![<?echo $_SERVER['SERVER_NAME'];?>](/template/twentyseventeen/skin/images/header.jpg)
On May 23, in the 103-year-old San Francisco Art Palace, Intel's new technology conference - Artificial Intelligence Developers Conference ("AIDC") arrived on schedule. This time, Intel focused on broadening the artificial intelligence ecosystem.
Between the Roman-style architecture and the AI ​​scene of the technological sense, Intel's AI helm Naveen Rao talked about Intel's artificial intelligence software and hardware combination, and the most important information is the release of the Nervana neural network chip, according to the plan. Intel's latest AI chip, the Nervana NNP L-1000, will be officially launched in 2019, which is Intel's first commercial neural network processor product.
Two years ago, Naveen Rao was the CEO and co-founder of deep learning startup Nervana Systems. After the company was acquired by Intel, Nervana became the core battleship of Intel's artificial intelligence, and the Nervana NNP series came into being. Naveen Rao was appointed as the head of the artificial intelligence product division.
Carey Kloss, vice president of Intel's Artificial Intelligence Products Division and Nervana team member, said in an interview with a reporter from the 21st Century Business Herald: "We started researching Lake Crest (Nervana NNP series initial chip code) in the early days of our business. At that time, our entire team was about 45. People, are building a largest Die (silicon chip), we have developed Neon (deep learning software), and also built a cloud stack, which is done by small teams. But this is also the challenge, small team growth will have pain It took us a long time to get the first batch of products out, and Nervana was founded in 2014, until the chip was finally released last year."
Image from the web
However, after joining Intel, Nervana can use all kinds of resources from Intel. "Of course, calling resources is not an easy task, but Intel has extensive experience in the marketization of products. At the same time, Intel has seen me so far. The best post-silicon bring-up and architecture analysis.†Carey Kloss told 21st Century Business Herald, “In terms of production chips, we have hundreds of systems running simultaneously, Nervana employees and 6 months. The members who have just joined the team are working together for the new product day and night.†In his view, Nervana is now in a reasonable rhythm and has all the elements for success next year.
In addition to Nervana, Intel's acquisition of artificial intelligence flagship companies include Movidius, which focuses on visual processing, Altera, an FPGA (Field Programmable Gate Array) giant, and Smartye-related Mobileye. In fact, since 2011, Intel has been investing in artificial intelligence-related companies, including China's Cambrian and Horizon.
At the same time, Intel's competitors are growing. NVIDIA's GPUs are advancing in the field of artificial intelligence; Google recently released the third-generation AI chip TPU, which is optimized for Google's deep learning architecture TensorFlow, and Google provides developers with the underlying services such as TPU; last year, Baidu United ARM, Ziguang Zhanrui and Hanfeng Electronics released DuerOS smart chips, mainly providing voice interaction solutions; Facebook and Alibaba have also entered the chip field, among which Alibaba Dharma is developing a neural network chip called Ali-NPU. Mainly used in scenes such as image, video recognition and cloud computing.
In the "encounter battle" of this artificial intelligence chip, how will Intel respond?
Three major factions
From a holistic point of view, the current pattern of global artificial intelligence is not yet clear, and it belongs to the local warfare of their own technological exploration, and has not yet entered the overall battle of the masses. Artificial intelligence is a general concept. The specific application scenarios are quite different. The focus of each company is different. If classified according to technology and business genre, global companies can be divided into three factions.
One is the system application, the most typical representatives are Google and Facebook. Not only do they develop system-level frameworks for artificial intelligence, such as Google's famous artificial intelligence framework Tensorflow, Facebook's Pytorch, but also large-scale applications. For example, Google has invested heavily in R&D autopilots and launched 2C services such as translation. And Facebook also applies artificial intelligence technology to image processing, natural language processing and many other fields in social networks.
The second category is the chip faction. Currently, it mainly provides computing support. The biggest players are Intel and NVIDIA. NVIDIA's GPUs capture the critical moments of computing device demand, with outstanding performance in graphics rendering, artificial intelligence, and blockchain, and put pressure on Intel in these areas. At the same time, Nvidia seems to be different from Intel's "Intel Inside". It hopes to become a real computing platform and successfully launch its own CUDA platform.
On May 30th, Nvidia released the world's first computing platform that combines artificial intelligence and high-performance computing, the HGX-2, which is the computing platform behind the DGX-2, the largest GPU available today.
