Deep thinking series AI server covers a variety of CPU platforms, and supports pre-installation of OS, drivers, DL framework and common DL libraries according to customer requirements, saving you a lot of pre-debugging time and being ready to use.

A simple task, if you want to form a clear idea in AI’s mind, needs a lot of experiments and practice. From the perspective of AI training, artificial intelligence will learn all the necessary knowledge in a short time through a single reinforcement learning model.
The higher the calculation efficiency, the better the trained model. Adding a lot of experimental data and training frequency can increase the training frequency of the model. However, when GPU trains large data sets, it will also have the problem of "low training efficiency". However, assuming that this model exceeds the training level of human beings, the cost of training may be too high.
So, how to train artificial intelligence efficiently and at low cost?
IW4213-4GR It is Siteng Heli specially designed forArtificial intelligence training, artificial intelligence project deployment (OEM), high-performance computingThe standard rack-mounted 4 GPU card high-performance server is built with Intel? Scalable processor.
Powerful heterogeneous computing power: Support Intel Xeon? Scalable series processors, Intel C621 chipset and NVIDIA data center-class GPU.
Rich scalability: Each CPU is equipped with 8 memory slots, with a total of 16 memory slots, which can reach 4T of large memory, support Intel Aoteng data center-level persistent memory, and provide flexible and powerful memory configuration options; It comes standard with 8 hot-swappable SATA hard disk bits and dual 10 Gigabit Ethernet high-speed interconnection, which can be used as a high-performance computing node of GPU computing cluster. 4 full-height and full-length PCI-E 3.0 X16 slots, which can be configured with 4 high-performance GPU cards.
Excellent product designStandard 4U rack structure, supporting the installation of rear suction fan, unique and excellent heat dissipation structure design, integrating high density, high performance and high reliability. It has a very balanced performance and is suitable for personal AI model research or small-scale HPC computing.
| Product technical specifications

IW4221-8GR
Support two new generation Intel Xeon ?Icelake scalable processors, TDP 270W.

Eight 350W GPU cards can be supported on PCIe Gen4 platform with double bandwidth.
Support one-button topology switching to meet different AI application requirements
Widely used in Internet AI public cloud, enterprise AI cloud platform, intelligent security, video codec and other scenarios.
| Product technical specifications

IW4220-10GR
Support for 2 new generation Intel Xeon IceLake scalable processors | Support for 10 GPU cards.

Deep thinking IW4220-10GR GPU server has the characteristics of high performance, high density and strong scalability, and supports two new generation Intel Xeon Ice Lake scalable processors, TDP 270W.
Designed with Intel C621A chipset, it supports 8TB of large memory, comes standard with 16 SATA hard disk bits and 8 NVMe hard disk bits, and is connected with dual 10 Gigabit Ethernet at high speed, which can be used as a high-density and high-performance node platform for GPU computing cluster.
It is the best choice for large-scale artificial intelligence training and HPC high-density computing.
| Product technical specifications

IW4222-8GR
Multi-scene adaptation server supporting two third-generation Intel? Xeon? Icelake extensible processors.

Sten Heli Deep Thinking Series IW4222-8GR supports two third-generation Intel? Xeon? Icelake series scalable processors, TDP 270W.Provide a powerful X86 processing platform for reasoning and training applications.
PCIe Gen4 platform with double bandwidthSupports 8 dual-width GPU cards., designed to provide multi-directional performance support for high-density GPU computing.
Support high-speed network, storage and other IO expansion, 8 U.2 hard disks.Provide high-speed transmission and data storage capabilities for high-end computing platforms.
At the same time, the operation and maintenance efficiency is further optimized, and a professional management platform can realize the functions of identification, monitoring, fan speed control support, fault alarm and so on for various mainstream GPUs in the market.
| Product technical specifications

Sten Heli is not just making products. We also have a professional team to provide after-sales service for you. No one will wait for you forever, but someone will always wait for you. The 7 * 24-hour after-sales service hotline can be dialed by users all over the country, and the after-sales engineer on duty will answer your questions within 2 hours, and we will initially locate the problems within 4 hours. We have after-sales offices in Beijing, Tianjin, Nanjing, Wuhan, Chengdu, Shenzhen, Xi ‘an and Baotou.