Shenzhen Gooxi Digital Intelligence Technology Co., Ltd.

PRODUCT CENTER

Flagship Dual-Socket Server, Strong Scalability with Massive Storage

Current location：

Home > Products > Product List > Intelligent Computing > AMD AI Server

AMD AI Server

Integrated Training and Inference

Powerful Computing Performance and Ultra-High Scalability

Supports 1 or 2 AMD EPYC Turin processors, compatible with AMD Genoa series, with TDP up to 500W; up to 24 DDR5 memory slots, supporting frequencies of 4800/5600/6400 MHz; CPU-GPU direct connection configuration, supporting 8 × 600W dual-width / 2.5-width / 3-width / 3.3-width / 3.5-width / 4.5-width GPUs; up to 12 × 3.5"/2.5" SATA/SAS/U.2 drives, meeting the needs of high-speed transmission and data storage.

Applicable Solutions

Artificial Intelligence ∣ Cloud Gaming ∣ Virtualization ∣ Big Data Analytics ∣ Digital Twin

Add to Compare

Product Overview Product Specifications Relevant Information

New Architecture, Extreme Performance

Supports 1 or 2 AMD EPYC Turin processors, compatible with AMD Genoa series, with TDP up to 500W; equipped with 24 DDR5 slots and 12 PCIe 5.0 slots for the latest technology, accelerating data transfer speed; GPU-CPU direct connection unleashes ultimate multi-GPU computing power.

Flexible Topology, Critical Applications

CPU-GPU supports passthrough, ensuring high-speed and stable operation of up to 8 × 600W dual-width, 2.5-width, 3-width, 3.3-width, 3.5-width, and 4.5-width GPUs; supports different CPU-GPU interconnect topologies such as common and performance, adaptable to a wide range of application scenarios.

Comprehensive Ecosystem, Stable and Reliable

Compatible with both international and domestic AI accelerator cards; redundant design of key system components and modular functional parts enhance product reliability and maintainability; supports intelligent remote management and smart remote fault diagnosis, ensuring efficient and stable 24/7 operation.

System Model	SYR8108G-D12R-G5
Processor	Supports 2 AMD EPYC Turin 9005 series processors, compatible with 9004 series processors, with TDP up to 500W
GPU Card	Passthrough 8-GPU AI Server (Fan Version): Supports 8 × 600W 3-width, 3.3-width, 3.5-width, or 4.5-width RTX 5090 fan GPUs or other 600W fan GPUs. The GPU adopts a front-rear 4+4 separated vertical airflow layout, supporting a maximum GPU width of 89 mm. Passthrough 8-GPU AI Server (Blower Version): Supports 8 × 600W 2-width or 2.5-width RTX 5090 blower GPUs or other 600W blower GPUs. The GPU adopts a rear 8-GPU side-by-side centralized airflow cooling layout, supporting a maximum GPU width of 50 mm.
Memory	24 DDR5 memory slots, supporting DDR5 4800/5600/6400 MHz memory
Storage	12 hot-swap 3.5"/2.5" SAS/SATA/NVMe drives
I/O Ports	Front Ports: 2 × USB 3.0, 1 × VGA Rear Ports: 1 × Serial, 2 × USB 3.0, 1 × VGA, 1 × RJ45 management port, 2 × Gigabit RJ45 network ports
PCIe Expansion	Passthrough 8-GPU AI Server (Fan Version): Supports 12 PCIe 5.0 expansion slots (including 1 OCP): 8 PCIe 5.0 4.5-width slots, 4 PCIe 5.0 x8 single-width slots Passthrough 8-GPU AI Server (Blower Version): Supports 12 PCIe 5.0 expansion slots (including 1 OCP): 8 PCIe 5.0 2.5-width slots, 4 PCIe 5.0 single-width slots
Power Supply	Supports 8 CRPS power modules with N+N / N+M redundancy (1600W/2200W/2700W/3200W)
Fan	Standard configuration: 6 × 6056 and 6 × 12038 fans
Management	1 dedicated RJ45 management port
Security	Supports TPM 2.0 module, chassis intrusion alert, and BMC/BIOS redundancy
Operating System	Microsoft Windows Server, Red Hat Enterprise Linux, SUSE Linux Enterprise Server, CentOS, Ubuntu, VMware ESXi, etc.
Server Size	364.3mm (H) x 446mm (W) x 910mm (D)
Working Temperature	5℃～35℃

Please note that the above specifications are subject to change without prior notice. (Gooxi reserves the right for the final interpretation)

Related Downloads

User Manual

Download Download Share

Product Specifications

Download Download Share

Product Image

Download Download Share