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 > Modular Computing > AMD Milan 5-GPU > AMD AI Server

AMD AI Server

Excellent Scalability with Flexible Configuration Options

Gooxi AMD dual-socket accelerated computing server features onboard 10 PCIe expansion slots, 16 DIMM slots, 1 M.2 slot, 2 Gigabit Ethernet ports, and 1 RJ45 management port. It boasts excellent expandability and flexible configuration options, making it suitable for a wide range of applications, including artificial intelligence, virtualization, cloud computing, big data processing, distributed storage, enterprise markets, and telecommunications services

Applicable solution

Rendering, cloud gaming, digital twin, artificial intelligence

Add to Compare

Product Overview Product Specifications Relevant Information

2 high-performance AMD EPYC 7003/7002/7001 processors
10 PCIe high-speed expansion slots to provide rich expansion performance
5 full-height, double-width high-performance GPUs for the needs of high computing power
16 RECC DDR4 memory slots, the maximum frequency supports 2933MHz

Modular Design

Standard E-ATX form factor, compatible with any chassis configuration. The fans, power supply, and hard drives all feature a modular design, enhancing system reliability and heat dissipation performance, while facilitating easy maintenance

High Performance

Supports two AMD EPYC 7003 (Milan) series processors, compatible with the 7002 (Rome) and 7001 (Naples) series. It features a 12Gb SAS backplane and supports up to 5 full-height, double-width high-performance GPU cards. With 16 RECC DDR4 memory slots, supporting a maximum frequency of 2933MHz, and native NVMe cache interface on the motherboard, it meets the demands of Artificial Intelligence, Virtualization, Cloud Computing, Big Data Processing, Distributed Storage, Enterprise Markets, or Telecom Services applications

GPU Computing

Supports up to 5 full-height, double-width high-performance GPU cards with direct PCIe expansion and low latency, catering to the needs of Artificial Intelligence, Virtualization, Cloud Computing, Big Data Processing, Distributed Storage, Enterprise Markets, or Telecom Services applications

Intelligent PSU and Temperature Control

Utilizes platinum-level 1+1 redundant power supplies with hot-swappable capabilities to ensure system stability. It employs intelligent temperature-controlled fan modules with hot-swappable features and excellent shock resistance design

System Model	ASR4105G-D12R
Processor	Support 2 AMD EPYC 7003; Compatible with AMD EPYC 7002 and 7001 series
Chipset	/
GPU Card	Supports 5 RTX3090 cards and 4 A800 double-width, full-height, full-length cards
Memory	16 DDR4 memory slots, support 2400/2666/2933 MHz DDR4 memory
Storage Controller	1 M.2 (PCIe x 4) interface 2 MiniSAS HD interfaces
Storage	Front: 12x 3.5"/2.5" SAS/SATA hot swap hard drives Rear: optional support 1x 2x2.5" hard disk module
Network	2 onboard 1GbE
I/O Ports	Front: 2 USB2.0 Rear: 1 VGA, 1 COM port, 2 USB3.0, 1 management network port, 2 RJ45 network ports
PCIe Expansion	Support 6 PCIe x8 and 4 PCIe x16
GPU Quantity	5 full-length full-height dual-width GPU cards
Power Supply	Support 800W, 1200W, 1600W, 2200W hot-swap redundant power (adapt according to the actual power)
Fan	Standard 8x 8038 hot-swap fans
Management	1 RJ45 management network port IPMI2.0
Security	Support TPM module
Operating System	Microsoft Windows Server、Red Hat Enterprise Linux、 SUSE Linux Enterprise Server、CentOS、Ubuntu
Server Size	695 x 433 x 176.5mm (D x W x H)
Working Temperature	Operating temperature: 5ºC - 35ºC (indoor) Storage temperature: -40ºC - 70ºC

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 specification

Download Download Share

Product brochure

Download Download Share