FG4268-F4 AI Server
FG4268-F4 is an artificial intelligence computing server developed based on Intel's latest EagleStream platform with powerful computing power and ultra-high elastic expansion capabilities. It supports CPU-GPU direct connection and CPU-GPU switch topology to meet the application requirements of various AI business scenarios. The efficient use of computing power meets the needs of energy conservation and carbon reduction.
Product type: 8-card 12-slot AI server
GPU support: Supports up to 8 600W double-width GPUs
System size: 176.5mm(H)x444mm(W)x850mm(D)
Processor: Supports 1 or 2 4th/5th Generation Intel® Xeon® Scalable Processors, TDP up to 350W
Memory: 32 DDR5 memory slots, support DDR54400/4800/5600MHz memory
Internal storage interface: 2 Slimline4x interfaces (SATA), 2 2280/22110 size M.2 (SATA/PCIe3.0x1)
External hard drives: 12 hot-swappable 3.5/2.5-inch SAS/SATA/NVMe hard drives
External ports: Front ports: 2xUSB3.0, 1x VGA
Rear ports: 1 serial port, 2 USB3.0 interfaces, 1 VGA, 1 RJ45 management port (optional 2 10GRJ45 business network ports)
*PCIe expansion specifications: Supports up to 13 PCIe expansion slots (OCP3.0 card optional PCIe5.0X8 or X16 and one PCIe5.0 standard single-width slot is mutually exclusive)
Operating system: Microsoft Windows Server, Red Hat Enterprise Linux, SUSE Linux Enterprise Server, CentOS, Ubuntu, Vmware EXSi, etc.
Security: Support TPM2.0 module, cover opening alarm, BMC\BIOS redundancy
Power supply: 3+1 or 2+2 redundancy (2000W/2200W/2600W/3200W)
System fan: standard 12 6056 hot-swappable N+1 redundant fans
IPMI compatibility: Supports Redfish, SNMP, IPMI 2.0 and other standard interfaces
Management port: 1 dedicated RJ45 management network port
Working temperature: 5℃～35℃