Altos BrainSphere™ R785 F6 AI Server
Altos BrainSphere™ R785 F6 is a high-performance GPU server platform designed based on the NVIDIA MGX architecture, purpose-built to address demanding workloads such as Large Language Models (LLMs), High-Performance Computing (HPC), model training and optimization, and Omniverse applications.
Powered by dual 5th Gen AMD® EPYC® scalable processors and high-capacity DDR5 memory, the R785 F6 supports up to eight NVIDIA RTX PRO™ 6000 or H200™ GPU cards. With its highly flexible and modular 4U rackmount design, the R785 F6 delivers powerful computing performance to handle large-scale models and massive datasets with ease.
Its flexible resource allocation and scalable architecture support the complete AI workflow, from AI model training to real-time inference, helping enterprises accelerate application deployment and quickly build scalable and efficient AI solutions.
Powerful AI and HPC Computing Performance
The R785 F6 supports up to eight NVIDIA RTX PRO™ 6000 or NVIDIA H200™ GPUs. With NVLink™ and NVSwitch™ technologies, the system delivers up to 1.8 TB/s of GPU-to-GPU bandwidth, enabling high-speed multi-GPU communication and large-scale parallel computing.
The platform provides up to 1.1 TB of total GPU memory and 38.4 TB/s of memory bandwidth, delivering powerful performance for generative AI and large language model (LLM) training while maintaining excellent power efficiency and compute density.
(Calculated based on eight NVIDIA H200 GPUs, each with 141 GB memory and 4.8 TB/s memory bandwidth.)
Flexible Expansion with Future-Ready PCIe Gen6 Support
The R785 F6 offers up to 13 PCIe Gen5 slots, 8 × 2.5" NVMe drive bays, and 2 × M.2 NVMe slots, enabling flexible expansion for high-speed networking and various accelerator cards to support AI, HPC, and cloud infrastructure deployments.
The system also supports an optional 8 × 400G NVIDIA ConnectX-8 networking module, enabling high-speed data exchange between servers in large-scale clusters. With this configuration, selected PCIe slots are upgraded to PCIe Gen6, preparing the platform for next-generation GPUs and high-performance expansion devices while extending system lifecycle.
(With 8 × 400G ConnectX-8 installed: 8 × PCIe Gen6 + 1 × PCIe Gen5 slots)
Intelligent Management and AI Workflow Optimization
The R785 F6 is equipped with the ASPEED® AST2600 BMC, enabling comprehensive remote management capabilities. Combined with the Altos aiWorks platform, it provides intelligent resource orchestration and enhanced platform security. This intelligent management framework simplifies AI deployment, improves resource utilization, and maximizes the return on AI infrastructure investments.
Altos BrainSphere™ R785 F6 AI Server
- Processors:Dual 5th Gen AMD® EPYC® 9005 series processors, TDP up to 500W
- Power Supply:3+1 3200W 80 PLUS Titanium redundant PSUs
| Processors | Dual 5th Gen AMD® EPYC® 9005 series processors, TDP up to 500W |
| Number of Processor | 2 |
| Memory | 24 DIMM slots, Up to 6TB ECC DDR5 6400 MHz ECC RDIMM |
| I/O Expansion | 8 x FHHL dual-slot PCIe Gen5 x16 slots 5 x FHHL single-slot PCIe Gen5 x16 slots |
| NVIDIA ConnectX®-8 Version (optional) 8 x FHHL dual-slot PCIe Gen6 x16 slots 1 x FHHL single-slot PCIe Gen5 x16 slots |
|
| Network | 2 x 10GbE(NCSI supported) 1 x 1GbE Management Ports |
| NVIDIA ConnectX®-8 Version (optional) 8 x 400 GbE QSFP112 ports |
|
| Storage | Up to 8 x 2.5 |
| Rear I/O | 1x USB 3.0, 1 x Mini Display Port, 1 x 1 Gb Management LAN port |
| Power Supply | 3+1 3200W 80 PLUS Titanium redundant PSUs |
| OS Support | Microsoft® Windows Server, Red Hat Enterprise Linux Server |
| Dimension | 800 (D) x 438.5 (W) x 175 (H) mm |
- Extended warranty services may vary by country. Please contact Altos for more information.