As the leader of traditional computing power, Intel is naturally not to be outdone. The 50-year-old company is quite old-fashioned. In recent years, it has launched heavy mergers and acquisitions in the field of artificial intelligence: 2015: $16.7 billion acquisition of "Field Programmable Gate Array Giant" (Field Programmable Gate Array (FPGA) Altera, which lays the foundation for the future development of computing power, FPGA has great potential in cloud computing, Internet of Things, edge computing, etc. In 2016, Intel acquired Nervana and plans to use this company to learn deeply. In terms of the ability to fight against the GPU; in the same year also acquired the visual processing chip startup Movidius; in 2017, Intel acquired Israel to assist the driving company Mobileye for $ 15.3 billion, aiming to enter the field of automatic driving.
In addition to the system application and chip, the third category is the technology application, and most of the remaining companies fall into this category. Although different companies claim to have deep and even unique technology accumulation in the field of deep learning and artificial intelligence, in fact, most of them are based on the system application and chip technology platform. However, the technology application is more for C-end users, including autopilot, image recognition, enterprise applications. Objectively speaking, the application of technology belongs to "the gentleman is good at the object".
From the current competitive landscape, the system application faction has gradually occupied the overall advantage and has the core competitiveness in the field of artificial intelligence. In the era of traditional computers and mobile phones, systems and chips are more cooperative, and chips are even more dominant. Specifically, for example, in the computer market, Intel is completely dominated in the field of computing power, spanning PC and Apple's MAC machines. On the system side, Windows and iOS have their own advantages and cannot replace each other, but their common Intel cannot replace it. In the era of mobile phones, although the protagonist of computing power has changed from Intel to Qualcomm, the chip is still in the core position, and its importance and operating system are equally divided.
In the past 1-2 years, the situation has changed very quickly. Apple released its own tone of research and development and production of MAC chips. Intel’s share price once fell. In the field of artificial intelligence, such a trend is more obvious. Due to the great difference in the requirements of computing scenarios, it is necessary for Google to develop mature chips according to its own needs, and it is technically more feasible. If Intel wants to customize the chip for different scenarios, it means that Intel will fully transfer to the 2B field. Compared with the previous 2B2C model, the pure 2B business will obviously be more like Party B, and the complexity of the business line will increase dramatically. Historically, a company's overall shift from 2C to 2B has often been due to the loss of its core dominance in the industry.
Betting Nervana NNP
Then, in the fierce competition, how does Intel further increase the chip business?
After joining Symantec, Naveen Rao became Intel's vice president and head of the AI ​​Business Unit (AIPG), leading the launch of the Intel Neural Network Processor (Nervana NNP) series of chips. This time at the AIDC conference, it provides software tools, hardware, and ecology for developers. In the industry's view, with Intel's technical strength, software tools and hardware are not a problem, but the ecology is still open to question. In the PC era, the core of the ecology is the chip, so building an ecosystem around the chip can make Intel solid, but in the era of artificial intelligence, the artificial intelligence system is the core of the ecology. The chip that provides the computing power is part of the ecology, and the CPU can provide calculations. Power, GPU can also be provided, Intel can produce, NVIDIA can also be produced, and even Google, Apple itself can also produce.
Currently in the field of data science and deep learning computing, Intel's chip layout mainly includes Xeon (Xeon) chip series, Movidius's vision chip VPU, Nervana NNP series, and FPGA (field programmable gate array). These product lines correspond to several different sub-application scenarios.
The Nervana NNP series is a neural network processor. In the training and inference phase of deep learning, the Nervana NNP is mainly for the calculation of the training phase. According to Intel's plan, Deep Learning ("DL") will be implemented by 2020. The effect is increased by 100 times. This neural network processor is designed by Intel and Facebook. It can be predicted that the chip should have great support for Facebook's machine learning framework Pytorch. After all, Facebook's Pytorch ambition is definitely to work with Google's Tensorflow. A high score. However, the latest chip will be officially launched in 2019, and the change in the pattern of deep learning will be unpredictable.
Naveen Rao wrote in his blog: "We are developing the first commercial neural network processor product, the Intel Nervana NNP-L1000 (codenamed Spring Crest), which is scheduled to be released in 2019. Compared to the first generation of Lake Crest products, we The Intel Nervana NNP-L1000 is expected to achieve 3-4 times the training performance. The Intel Nervana NNP-L1000 will also support bfloat16, a numerical data format widely used in the industry for neural networks. In the future, Intel will be in artificial intelligence. The product line expands support for bfloat16, including Intel Xeon processors and Intel FPGAs."
