Lenovo H200 SR680a V3: Purpose-Built for NVIDIA H200
The Lenovo H200 (ThinkSystem SR680a V3) is purpose-built for NVIDIA H200 GPUs, featuring the latest SXM5 form factor with 141GB HBM3e memory per GPU. This 8-way GPU server delivers exceptional performance for memory-bound AI workloads and large-scale model training. Leveraging Lenovo's Neptune liquid-cooling technology, the H200 achieves remarkable density and efficiency.
Technical Specifications
| Component | Detailed Specification |
|---|---|
| GPU | 8x NVIDIA H200 141GB SXM5 (HBM3e memory, NVLink4) |
| GPU Memory | 141GB HBM3e per GPU (1,128GB aggregate) |
| CPU | 2x Intel Xeon Platinum 8568Y+ (48 Cores each, 2.3GHz base, 350W TDP) |
| System Memory | 1.28TB TruDDR5-5600 (32x 64GB 2Rx4 RDIMMs) |
| System Drive | 1x 960GB NVMe M.2 (boot/OS) |
| Storage Drives | 4x 3.2TB NVMe PCIe 4.0 (U.3 7450 MAX) |
| InfiniBand | 8x ConnectX-7 400G InfiniBand ports (OSFP) |
| Networking | 1x Intel X710-T2L 10GBase-T 2-Port Ethernet |
| Management | Lenovo XClarity Controller |
| Power | 6x 3000W Titanium PSUs (redundant) |
| Form Factor | 4U/5U rack-optimized chassis |
| Cooling | Lenovo Neptune liquid-cooling ready (hybrid design) |
Key Features
- 8x NVIDIA H200 141GB SXM5 GPUs with 141GB HBM3e each
- Intel Xeon Platinum 8568Y+ 48-core processors (96 total cores)
- 1.28TB DDR5-5600 memory capacity for CPU-side workloads
- 400G InfiniBand fabric for GPU-to-GPU communication across nodes
- Lenovo Neptune liquid-cooling ready design for maximum efficiency
- ThinkSystem reliability with 24/7 enterprise support
Lenovo Neptune Cooling Advantage
The H200 leverages Lenovo's innovative Neptune technology, offering direct-to-node liquid cooling for GPUs and CPUs, reduced fan power consumption by up to 40%, higher density deployments with more nodes per rack, quieter operation for data center environments, and improved PUE with lower operational costs.
Ideal Workloads
The Lenovo H200 excels in large language model training for LLaMA and GPT-class models, memory-intensive HPC simulations, generative AI and diffusion models, enterprise AI development and production, research and academic supercomputing, and climate modeling and weather simulation.
Request a Quote
Contact Omnixon Global for competitive pricing on the Lenovo H200 SR680a V3. Enterprise AI infrastructure with Lenovo reliability and Neptune cooling innovation.