatched compute density, bandwidth, and performance for modern AI workloads. With 8x NVIDIA H100 GPUs and a powerful dual-CPUarchitecture, it empowers enterprises and researchers to accelerate AI training and inference at scale.
Features
This next-generation DGX H100 system is built on NVIDIA cutting-edge Hopper architecture, equipped with 8x 80GB H100 Tensor Core GPUs connected via NVLink for a total of 640GB of GPU memory and lightning-fast interconnect. The system supports large-scale AI models, deep learning workloads, and complex simulations in HPC environments.
Advanced Options
Each unit integrates seamlessly with NVIDIA AI software stack, including Base Command and NGC containers, providing end-to-end workflow management. It’s ideal for AI research institutions, enterprise data centers, and scientific computing operations needing maximum scalability and reliability.
Product Key Features
-
-
Powered by 8× NVIDIA H100 80GB Tensor Core GPUs
-
Designed for large-scale AI training & inference workloads
-
PCIe Gen 5 and high-speed NVLink for ultra-fast GPU communication
-
Dual Intel Xeon CPUs with massive memory bandwidth
-
Comes with NVIDIA Base Command and enterprise-grade software stack
-
Fully integrated, tested, and optimized AI infrastructure solution
-
Reliability & Performance
Designed to deliver enterprise-grade stability, the DGX H100 features redundant components, high-efficiency cooling, and advanced workload balancing. With a total compute power exceeding 32 PFLOPS (FP8), it supports secure multi-user AI development environments and scales linearly in multi-node deployments. The system also includes full-stack enterprise support and optimized performance tuning tools.
Key Features :
-
-
8 × NVIDIA H100 Tensor Core GPUs (80GB each)
-
Total 640GB GPU memory with high-speed NVLink
-
2 × Intel Xeon Scalable CPUs for compute and orchestration
-
Up to 2TB of system memory
-
PCIe Gen 5.0, NVLink, NVSwitch interconnects
-
Optimized for FP8, FP16, BF16, TF32, and FP64 workloads
-
Supports NVIDIA AI Enterprise, Base Command, and NGC containers
-
Rack-mountable 6U form factor
-
Redundant power and cooling for enterprise uptime
-
Ideal for deep learning, generative AI, and scientific computing
-
Write a Review