Wen Yi Xiao Yi Guangsheng Recently, market research company CompassIntelligence released the latest research report. In the top 24 list of AI chip companies in the world, the top three are Nvidia, Intel (Intel) and IBM respectively. Chinese companies occupy seven seats, and the highest ranking is the twelfth HUAWEI. In the list of Top24, a total of seven Chinese companies were selected. HiSilicon (HiSilicon) ranked twelfth in the list. The MediaTek was ranked fourteenth. Imagination row fifteenth; Twentieth in Rockchip. Twenty-first names in Verisilcon. Twenty-third in Cambrian (Cambricon); The horizon (Horizon) ranks twenty-fourth. The report also mentioned that in the past three years, the 24 companies invested a total of 60 billion dollars in the field of AI in addition to the research and development input of the AI chip. At present, more than 1700 start-up companies are interested in AI chips, and of course, the demand for AI chips is increasing. Next, intelligent bacteria will sort out the situation of these seven Chinese AI chip companies. 1, HUAWEI HiSilicon - Kirin 970 Time of establishment: 2004 Headquarters: Shenzhen In October 2004, HUAWEI founded the Hass company, whose predecessor was the HUAWEI IC Design Center, which also officially opened the way of R & D of HUAWEIs mobile phone chips. In 2009, HUAWEI launched its first open market - oriented K3 processor, which was positioned to compete with the exhibition and the coalition to compete in the Shanzhai market, and HUAWEIs own mobile phone was not used. Because the K3 product is not mature enough and the sales strategy is not suitable, the chip has not been successful. This is also the first smart handset processor in China. In 2012, China launched K3V2 processor for Hass. This time it was used in its own mobile phones, and it was the flagship Mate1, P6 and other models. In 2012, the mobile phone processor has opened the multi-core process, and K3V2 has become the second quad core processor in the world. Then, Kirin 910 is the first SoC (on chip). If CPU is the brain of the mobile phone, the SoC is a system that integrates various functions and assigns tasks to them. A mobile SoC includes the baseband (Baseband), the graphic processor (GPU), the digital signal processor (DSP), and the image signal processing in addition to CPU. ISP and other important modules. After more than ten years of development, in September 2017, HUAWEI officially launched its first new AI chip kylin 970 (Kirin970) at the Berlin international electronic consumer goods exhibition (IFA), Germany. Kirin 970 uses the industrys high standard TSMC10nm process, which integrates 5 billion 500 million transistors on a nail size chip, reduces power consumption by 20%, and achieves a 1.2Gbps peak download rate. Kirin 970 integrates NPU dedicated hardware processing unit (Cambrian IP), innovating the HiAI mobile computing architecture, and its AI performance density is much better than CPU and GPU. Compared to the four Cortex-A73 cores, the new heterogeneous computing architecture has about 50 times energy efficiency and 25 times performance advantages in handling the same AI task. Moreover, HUAWEIs second generation AI chip Hirst Unicorn 980 will also be formally produced in the current quarter, using TSMC 7Nm process technology. This processor will be equipped with the second generation of NPU, which will support more scenario applications on the basis of previous generations, and the performance of NPU will be increased by more than 2 times. 2, MediaTek -- Xi Li P60 Time of establishment: 1997 Headquarters: Taiwan Compared with other chip manufacturers, the pace of access to AI chips by MediaTek is slightly slower. It was not until March this year that the Taiwan chip manufacturer released the first built-in AI function chip P60. According to the Department of development, P60 first built a APU dedicated to AI operations, carrying NeuroPilotAI technology, which is an independent NPU unit that runs heterogeneous by CPU, GPU, APU, and provides a complete API support and developer toolkit from entry to advanced, performance and power optimization, portability, and power consumption. The characteristics of custom-made and so on. In product performance, the P60 uses 4 CortexA73 and 4 A53 core architectures of 4 A53, which is more efficient than the previous generation of HelioP23 and P30 products, and the performance of CPU and GPU increases by 70%. At the same time, due to the improvement of the processing technology, compared with the previous generation, the power consumption of large games was reduced by 25%, which greatly extended the use time of mobile phone batteries. According to Li Zonglin, general manager of the wireless communications business department of the joint hair technology, HelioP60 has integrated the solutions from many artificial intelligence manufacturers, such as Tencent, Shang Tang, rainbow, and open vision. It has achieved advanced level in the field of mobile phone security, photography, face identification and so on. 3, Imagination - PowerVRNNA Time of establishment: in 2017, it was bought by Chinese capital Headquarters: UK Person in charge: Li Liyou Imagination was originally a British