- All products
- RACKMOUNT SERVERS
- GPU & AI Servers
- Godì 1.8SP5-D HPR PXL
- GPU & AI Servers
Mona Mainstream Platform
Double Width GPU
Powered by dual AMD EPYC™ processors, it can host up to 8 double-width GPUs, supporting up to 600W per GPU such as NVIDIA RTX Pro 6000 or H200,
delivering exceptional acceleration for AI training and inference.
Its architecture provides very high GPU density, making it ideal for companies deploying large-scale AI models and real-time inference workloads.
The system also features front-accessible hot-swap drive bays, allowing easy maintenance and continuous operation.
For high-speed connectivity and efficient data transfer, the server integrates up to two network interface cards at the rear, ensuring seamless networking capabilities for GPU-accelerated infrastructures.


AI Server
8x NVIDIA® RTX PRO Blackwell or H200 GPUs
The Godì 1.8SP5-D is a high-performance GPU server designed for AI and inference workloads in modern data centers. Powered by dual AMD EPYC™ processors, it can host up to 8 double-width GPUs, supporting up to 600W per GPU such as NVIDIA RTX Pro 6000 or H200, delivering exceptional acceleration for AI training and inference.
Its architecture provides very high GPU density, making it ideal for companies deploying large-scale AI models and real-time inference workloads. The system also features front-accessible hot-swap drive bays, allowing easy maintenance and continuous operation.
For high-speed connectivity and efficient data transfer, the server integrates up to two network interface cards at the rear, ensuring seamless networking capabilities for GPU-accelerated infrastructures.
Image courtesy of NVIDIA. 
NVIDIA® RTX PRO 6000 Server Edition
The NVIDIA RTX Pro 6000 Server Edition delivers cutting-edge performance with the Blackwell architecture, 96 GB ECC GDDR7 memory, and over 24,000 CUDA cores. Designed for AI, HPC, visualization, and large-scale analytics, it combines massive compute power, advanced Tensor and RT cores, and enterprise features like MIG and secure boot, making it ideal for modern data centers.
NVIDIA® H200 NVL GPUs 141GB
With AI models now exceeding hundreds of billions of parameters, the NVIDIA H200 NVL platform combines the Hopper architecture, 141GB of HBM3e memory, and high-speed NVLink interconnects between GPUs.
Designed for modern data centers, it helps accelerate AI and inference workloads. It also simplifies the deployment and scaling of AI models and LLMs.
Image courtesy of NVIDIA. 
AMD EPYC™ 9005 Series Processors
Built for the most demanding workloads, the AMD EPYC™ 9005 (Zen 5 / Zen 5c architecture) redefines server computing with up to 192 cores and 384 threads — delivering unmatched compute density for Cloud, HPC, AI, analytics, and mission-critical applications.
Boosting up to 5.0 GHz, it leverages 12 channels of DDR5 ECC memory at up to 6,000 MT/s, ensuring massive bandwidth for the most data-intensive workloads. With PCIe® 5.0 support up to 128 lanes from CPU and CXL 2.0 for memory expansion and accelerator support, it's ready for next-gen GPUs and high-speed NVMe storage.
Available in single- or dual-socket (1P/2P) configurations on the scalable Socket SP5 platform, EPYC 9005 is engineered for flexibility and long-term investment protection. AVX-512 vector extensions further accelerate HPC and AI workload performance.

|
System |
|
|
Form Factor |
19" 4U |
|
Dimension |
867 x 438 x 176.5 mm (34.1'' x 17.2'' x 6.9'') |
|
Storage |
|
|
Internal type per node |
1x M-key (PCIe3.0 x4 or SATA 6Gb/s), supports 22110/2280 form factor [CPU0] 1x M-key (PCIe3.0 x4 or SATA 6Gb/s), supports 22110/2280 form factor [CPU1] |
|
External type per node |
4x Hot-swap 2.5" NVMe (PCIe5.0 x4) drive bays 20x Hot-swap 2.5" SATA/SAS* drive bays *Additional RAID/HBA card required to support the 20 x 2.5" SATA/SAS drive bays |
|
GPU |
|
|
Motherboard |
|
|
CPU |
Dual AMD EPYC™ 9xx5/9xx4 Turin, Genoa, Bergamo and Genoa-X with AMD 3D V-Cache™ Technology Series Processor families, up to 192-core, 384 threads per processor, cTDP up to 500W
Note: 2 CPUs must be installed. |
|
Socket |
Dual Socket SP5 (LGA 6096) |
|
Expansions slots |
8x FHFL dual-slot PCIe5.0 x16 [CPU] 1x FHHL PCIe5.0 x8 * For GPU TDP up to 600W (RTX PRO 6000 and H200 NVL4 supported) |
|
BMC |
Integrated Aspeed® AST2600 |
|
TPM |
1x header (13-pin, SPI) |
|
Memory |
|
|
Total Slots |
24 (12-channel per socket, 1DPC) |
|
DIMM Capacity |
RDIMM: 128GB (2Rx4), 3.0TB total RDIMM-3DS: 512GB (2Rx4), 12.0TB total |
|
Memory Type |
AMD EPYC™ 9005: RDIMM: Up to 6400 MT/s AMD EPYC™ 9004: RDIMM: Up to 4800 MT/s |
|
Network |
|
|
Onboard |
1x 1GbE Management Port dedicated to the IPMI 2x 1GbE Intel® i350 |
|
NIC option |
1x FHHL PCIe5.0 x8 |
|
I/O |
|
|
Front I/O Ports |
1x UID button 1x DB15 (VGA) 2x Type-A (USB3.2 Gen1) 2x RJ45 (1GbE) by Intel i350, shares with Front Panel 1x dedicated IPMI, shares with Front Panel |
|
Management Solution |
|
|
Remote Management |
KVM module, Remote Update (OoB), Platform Firmware Resilience |
|
Software |
WebGUI, IPMI 2.0 and RESTful APIs (Redfish) |
|
Other Specifications |
|
|
Power Supply |
3+1 redundant 3200W Titanium CRPS power supply |
|
Operating Environment |
Operating temperature: 10°C ~ 35°C / Non-operating temperature: -40°C ~ 70°C Non-operating humidity: 20% ~ 90% (non-cond.ensing) |
Skus : | Front Storage | Rear Expension Slots |
BRB-GI1-408-SP5D-R1101 | Up to 4x U.2 NVMe (PCIe G5) + 20 Hot-swapable 2.5 inch (depending on the configuration | Front: 1x PCIe x8 (Gen5 x16 link), FFHL |
BRB-GI1-408-SP5D-R1102 | 16 hot-swap 2.5" NVMe (PCIe5.0 x4) drive bays or 24 Hot-swap 2.5" SATA/SAS *drive bays *Additional RAID/HBA card required | Front :2x PCIe x16 FHHL (PCIe5.0) Rear: 10 FHFL dual-slot PCIe5.0 x16 (with dual PEX89104 PCIe Switch) 1 FHFL single-slot PCIe5.0 x16 (CPU direct link) |