The most popular nvidiaturingt4 cloud GPU continue

2022-07-26
  • Detail

NVIDIA Turing T4 cloud GPU continues to accelerate strongly

Internet companies such as search, social media and shopping malls are early users of T4 and the largest end customer group. The first batch of enterprises in China to start using T4 to expand and improve the horizontal expansion of workload include Baidu, Tencent, JD and iFLYTEK. China's leading computer manufacturers will also launch a series of T4 based servers, including Inspur, Lenovo, dawning, Inspur business machines and Xinhua

NVIDIA announced last week that T4 had played an important role in 57 independent server designs of the world's leading computer manufacturers only two months after its launch. In addition, customers of Yunxuan cuckoo cloud platform can use T4

nvidia vice president and general manager of accelerated computing ianbuck said: "it is not surprising that T4 is adopted in China so fast. We have never launched such a GPU before to provide the public cloud and private cloud with the required comprehensive performance and energy efficiency, and can run computing intensive workloads on a large scale and more economically. Moreover, the 'scale' of the Chinese market is unparalleled, and we predict that T4 will be very popular."

t4 is based on the new NVIDIA Turing TM architecture, adopts multi-precision Turing tensor core and new rtcore, and combines with the accelerated container software stack to provide unprecedented performance

Chinese server companies adopting T4 include Langchao, Huawei, Lenovo, Shuguang, Langchao business machine and Xinhua III. These new systems include:

● Inspur nf5280m4/nf5280m5/nf5288m5/nf5468m5

● Lenovo thinksystemsr630/sr650

● Huawei g2500/2288hv5/5288v5/g530v5/g560v5

● dawning x580-g30/x745-g30/x780-g30/x780-h30

● Inspur commercial machine fp525 95g2

● Xinhua 3 uniserverr4900g3

the above systems are expected to be supplied before the end of the year

flexible design and amazing performance

t4 is designed to meet the unique needs of horizontally expanding public cloud and Enterprise Cloud environments, maximize throughput, utilization and user concurrency, and help customers effectively cope with the explosive growth of users and data

the compact 70 watt t4gpu is roughly the size of a bar of chocolate candy. It can be flexibly adapted to the design of standard servers or super large servers for any open computing project. The server design can range from a single t4gpu to 20 GPUs in a single node

The multi precision function of

t4 can support four types of AI workloads with different precision levels, helping to achieve breakthrough AI performance. Fp32 precision can provide 8.1tflops, fp16 precision can provide 65tflops, int8 precision can provide 130tops, and int4 precision can provide 260tops. For AI inference workloads, a single server with two t4gpus can replace up to 54 CPU servers. For AI training, a single server equipped with two T4S can replace nine dual socket CPU servers

support from technology leaders

based on the excellent characteristics of nvidiagpu in ease of use and energy efficiency, Inspur and NVIDIA have jointly assisted many AI commercial companies around the world to deploy AI online reasoning platforms with high performance, low power consumption, high bandwidth and low delay, which greatly promoted the realization of AI business value. With the release of NVIDIA's new generation Turing architecture T4 and tenso and the extensive use of rrt5, Inspur will bring the ultimate AI acceleration experience to global AI users at the first time by virtue of its strong product innovation ability, and work with NVIDIA to promote the sustainable development of global AI ecology-- Pengzhen, vice president of Inspur Group, Lenovo has accumulated considerable technology and experience in the field of artificial intelligence and HPC, and has an absolute advantage in the field of HPC. In the data center, GPU application performance explosion has promoted the leapfrog development of HPC and cloud computing. Combined with the latest nvidiat4gpu, Lenovo will introduce more powerful products and solutions to the market, maintain its leading position in the HPC field, and help users make breakthroughs in data science applications in data mining, data retrieval, big data analysis, artificial intelligence and other fields-- Wu Bin, chinabu of Lenovo Group's data center business and general manager of strategic operations in China