chip company and used to be Apple Corps GPU supplier. In March 2017, Imagination was abandoned by apple, and its share price fell overnight by more than 7. Six months later, CanyonBridge, a Chinese private equity fund, invested 550 million pounds (about 4 billion 900 million yuan) to acquire it. However, Imaginations MIPSCPU business was stripped off by the interference of the US government and not acquired by CanyonBridge. After the acquisition, the Imagination company CEO was appointed by Li Liyou, Li Liyou previously served as the chairman of the exhibition, the post news was bought by purple light, and Li Liyou was the co - President of the purple light group. In addition to GPU, Imagination launched the PowerVRNNA neural network accelerator for hardware IP products for AI applications in September 2017. At present, Imagination has three main product lines, namely, GPU, neural network accelerator, IPEnsigma of communication and communication. 4, Ruixin micro (Rockchip) - RK3399Pro Time of establishment: 2001 Headquarters: Fuzhou Founder: Li Min Ruixin micro (Rockchip) is headquartered in Fuzhou, Fujian, the company currently has nearly 700 employees. As a Integrated Circuit Design Company, Ruixin provides chip solutions for smartphones, tablet computers, communication plates, TV set-top boxes, vehicle navigation, IoT Internet of things and multimedia audio and video products to provide professional chip solutions. On the CES2018 held in January this year, Ruixin released its first AI processor RK3399Pro, which was the first AI chip designed by Ruixins CPU+GPU+NPU hardware structure, and its chip NPU (neural network processor) operation performance was as high as 2.4TOPs, with high performance, low power consumption and easy opening and other advantages. The RK3399ProAI chip uses big.LITTLE size core CPU architecture, dual core Cortex-A72+ quad core Cortex-A53, quad core ARMGPUMali-T860, and integrates more bandwidth compression technology. Headquarters: Shanghai Founder: Dai Weimin Verisilcon was formerly a subsidiary of CelestryDesignTechnologies (Inc.) in Shanghai. The company expanded its capital in August 2002 and transferred the stock of the United States. It was renamed the core. It is the first company in China to provide the standard unit library of the chip. Headquartered in Shanghai, China, there are more than 600 employees worldwide. The core itself is a chip design platform, SiliconPlatformasaService (SiPaaS) provider, providing a comprehensive system level chip (SoC) and system level package (SiP) solution for a wide range of terminal markets including mobile interconnected devices, data centers, Internet of things (IoT), automotive, industrial and medical devices. In November 2004, Hongkong Zhonghua Technology Co., Ltd. and its subsidiary Shanghai Zhonghua Electronics Co., Ltd. merged into the core. In July 2006, the company successfully purchased the ZSP digital signal processor department from LSILogic. In 2015, the company also acquired graphic technology (Vivante) for embedded graphics processor designers in the United States. In May 2017, the mission launched a processor VIP8000 for computer vision and artificial intelligence applications, which could directly import neural networks generated by frameworks such as Caffe and TensorFlow. It is worth mentioning that NXPs flagship application processor i.MX8 also uses the core visual graphics processor. Dai Weimin is the founder, chairman, President and CEO of the company. Prior to that, he served as co chairman and chief technology officer of the US Celestry company (acquired by Cadence in 2002). 6, Cambrian (Cambricon) - 1A/1H/1M+MLU100 Headquarters: Beijing Founder: Chen Tianshi brothers In 2009, the team applied artificial intelligence to processor design verification; in 2011, the team worked with Nanjing University Zhou Zhihua to apply AI to processor architecture optimization to improve processor performance; in 2013, the worlds first deep learning processor architecture DianNao was developed to obtain the best ASPLOS2014 paper; in 2014, the team launched the global first A multi core deep learning processor architecture, DaDianNao, was once again obtained the best MICRO2014 paper; in 2015, the team successfully developed the worlds first dedicated processor chip for deep learning Cambrian, and Cambrian science and technology was registered in Beijing and Shanghai in 2016. In 2016, Cambrian issued the worlds first commercial deep learning processor Cambrian 1A, including the worlds first special artificial intelligence instruction set, CambriconISA, with fully autonomous intellectual property rights, theoretical peak performance 1GHz, and support for a variety of intelligent tasks such as vision, voice, Natural Language Processing, and so on. In 2017, Cambrian authorized HUAWEI Hass to use the Cambrian 1A processor, which was installed in the Kirin 970 chip and Mate10 series mobile phones. In August 2017, Cambrian science and technology announced the completion of the A round of financing, by the association of venture capital, Alibaba, investment, national investment, national investment, China branch Turin, Yuan Wo origin (angel Ring Leader), investment (angel