In fact, the rumors that Spring Crest launched at the end of 2018 have already existed, but it seems that the official announcement of the 2019 time is slightly delayed. In this regard, Carey Kloss explained to the reporter: "Into the more modern process node, we integrated more Die (silicon chip), can get faster processing speed. But it takes a certain time to manufacture silicon wafer, also need Time turns the silicon into a new neural network processor, which is the reason for the delay."
For the difference between the two generations of chips, he analyzed: "Lake Crest, as the first generation processor, achieves very good computational utilization on both GEMM (matrix operations) and convolutional nerves. This is not just about 96% throughput. The utilization of the quantity, but in the case of insufficient customization, we have achieved the calculation utilization of GEMM higher than 80% in most cases. When we develop the next generation chip, if we can maintain high computing utilization Rate, the new product has a performance improvement of 3 to 4 times."
When it comes to competition, Carey Kloss said: "I don't know what our competitors' roadmap is, but our response is relatively fast, so I don't think we will be at a disadvantage in neural network processing. For example, bfloat16 has been around for a while. However, it has recently become more popular, many customers have asked for support for bfloat16, and we have gradually turned to support bfloat16." Compared with Google's TPU, he thinks that TPU II is similar to Lake Crest, and TPU is similar to Spring. Crest.
Attack on all sides
In addition to the highly regarded Nervana NNP, Intel's Xeon chips are mainly for servers and large computing devices. For example, China's supercomputers Tianhe No. 1 and No. 2 use Intel Xeon six-core processors.
In terms of visual chips, Intel's business volume has grown rapidly. Movidius VPU chips have long been used in emerging hardware markets such as automobiles and drones, such as the Uighur UAV, Tesla, and Google Clips cameras.
Gary Brown, Movidius's marketing director, told 21st Century Business Herald: "At Movidius, the chip we developed is called the Visual Processing Unit VPU. VPU is a chip that combines computer vision and smart camera processors. So our chips There are probably three types of processing: ISP processing, which is image signal processing, processing based on camera capture technology, and computer vision and deep learning."
For example, his specific usage scenarios include VR products and robotics, smart homes, industrial cameras, AI cameras, and surveillance and security. Among them, "monitoring and security is a huge market, especially in China, the market for surveillance and security cameras is particularly large, and some large companies are developing surveillance cameras such as Hikvision and Dahua."
Gary Brown also mentioned that the smart home sector is currently growing rapidly, and although the market is small, it is growing rapidly. “There are many companies that are developing smart devices such as smart home security, personal home assistants, smart doorbells, and access control for apartments and homes. But in the home sector, low cost, low power consumption, long battery life, and very precise It is very challenging. Because the outdoor shade is moving, it is possible to trigger an burglar alarm, so a very low false alarm rate is very important and has good accuracy."
One of the company's challenges is how to continue to create high-performance chips. "We have some strategies, such as using a front-end algorithm to reduce power consumption, so that we can turn off most of the chips and only run a small part of the optimized face detection function. When a face appears, other chips will be activated. This will keep the face monitoring system open. We also have a lot of energy-saving technology to make the home smart camera last for about 6 months," Gary Brown explained.
In addition, the FPGA line is dominated by Altera. With the arrival of the 5G wave, the data analysis and computing requirements of the IoT IoT will increase sharply. The access nodes of the Internet of Things are at least tens of billions of scales, which is 1-2 orders of magnitude higher than the scale of mobile phones. The typical requirement of the Internet of Things is the need to flexibly use algorithmic changes. This is the strength of FPGAs. FPGAs can adapt to the needs of customized computing scenarios through changes in their own structure, which also makes Intel more efficient for more different types of devices in the future. It is possible to provide chips. From the $16.7 billion acquisition amount, it can be seen that Intel’s purchase is obviously not just the immediate value.
Quick attack enterprise level scene
A recent Intel survey found that more than 50% of US corporate customers are turning to existing cloud solutions based on Intel Xeon processors to meet their initial needs for artificial intelligence. Many Intel executives told reporters in an interview that no solution is available for all artificial intelligence scenarios, and Intel will match technology and business according to customer needs. For example, Intel will configure Xeon and FPGA, or Xeon and Movidius together to achieve higher performance artificial intelligence.
