With the advancement of science and technology, artificial intelligence (AITI) has become more and more important. Not only Internet giants such as Google, Facebook, and Amazon are competing for each other. Intel, AMD, and NVIDIA have gradually turned to the field of artificial intelligence in the PC market. In addition to the accelerated upgrade of processors and graphics cards to better match AI computing, Intel is launching a series of acquisitions, from cognitive computing company Saffron, machine learning new Nervana to computer vision chip company Movidius, Intel wants to do something. what?
Most of today's AI companies use GPUs for deep learning training and recognition. AMD and nVIDIA have launched a series of products based on this. NVIDIA has even launched a GPU-based deep learning dedicated supercomputer DGX-1, which is promoted in artificial intelligence. Intel is clearly much lower-key than its rivals AMD and NVIDIA.
The opponents are coming, and Intel, which has always been silent, finally announced the latest artificial intelligence strategy in mid-November, predicting that a series of AI solutions will be launched in the future. On November 17 and 30, Intel held artificial intelligence forums in San Francisco, China, and Beijing, China. (AI Day) evangelism, expressing Intel's determination to launch a series of AI revolutions from chip hardware and computing software to architecture.
Intel CEO Brian Krzanich called artificial intelligence a new turning point in human history and emphasized that Intel is the only company capable of accelerating the promotion of AI, and Intel is betting a lot of resources on artificial intelligence to promote business and society. development of. Ke Zaiqi even pointed out that all the acquisitions are to make artificial intelligence spread throughout Intel's products. The ambition to rejoin AI is in words.
AI portfolio from Nervana
In the development of Intel artificial intelligence, Nervana became the focus of attention first, and Nervana bought a machine learning startup company in August this year as a new product name, including Xeon (Chinese to Xeon) and Xeon Phi. (Chinese translation is Xeon Phi) Intel's existing CPU products are included.
Intel's Xeon dominates the server chip market, and official data emphasizes that the data center server currently running AI operations is up to 97%. However, in the near future, artificial intelligence deep learning technology has gradually attracted attention. The demand for parallel computing in computing models has made the GPU more important in the field of AI. Intel obviously does not want manufacturers such as nVIDIA to grab their own important markets.
Diane Bryant, Intel's Executive Vice President and General Manager of the Data Center Business Group, is even more vocal. By 2020, the Nervana platform trained deep learning model time, which is 100 times shorter than current GPU training time, accelerating the pace of deep learning. At this Beijing AI Days, Intel China Research Institute Dean Song Jiqiang pointed out that Nervana's ability to identify images and speech in the case of large enough data can surpass humans.
As to why Nervana can accelerate the effectiveness of deep learning, Song Jiqiang explained that in the field of machine learning, researchers are expecting performance to increase with the increase of data, which can be achieved by increasing the number of processors at the beginning, but when dealing with The device is increased to a certain extent, performance growth is stagnant, and it is faced with I/O problems, bottlenecks caused by communication between processors and memory access. Nervana technology can more efficiently define memory bandwidth and computational density to match the increased data volume, while allowing multiple nodes to support large model training in parallel, with near-linear performance extensions.
In these activities, Intel also revealed the product blueprint after integrating Nervana technology. In addition to the first chip designed for deep learning, Lake Crest will be tested in the first half of 2017. Nervana will also be equipped with new Xeon and Xeon Phi processors. 2017 Qifa.
The first Xeon processor integrated with Nervana technology, Knights Crest, was developed into a core technology-like neural network for deep learning and supports high-bandwidth interconnects. Xeon Phi, a 64-core, 72-core super multi-core coprocessor, can greatly improve computing performance with software. For example, Song Jiqiang, Caffe AlexNet did not change much when running on Xeon Phi, but after a six-month optimization by software department researchers, the performance improved by 400 times.
And Intel's next-generation Nervana Xeon Phi processor, the Knight Mill, will be four times more powerful than the existing Xeon Phi processor Knight Landing. The Altera FPGA products acquired recently have also been integrated into Nervana's AI product line, emphasizing that it can be used in some specific areas for specialized customized architectures, or that the framework is not yet determined to be easy to experiment and modify.
Steel pole
Steel poles are commonly used to carry several types of electric power lines, distribution lines and lighting system. Distribution lines carry power from local substations to customers. They generally carry voltages from 4.6 to 33kV for distances up to 30 miles, and include transformers to step the voltage down from the primary voltage to the lower secondary voltage used by the customer. A service drop carries this lower voltage to the customer's premises.
Steel Pole,Steel Street Light Poles,Galvanized Steel Pole,Steel Pole Tower
Yixing Steel Pole International Trading Co., Ltd , https://www.yx-steelpole.com