wheel investment) joint investment. A financing makes Cambrian become the first Unicorn enterprise in the field of AI chip. In November 2017, a new intelligent processor was released in Cambrian. The first is the upgraded version of the 1A processor, with a higher performance Cambrian 1H16 processor. Another product is the 1H8 processor for the visual field. There are 4 configuration options, taking photo assistance, picture processing, security monitoring, and 2.3 times the performance power ratio of the 1A processor. In May 3rd this year, Cambrian issued a new generation of terminal IP products in Shanghai, using the 7Nm technology terminal chip Cambricon1M, the first cloud smart chip MLU100 and the MLU100 cloud intelligent processing card. The third generation machine learning terminal processor 1M has 10 times higher performance than the previous Cambrian 1A. In terms of configuration, Cambrian 1M is produced by TSMC 7Nm process, and its 8 bit operation efficiency is 5Tops/watt (5 trillion times per watt). Cambrian provides three processor cores of 2Tops, 4Tops and 8Tops to meet the needs of different levels of intelligent processing in different scenarios. MLU100 uses the latest Cambrian MLUv01 architecture and the TCP 16nm process, which can work in two different modes of equilibrium mode (primary frequency 1Ghz) and high performance mode (main frequency 1.3GHz). The equivalent theoretical peak velocity can reach 128 trillion fixed-point operations and 166 trillion and 400 billion fixed-point operations respectively, and their power consumption is 80W and 110W. The Cambrian introduction, MLU100 cloud chip also has high throughput, can support all kinds of deep learning and common machine learning algorithms. They also put forward the idea of end cloud collaboration, that is, MLU100 cloud chip can adapt to the Cambrian 1A/1H/1M terminal processor and cooperate with the complex intelligent processing task. . 7, horizon (Horizon) - journey + rising sun Time of establishment: 2015 Headquarters: Beijing Founder: Yu Kai The horizon (Horizon), founded by Yu Kai, the former vice president of Baidu IDL, is committed to providing high performance, low power, low cost, and full open embedded artificial intelligence solutions. At present, the worlds first world leading embedded AI vision chip based on the BPU (Gauss Architecture) has been successfully developed, introducing a Journey series processor and a Sunrise series processor for smart cameras, and providing algorithm + chip + to industry customers. A complete solution to the cloud. Journey 1 is oriented to intelligent driving. It can simultaneously monitor and identify many kinds of targets, such as pedestrians, motor vehicles, non motor vehicles, lane lines, traffic signs, traffic lights and many other targets, and meet the harsh environment requirements of the vehicle and the visual perception requirements under different environments. Sunrise 1 is an intelligent camera, which can be used for large scale face capture and recognition, video structured processing and so on. It can be used in many practical applications, such as business and security. There is still a long way to go for Chinese AI chips to catch up. On the whole, Chinese companies have a lot of seats on the global AI chip list, but if they want to reach the worlds leading level, it is still a long way to go, even to overtake the NVIDIA, Intel, and NDK. First, develop the accumulation and high tolerance of domestic chips for long-term R & D investment. This long-term investment is reflected in the positive cycle of heavy capital investment and high production; on the other hand, it is embodied in the design of micro architecture and the design ability of the underlying operating system. Second, if China wants to catch up with international rivals in the AI chip field, it has to do very well in a vertical field and really do all the stack things to the user and make it really applied. Again, in the long run, domestic manufacturers must foster the ecosystem of hardware developers. Intel and MS have rarely adopted such cross level strategic operation in domestic universities for many years to develop curriculum system, certification system and ecological cultivation system. (Note: this part of the data data from the network) is concerned about the NetEase intelligent public number (smartman163), for you to read the big events in the AI field, new ideas and new applications. Source: NetEase intelligent responsible editor: Wang Chao _NT4133 Second, if China wants to catch up with international rivals in the AI chip field, it has to do very well in a vertical field and really do all the stack things to the user and make it really applied. Again, in the long run, domestic manufacturers must foster the ecosystem of hardware developers. Intel and MS have rarely adopted such cross level strategic operation in domestic universities for many years to develop curriculum system, certification system and ecological cultivation system. (Note: part of the data is from the Internet). Pay attention to NetEase intelligent public address (smartman163), to interpret the big events, new ideas and new applications of large companies in AI field.