Configuration
Tailored to your project
Rack Servers
TFNEW S768G5 6U 8x GPU AI Supercomputing Server
Model: S768G5 · SKU: 0102010
S768G5: an enterprise GPU server supporting up to 2x 4th/5th Gen Intel® Xeon® Scalable processors (TDP 350W), with 32 DDR5 DIMM slots (up to 8 TB) and up to 8 high-performance GPU cards (600W per card), a 6U rackmount AI supercomputing server.
Reference configuration
Integration
Fits your environment
Support
Selection & delivery
WHY THIS PRODUCT
S768G5 GPU Server — Enterprise Server Platform
The S768G5 supports up to 2x 4th/5th Gen Intel® Xeon® Scalable processors (TDP up to 350W), with 32 DDR5 DIMM slots up to 8 TB and up to 8 high-performance GPU cards.
High-Density Computing
Powered by 2x 4th/5th Gen Xeon Scalable processors to meet enterprise-grade workload requirements.
Flexible Storage Expansion
12 3.5-inch drives and 32 DDR5 DIMM slots for elastic scaling of business data.
Enterprise Reliability
N+N redundant power and efficient thermal design ensure stable 7x24 operation.
Specifications
| Model | S768G5 |
|---|---|
| Chassis | 6U rackmount AI supercomputing server |
| GPU Module Scheme | 8-card passthrough module / 8-card Switch module |
| Processor | Up to 2x 4th/5th Gen Intel® Xeon® Scalable processors, up to 64 cores per CPU, HBM support, TDP up to 350W |
| Memory | 32 DDR5 DIMM slots, up to 5600 MT/s, RDIMM/LRDIMM support, up to 8 TB; CXL 1.1 support |
| Storage | 12x 3.5-inch drives; 2x 2280/22110 SATA or 1x 2280/22110 NVMe M.2 SSD |
| PCIe Expansion | Passthrough: up to 11 standard PCIe expansion slots; Switch: up to 27 PCIe expansion slots, up to 27 x16 standard PCIe expansion slots, 1x OCP3.0 slot |
| GPU Support | Up to 8 high-performance GPU cards, up to 600W per card |
| Network | 1x OCP3.0 network card, optional 1/10/25/50/100/200/400GE PCIe standard network cards |
| Power | Supports 8x 2000W/2400W/2600W/2700W/3200W Platinum power supplies, N+N redundancy/N+1 hot-swap redundancy |
| Fans | 2 fan boards; lower 2U fan board supports 4x 8056 or 8038 fans N+1 hot-swap redundancy; middle fan board supports 10x 8056 or 8038 fans N+1 hot-swap redundancy |
| Dimensions | 263.7 mm (H) x 447.4 mm (W) x 898 mm (D) (without ears) |
| Operating Environment | Temperature 0°C ~ 35°C, humidity 8% ~ 90% non-condensing |
| SKU | 0102010 |
| Warranty | - |
| Lead time | - |
Applications
Built for Real Business Workloads
Use this platform as part of a complete enterprise AI infrastructure solution.
AI Training and Inference
A high-performance compute node supporting private model training, inference and concurrent workloads.
Contact our team →Private Compute Infrastructure
Serves as the compute foundation of enterprise data centers for virtualization and hyper-converged scenarios.
Contact our team →HPC and Data-Intensive Workloads
High-throughput I/O and low-latency interconnect for scientific computing, engineering simulation and data analytics.
Contact our team →Deployment
A Clear Deployment Path
Move from requirement discovery to a validated configuration and supported delivery.
Requirements and Specification Review
Confirm key parameters such as compute selection, power consumption, storage and facility environment.
Configuration and Compatibility Validation
Validate PCIe topology, thermal airflow and full-load system stability.
Delivery and Technical Support
Provide factory test reports, environment pre-installation and full lifecycle technical support.
Resources
Project Resources
Request the current documents for the exact configuration your team is evaluating.
Product Datasheet
Review detailed technical specifications, dimensions and interface definitions.
Contact our team →Installation and Configuration Guide
Best practices for rack mounting, cabling and data center deployment.
Contact our team →Technical Consulting
Discuss customized solutions with TF NEW senior hardware engineers.
Contact our team →REQUEST A QUOTE
Get a Configuration and Quote for the S768G5 GPU Server
Tell us about your required configuration scale and deployment timeline, and we will provide professional advice and a formal quotation.