Inspur · NF5688M7
Inspur NF5688M7
Inspur NF5688M7 is a 6U flagship AI training server with NVIDIA HGX 8-GPU module, designed for trillion-parameter LLM training at hyperscale.
- 2 4th/5th Gen Intel Xeon
- 8 Card NVIDIA HGX H100/H200/H20
- NVLink+NVSwitch Fully interconnected
- 6U Chassis·Up to 128TB NVMe
Published model specifications describe available options, not the delivered configuration.
Configurations & prices
Contact our team to confirm the configuration and quotation.
Listed prices are indicative; confirm tax, stock, compatibility, lead time and service in a written quotation.
Specifications & selection notes
Technical specifications
| Model scope | This table covers NF5688-M7-A0-R0-00 from the NF5688G7 specification table, not the AMD version |
|---|---|
| Chassis | 6U rack chassis |
| Processors | 2 fourth- / fifth-generation Intel Xeon Scalable processors; up to 350W TDP |
| Memory | 32 DDR5 DIMM slots, up to 5600MT/s |
| GPU module | 1 NVIDIA HGX-Hopper-8GPU module; this is not an individual PCIe GPU-card chassis |
| Storage | Up to 24 x 2.5-inch SSDs, including up to 16 NVMe; optional 2 internal M.2 NVMe / SATA drives |
Configuration and expansion
| PCIe expansion | 10 PCIe 5.0 x16 slots; 1 may be replaced by 2 physical x16 slots operating at x8, not counted simultaneously |
|---|---|
| Networking | 1 optional OCP 3.0 NIC with NCSI; BlueField-3, CX7 and smart-NIC options |
| Power supplies | 2 x 12V 3200W and 6 x 54V 2700W Titanium CRPS power supplies with N+N redundancy |
Enquire about NF5688M7
Provide quantity, workload and component requirements to confirm the configuration and quotation.