Intel is developing a new chip that can be learned like the human brain and it will make robots more autonomous. The test chip, code-named "Loihi," is still in the prototype stage, but it also scares many people who have just recovered from the Western World. When artificial intelligence is affecting everyone in a fun way, there is no second way to implement it except for the use of chips.
As a result, the position of the chip has increased again and competition has intensified. In this new era of computing infrastructure and the layout of future ecological strategies, global IT giants, Internet companies, and startup companies have all been involved, and China has also accelerated its entry.
The self-learning neuron chip Loihi that Intel is developing causes the industry's attention because they are trying to "copy the human brain." The neural network of the human brain relies on the information of nerve impulses and synapses. Loihi simulates the function of the human brain. It integrates more than 130,000 neurons and 130 million synapses. It uses the method of replicating nerves to perform calculations and reflections without the need of tradition. The method of training can achieve self-evolution and decision-making. Compared to chips that are commonly used to train artificial intelligence systems, the Loihi chip's energy efficiency has increased by a factor of 1,000.
According to Dr. Michael Mayberry, Intel’s Global Vice President and Director of Intel Labs Research, the potential for self-learning chips is limitless compared to machine learning. It can improve automotive and industrial applications as well as personal robots. Such devices can learn and execute. The most complex cognitive tasks include any application that benefits from autonomous operation and continuous learning in an unstructured environment. For example, identify the motion of a car or a bicycle, explain key heart rhythms, detect abnormal conditions, prevent cyber attacks, and arrange music. In the first half of 2018, Intel will share Loihi test chips with many famous universities and research institutions.
Amir Cossasi, chief technology officer of Intel's artificial intelligence product division, was the co-founder and chief technology officer of Nervana, which was acquired by Intel. He believes that the focus of artificial intelligence is not to rebuild the human brain, but to transcend the human brain: "A lot of people think that we are creating a brain. We want to go beyond this and hope to create a new kind of artificial intelligence that can understand data statistics in all areas of the enterprise, healthcare, etc., and these data are essentially indistinguishable from the real world. different."
Only at present, artificial intelligence is still in its infancy, and it is not easy to use machines to realize human brain functions.
Decide on future competition
The Morningstar report from a third-party agency shows that in 2021, the total value of the artificial intelligence chip market will reach 20 billion U.S. dollars. TechNavio predicts that by 2021, the global compound annual growth rate of artificial intelligence chips will exceed 54%. Holding their own technical path, giants in the chip field have entered the war where the smoke has not yet fully ignited.
NVIDIA, which was once tepid in the semiconductor industry and leaned on the graphics processor (GPU) market, performed outstandingly in this round of artificial intelligence chip wars. Establishing the "CUDA" (Unified Computing Device Architecture) project around 2007 laid the foundation for Nvidia The first-mover advantage in the field of intelligence. During the process, Nvidia underwent significant cost pressures. The project spent around US$500 million annually and accounted for one-sixth of the company’s total annual revenue at the time. This year, Nvidia set a record high of 20 historical closings. The stock price has risen by about 69%, outperforming the gains of funds tracking the semiconductor sector.
Liu Jun, General Manager of Artificial Intelligence and High Performance Products Division of Inspur Group, told the reporter of Science and Technology Daily: “Before and after 2014, those companies that are well-known in the field of artificial intelligence began to look for GPU experts to help them integrate the existing DNNs (Deep Neural Networks). ) Migrate from CPU to GPU."
The 2nd quarter report of NVIDIA's fiscal year 2017 shows that the growth of artificial intelligence-related business is bright, and GPU business revenue has increased by 59% year-on-year.
Intel, which overwhelmingly dominated the PC central processor market for many years, has fallen behind Nvidia in the chip field. In recent years, Intel has been working hard to change this situation and accelerate the deployment of artificial intelligence through acquisitions. In 2015, Intel acquired Altera, a Silicon Valley chip maker, with the largest amount of history. Intel CEO Ke Zaiqi once stated clearly that FPGA is an important core of Intel's future development. In August 2016, Intel acquired Nervana, a US startup company. Its deep learning engine, the Engine, is 10 times faster than the GPU. On October 17th, Intel announced that it will launch the first chip of the first artificial intelligence chip Nervana (NNP) series, which has high hopes for this industry's first chip for neural network processing, hoping to help them achieve 2020 artificial intelligence. The goal of 100 times improvement in performance.
Qualcomm has a large number of terminal card capacity, the main force in the direction of mobile phones and cars. Jeff Gelhart, vice president of Engineering Technology at Qualcomm, stated that the importance of processing data on the terminal side is reflected in the fact that it can better protect the privacy of users and help solve the need for real-time decision-making such as unmanned driving. Use network bandwidth. Mobile chip manufacturers and terminal vendors are supporters of this view. Apple iPhoneX uses artificial intelligence chip "A11 biological neural network engine" to support face recognition and other artificial intelligence programs. Huawei Mate10 equipped with the "world's first mobile phone artificial intelligence chip" Unicorn 970 has been listed in China.
Whose chip will be the preferred artificial intelligence in the future? Song Jiqiang, president of the Intel China Research Institute, told the reporter of the Science and Technology Daily: “The chip needs to be determined by application, and the current industry has no
Chinese chips short boards need to break
The huge impact of small chips on the global economy and technology in the future cannot be ignored. However, the vast majority of chips on the Chinese market have still been imported from abroad. Fortunately, with AI chips as the representative, China is becoming a global The most potential part of the smart layout.
On October 24th, Zhong Ke Dawning announced the successful development of the world's first artificial intelligence server "Phaneron" equipped with Cambrian AI chips. This is the first achievement since the strategic cooperation between China Dawning and Cambrian Technology started last year. . The president of Dawning Li Jun told the reporter of the Science and Technology Daily: “The Cambrian era entrusted the Chinese artificial intelligence industry with the hope of changing lanes and overtaking. The Cambrian age will play an important role in the smart age.†It may have required three Nvidia chip cards. Now it only takes a Cambrian period to deal with the consumption of the entire data center. In 2016, it was ranked as one of the 15 "World's Internet Leading Scientific and Technological Achievements" by the World Internet Conference.
Lijun said: “Artificial intelligence has become a new “national sport†in China. The series of advantages such as strong calculation ability, large data volume, national policy support, and good industrial situation have laid a solid foundation for artificial intelligence to usher in gold development in China. However, it is undeniable that China still lacks a profound accumulation of artificial intelligence theories and algorithms, and most of the deep learning frameworks also use foreign open source frameworks. This is a shortcoming that China will break in the future."
Artificial intelligence chips are undoubtedly a huge industry development trend, but the original chips will not become artificial intelligence chips overnight. The Lijun said: “If artificial intelligence is to maintain high growth in China and even to achieve a curve overtaking, it will require both policy and support from the big ecosystem. No matter whether it is technology or application, it will not be possible to go it alone and prosperity is the only way for China to survive. The success of the artificial intelligence industry."
500W Digital Tv Transmitter,Wireless Digital Tv Transmitter,Digital Tv Broadband Transmitter,Digital Tv Signal Transmitter
Anshan Yuexing Technology Electronics Co., LTD , https://www.yxhtfmtv.com