For Intel, these enhanced artificial intelligence capabilities will be widely used in enterprise-level scenarios. Naveen Rao said: “When accelerating the transition to AI-driven future computing, we need to provide a comprehensive enterprise-class solution. This means our solution offers the widest range of computing power and is able to support from milliwatts. A variety of architectures ranging from kilowatts to kW."
Carey Kloss further explained the application scenario of artificial intelligence chips to 21st Century Business Herald: "Spring Crest is arguably the highest-ranking Nervana neuron processor architecture. Therefore, its customers include hyperscale computing centers and already have quite powerful data. Large companies, governments, etc. that work scientifically. If you need a low-profile and small model, Xeon can help you, and it can get data from the cloud to the end."
Specifically, Intel has also explored the scenes of medical, driverless, new retail, and Internet of Things. For example, in the medical field, it is reported that Intel is working with Novartis (NovarTIs) to use deep neural networks to accelerate high-content screening – a key element in early drug development. The cooperation between the two parties reduced the time to train the image analysis model from 11 hours to 31 minutes - an efficiency increase of more than 20 times.
In the unmanned store, Intel provides “computing brain†for the Jingdong unmanned convenience store, which has been deployed in several smart stores (Sinopec EasyJet convenience store, Jingdong home) and smart vending machine projects. In terms of algorithm, Jingdong said that the machine learning algorithms used by unmanned stores are mainly concentrated in three directions: knowing people, knowing goods, and knowing the field. Because of the involvement of online and offline data, unstructured data such as video is converted into structural data. Etc., you need to use the popular machine vision field CNN (convolution neural network) algorithm, the traditional machine learning algorithms used in the smart supply chain, such as SVM, statistical linear regression, logistic regression and so on. In the case of better network conditions, most video data can be completed in the cloud using a larger model. In the case of poor network, the edge calculation is done using a small network through end computing such as mobile. The hardware used includes Intel's edge server and so on.
Despite the strong enemy of Intel, the pace of transformation and expansion is very firm. From the perspective of R&D values, according to IC Insights statistics, the top 10 semiconductor manufacturers in 2017 had a total R&D expenditure of US$35.9 billion, and Intel ranked first. According to the report, Intel's R&D expenditure in 2017 was US$13.1 billion, accounting for 36% of the Group's total expenditure, which is about one-fifth of Intel's 2017 sales.
With the huge investment of each family, the battle of AI chips will intensify. (Source: tuicool)
Circular Connectors
Circular connectors, also called [circular interconnects," are cylindrical, multi-pin electrical connectors. These devices contain contacts that transmit both data and power. Cannon (now ITT Tech Solutions) introduced circular connectors in the 1930s for applications in military aircraft manufacturing. Today, you can find these connectors in medical devices and other environments where reliability is essential.
Circular Connectors is designed with a circular interface and housing to quickly and easily connect and disconnect signal, power, and optical circuits. Circular connectors are often preferred in military, aerospace, and industrial applications: these can be connected and disconnected without the use of coupling tools such as torque wrenches. Antenk`s circular connectors offer rugged solutions that have been engineered for reliable performance in a wide variety of harsh environment applications.
I/O connectors provide secure electrical contact and smooth, safe disconnect. They are used across a range of industries for communications devices, business equipment and computers. Top considerations when purchasing I/O connectors include pinout, gender, voltage rating, contact plating, and termination style.
Types of Circular Connectors
Circular connectors typically feature a plastic or metal shell surrounding the contacts, which are embedded in insulating material to maintain their alignment. These terminals usually pair with a cable, and this construction makes them especially resistant to environmental interference and accidental decoupling.
Circular I/O Connectors Types
Audio Connector
BNC/TNC Connector
DC POWER Connector
Mini din connector
DIN Connector
M5/M8/M12/M16/M23 PP9 Connector
SMA/SMB/FME Connector
RCA Connector
MCX MMCX Connector
Power Connector,Circular I Connectors,Circular O Connectors,Circular Power Connector,SMA/SMB/FME Connector,RCA Connector,MCX MMCX Connector,and we are specialize in Circular O Connectors,Circular Power Connector
ShenZhen Antenk Electronics Co,Ltd , https://www.atkconnectors.com