Supports 4 new-gen Intel® Xeon® Scalable processors, with up to 60 cores and 120 threads per CPU, up to 350W TDP, and 3 Ultra Path Interconnect (UPI) links per CPU at up to 16GT/s per link, significantly improving its cost performance in virtualization scenarios. 

It supports 64 DDR5 ECC DIMMs (4,800MT/s, RDIMMs), delivering superior speed, high availability, and a memory capacity of up to 16TB. 

The all-flash configuration of 24 NVMe SSDs provides a high IOPS ten times higher than that of high-end enterprise-level SATA SSDs, boosting the storage performance. 

With 4 dual-width GPUs or 8 single-width GPUs, it provides ultra-high AI training performance to meet customers' computing needs in deep learning, inference, training, etc.

The I/O modular design supports 2 full-height modules that can be flexibly deployed to fit customer requirements on scalability and balance. 

The modular design of drive modules enables the server to support multiple flexible drive configurations, providing elastic, scalable storage capacity to meet different storage capacity and upgrade requirements. 

It supports up to 24 front 2.5-inch SAS/SATA/NVMe drives or 21 front 2.5-inch SAS/SATA drives and 4 NVMe drives, and supports 8 directly-connected SATA drives onboard. It supports 2 internal M.2 SSDs and 12 3.5-inch drives to store the data in layers. 

Supports 4 PCIe expansion slots onboard and 12 PCIe expansion slots through riser cards, as well as 4 dual-width GPUs or 8 single-width GPUs. 

Supports 2 hot-swap OCP 3.0 slots that can flexibly support 1/10/25/100/200Gb OCP 3.0 cards (multi-host); the OCP 3.0 cards can connect to 4 CPUs at the same time, improving the performance by 20%.

Supports hot-swap SAS/SATA/NVMe drives. With RAID cache and data protection enabled by the super-capacitor in case of power failures, SAS/SATA drives can be configured to RAID 0/1/1E/10/5/50/6/60 depending on the RAID card in use. 

The LEDs for fault diagnosis on the front and rear panels, the plug-in LCD module, and the ISBMC Web GUI indicate the statuses of key components and quickly lead engineers to failed (or failing) components, simplifying maintenance, speeding up troubleshooting, and enhancing system availability. 

The BMC management network port on the rear panel enables local O&M through ISBMC, improving O&M efficiency. 

Provides 4 hot-swap PSUs with N+M (M≤N) redundancy; provides 12 hot-swap fan modules with N+1 redundancy, improving overall system availability. 

The ISBMC monitors system parameters in real time and sends alerts in advance, enabling engineers to take corresponding measures in time to minimize system downtime.

Provides security-enhanced TPM 2.0 and TCM modules and secure measurement for trust, establishes a complete hardware and software trust chain, and discovers malicious intrusion and device replacement in time, so that worldwide customers can ensure the security and controllability of their information systems. 

Supports the trusted Intel PFR boot to effectively prevent network attacks against firmware and safeguard customer's data and assets. 

The security panel design supports the intrusion alert function. 

Online memory diagnosis helps service engineers quickly locate the failed DIMMs, improving maintenance efficiency. 

ISBMC4, an intelligent remote server management system developed independently by us, supports mainstream management specifications in the industry such as IPMI 2.0 and Redfish 1.8, greatly simplifying device deployment, management and maintenance of users, and providing higher operational reliability, easier maintenance and more accurate and comprehensive fault diagnosis capabilities.