标签： 花坊上海后花园论坛

The finals of Beijing Three-Man Basketball Open ended, and the two teams advanced to the national finals.

BEIJING, Beijing, September 18 (Liu Xingchen) On the evening of the 17th, the final of FIBA Open 3×3 Yanjing Beer Cup Beijing Three-Man Basketball Open ended at the National Speed Skating Hall. In the battle for the championship and runner-up, V10 Junda beat Beijing Jianyou overtime to win the Beijing championship. In October, the two teams will join hands to participate in the national finals held in Nanchang, Jiangxi Province, and compete with teams from other seven provinces (cities) for the national championship.

Since its launch on August 19th, FIBA Open 3×3 Yanjing Beer Cup Beijing Three-Man Basketball Open has taken the form of "5+1", which has gone through five sub-races in Shougang Park, Olympic Sports Center, Yizhuang, Changying and Wukesong, and the Beijing Finals in the National Speed Skating Hall. The 20 teams in the finals are composed of the top four in each of the five sub-races. According to reports, the competition attracted nearly 100 teams and more than 500 players to sign up through the official FIBA basketball APP.

Photo courtesy of the organizer

On the day of the finals, before the game started, the young players from Youken Basketball Club played a juvenile version of 3×3 basketball. Jiao Jian, president of the Youth Committee of the Beijing Basketball Association, said: "The popularization and promotion of youth basketball is very important. The 3×3 basketball is relatively simple to set up, and two teams and six people can play it, making it easier for children to get started and experience the joy of basketball more."

In this finals, the lion king eliminated all the winners in several subjects and took the lead in entering the semi-finals. Since then, Beijing Jianyou, TOJOY and V10 Junda have successively eliminated Yanjing U8, Refreshing Fast Star and Baokuang Bath, and locked in the semi-finals.

Later, Beijing Jianyou and V10 Junda won the final meeting from the semi-finals, which was also a strong dialogue between the champion of the Olympic Sports Center and the champion of Wukesong Station.

In the final, the two teams tied 18 in the regular time. In overtime, V10 Junda took the lead in winning the game with 2 points, and was crowned the champion of Beijing finals.

Photo courtesy of the organizer

Meng Qianghua, member of the party group and deputy director of Beijing Sports Bureau, said: "Beijing Sports Bureau is committed to building a nationwide fitness service system. The FIBA Open 3×3 Beijing Three-person Basketball Open is an important measure in the process of building a sports event with the whole crowd and the whole sports cycle. When the masses personally participate in and immerse themselves in the competition, they can feel the happiness brought by sports and play a good role in demonstrating and driving the implementation of the national fitness policy. "

The official of Yanjing Beer, the title sponsor of this competition, said that he would continue to practice social responsibility and make contributions to the development of sports in China.

In addition, Fang Liang, general manager of FIBA China, Sun Guohua, president of Beijing Basketball Association, Min Lulei, head coach of Beijing Basketball Club, and others also attended the game. (End)

Siteng Heli Thinking Series ｜ Four High Performance AI Servers

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.

How to create a safe, intelligent and efficient terminal new logistics？

Terminal new logistics refers to the last logistics link of goods from distribution center to consumers. It is the last link of the whole supply chain, which is very important for consumers’ shopping experience and the quality of logistics service of e-commerce enterprises. Therefore, it is very important to create a safe, intelligent and efficient terminal new logistics. Here are some suggestions:

First, the use of intelligent technology

The terminal new logistics needs to use intelligent technology to improve efficiency and reduce errors. Technologies such as Internet of Things, big data and artificial intelligence can be used to realize the intelligentization of terminal new logistics. For example, intelligent sensing devices can monitor the location, quantity and transportation status of goods in real time, optimize distribution routes and strategies through big data analysis, reduce human intervention in logistics links and improve logistics efficiency.

Second, establish a standardized process

The terminal new logistics needs to establish a standardized process. By formulating a series of norms and processes, the whole distribution process will be more standardized, reduce errors and mistakes, and improve the quality and stability of logistics services.

Third, strengthen safety management

Terminal new logistics needs to strengthen safety management and take safety measures to ensure the safety of logistics services. For example, carry out strict safety inspection and supervision on transport vehicles and goods to prevent theft and accidents during transportation. At the same time, provide necessary safety equipment and training for distributors to improve their safety awareness and coping ability.

Fourth, optimize the distribution network

The terminal new logistics needs to optimize the distribution network, reduce the logistics time and cost and improve the distribution efficiency by optimizing the route and distribution strategy. Intelligent algorithm and big data analysis can be used to analyze the distribution demand in different regions and different time periods, and make more scientific and reasonable distribution strategies.

V. Building a trust system

Terminal new logistics needs to establish a reliable trust system, including trust in distributors and logistics enterprises and trust in consumers. We can improve the credibility of logistics enterprises and distributors and enhance consumers’ trust and loyalty by establishing credit rating system and adopting real-name registration system and insurance system.

In a word, to build a safe, intelligent and efficient terminal new logistics requires comprehensive measures such as using intelligent technology, establishing standardized processes, strengthening safety management, optimizing distribution network and establishing trust system. Only through continuous innovation and improvement of service quality can we meet consumers’ demand for fast, safe and convenient terminal logistics services, and also enable logistics enterprises to gain greater market share and commercial profits in the fierce market competition. At the same time, the government should also increase its support for terminal new logistics, and create a good policy and business environment for the development of terminal new logistics by introducing policies and strengthening supervision.