Huawei's intelligent computing is committed to making computing ubiquitous and intelligence omnipresent. Huawei G-Series servers support NVIDIA Tesla T4 for the first time, and together provide customers with competitive AI solutions. Facing the future, Huawei will always maintain an open ecology and accelerate the popularization of AI with industry partners. As the leader in AI, NVIDIA will continue to cooperate with Huawei to build an omnipresent intelligent society-- Qiu long, President of Huawei's intelligent computing business department, dawning, as the leader of China's computing technology, has always believed that "intelligence changes the world, and computing determines the future". One of the foundations of AI industrialization is computing technology. Tesla T4 GPU based on NVIDIA turning architecture is the best choice for AI training and reasoning calculation facing the horizontal expansion of HPC. The newly released xmachine series AI dedicated servers and the latest t4gpu, which serve the deployment of AI reasoning applications, will become an important technical guarantee for the industrialization of AI technology-- Li Bin, general manager and assistant to the president of Shuguang HPC products division, Xinhua San has always been committed to the continuous exploration of new technologies and applications, providing leading infrastructure services for users' digital transformation. In the field of artificial intelligence, Xinhua San has always maintained close cooperation with NVIDIA, the industry leader, to provide domestic users with the latest AI technology and scenario based transmission stability application solutions. NVIDIA recently released a new generation of gputeslat4. Xinhua III r4900g3 immediately supported this GPU with excellent performance, helping users greatly improve the AI training and reasoning performance of voice, video, search engines and other AI applications-- Chenzhenkuan, vice president of Xinhua Group 3, has created a very large application system with the technology of cloud computing and hybrid heterogeneous computing, which has brought huge computing support to the development of artificial intelligence. Facing the new business load of artificial intelligence, Inspur business machine is expanding the platform horizontally with open power. Under the openpower cooperation ecological environment, Inspur business machine and NVIDIA are jointly building the next generation reasoning hardware and software platform to build a new standard for efficient reasoning load efficiency. Power9 and Tesla T4 are used to stimulate great potential, promote the development of various new applications, and provide an important step for the transition of the whole system-- Huleijun, general manager of Inspur business machines Co., Ltd., "JD cloud is committed to providing users with one-stop and all-round cloud computing solutions, and has profound technical accumulation in cloud computing and AI technology. JD cloud and NVIDIA are jointly based on the most cutting-edge IT technology → steel for prestressed concrete test methods (gb/t21839 ⑵ 008) → ISO 15630 ⑶: 2002 Test methods for steel for prestressed concrete and steel for prestressed concrete Part 3: Steel for prestressing (ISO 15630 ⑶: in 2002, GPU cloud products were developed to provide the performance and efficiency required by large-scale computing intensive workloads for today's public and private clouds. It is expected that the cooperation between the two sides will bring users more convenient and efficient experience and services. "-- He Gang, vice president of JD group and head of JD cloud product development department, iFLYTEK, as a leading artificial intelligence company in China, has always maintained the platform + track AI strategy It holds a leading position in the fields of speech recognition, machinetranslation and image recognition. In this process, iFLYTEK and NVIDIA have maintained very close technical communication and product cooperation. As the first domestic customer of this kind of graphene material lat4 developed by Professor chenyongsheng's team, after detailed testing, we have fully felt the excellent AI computing power of Turing architecture. We believe that we will work together to provide a better AI experience for AI users by taking advantage of the advantages of the two companies-- Zhangzhijiang, President of Cloud Computing Research Institute of iFLYTEK Co., Ltd.,

about nvidia

nvidia (NASDAQ: NVDA) invented GPU in 1999, which stimulated the growth of PC game market, redefined modern computer graphics card, and innovated parallel computing. Recently, by using GPU as a brain that can sense and understand the world, such as computers, robots and even autonomous driving, GPU deep learning has ignited a new era of computing - modern artificial intelligence

Copyright © 2011 JIN SHI