Lenovo ThinkSystem SR680a V3
Massive Computational Performance
The ThinkSystem SR680a V3 delivers groundbreaking performance with two 5th Gen Intel® Xeon® Scalable processors. Supporting 32x DDR5 memory at 5600MHz (up to 4TB), this system is engineered to maximize GPU performance with support for eight of the latest high-speed interconnected GPUs.
Ideal for AI innovations like LLM, machine learning, model training, and demanding HPC workloads, the SR680a V3 features NVIDIA NVLink technology with transfer rates up to 1.8TB/Sec.
Flexible Acceleration
Supporting industry-leading GPU platforms for AI, including NVIDIA's HGX™ H100/H200/B200 GPUs in an 8U frame compatible with standard 19″ racks. These Tensor Core GPU platforms provide unprecedented acceleration at every scale, powering the world's highest-performing elastic data centers for AI and HPC applications.
| Form Factor |
8U rack |
| Processor |
2x 5th Gen Intel® Xeon® Scalable processors, up to 350W |
| Memory |
Up to 4TB using 32x DDR5 DIMMs with maximum frequency at 5600MHz |
| GPU |
Supports 8x high-performance GPUs: 8x NVIDIA HGX™ H100/H200/B200 GPUs with NVLink interconnects at up to 1.8TB/s |
| I/O Expansion |
Up to 10x PCIe Gen5 x16 FHHL adapters (8x front connected to PCIe switch for GPU connectivity, 2x rear connected to CPU) |
| Storage |
Up to 16x 2.5″ hot-swap NVMe SSDs + 2x non-hot swap M.2 for boot (RAID via VROC) |
| Power |
Up to 8x hot-swap power supplies with full N+N redundancy |
| Cooling |
Air-cooled with N+1 hot-swap fan solution |
| Management |
XClarity Controller2 (XCC2) with advanced service-processor control, monitoring, and alerting functions |
| OS Support |
RHEL, Ubuntu, ESXi |