Ethereum bond
Publish: 2021-03-28 23:54:53
1.
1、 Different definitions:
1. virtual currency:
virtual currency refers to non real currency
digital currency:digital currency is an alternative currency in the form of electronic currency. Both digital gold coin and cryptocurrency belong to digiccy
3. Cryptocurrency:
cryptocurrency is a kind of transaction medium that uses cryptography principles to ensure transaction security and control the creation of transaction units
4. Token (token):
a kind of article whose shape and size are similar to currency, but the scope of use is limited and has no currency effect, and its token is the homonym of token in English
Second, the characteristics are different:1; It can also be said that virtual currency is personalized currency. In another way, it can also be called information currency
2. Digital currency:
is an unregulated and digital currency, which is usually issued and managed by developers and accepted and used by members of specific virtual communities
Cryptocurrency:cryptocurrency is based on the decentralized consensus mechanism, which is opposite to the banking and financial system relying on the centralized regulatory system
4. Token (token):
usually needs to be exchanged for money, used in shops, playgrounds, mass transportation and other places, as a voucher to use services and exchange goods
extended data
at present, digital currency is more like an investment proct, because it lacks a strong guarantee agency to maintain its price stability, and its role as a value measure has not yet appeared, so it can not be used as a means of payment. As an investment proct, digital currency cannot develop without trading platform, operating company and investment company
digital currency is a double-edged sword. On the one hand, the blockchain technology it relies on has been decentralized and can be used in other fields except digital currency, which is one of the reasons why bitcoin is popular; On the other hand, if digital currency is widely used by the public as a kind of currency, it will have a huge impact on the effectiveness of monetary policy, financial infrastructure, financial market and financial stability
2. When I first answered this question, it was on the 70th day of the crowdfunding period of EOS (a crowdfunding day of EOS is 23 hours). At that time, there was a lot of sadness under this answer, because it had dropped from the high of $3 in the initial crowdfunding period to nearly $0.50. When I started talking, there was little attention. In the next two months, the subscription price of EOS has been hovering around this position. If we follow up the subscription, one eth can get about 500 EOS, then the revenue should exceed most of the participants
reviewing the financing of EOS, we not only set many records of ICO, but also a significant milestone in the whole history of open market financing. During the 350 day subscription cycle, 200000 people participated in ICO, accumulating 7.2 million ether coins. According to the market price of the day, the total amount reached US $4.3 billion, which is close to the total IPO amount of Google (US $2.7 billion) and twitter (US $2.1 billion). It is the largest ICO in history so far. It's very worthwhile to make a comprehensive summary. I very much hope that what I write can penetrate the surface of the facts, catch the essence hidden behind, and become a part of historical records
EOS's ICO uses Ethereum's smart contract to complete the automatic subscription and allocation of funds. There is no middleman, and there is no role of investment bank and underwriter of traditional securities listing. This is a primary market. The total number of EOS coins is 1 billion, of which 900 million are distributed to subscribers and 100 million are reserved by block.one. All the funds obtained from the subscription belong to block.one company. Meanwhile, according to the subscription agreement, these funds directly become the profits of the company, and there is no other responsibility and obligation relationship with the subscribers, nor the responsibility and obligation to maintain and develop the EOS software and network. The main network of EOS will be launched by the loosely organized technology forum community, The network operation funds are automatically generated by the code rules, and new EOS tokens are issued to pay to the block procer nodes
in the process of ICO, the secondary market (exchange) is fully opened at the same time. After the completion of the subscription in the primary market, the general financing will enter the secondary market for a period of time. However, the financing cycle is as long as 350 days. The obtained Ethereum EOS token can enter the secondary market for circulation on the same day, resulting in a large number of turnover transactions, The prices of the two markets fluctuated simultaneously
after the end of ICO, the total market value of EOS was US $11 billion, ranking fifth in the list of digital currencies. BCH (US $17 billion), which is closer to fourth place, is more than twice as big as ripple (US $24 billion), which is third place. Compared with the total funding of $4.3 billion, according to the popular view of ICO, it is only three times as much
reviewing the financing of EOS, we not only set many records of ICO, but also a significant milestone in the whole history of open market financing. During the 350 day subscription cycle, 200000 people participated in ICO, accumulating 7.2 million ether coins. According to the market price of the day, the total amount reached US $4.3 billion, which is close to the total IPO amount of Google (US $2.7 billion) and twitter (US $2.1 billion). It is the largest ICO in history so far. It's very worthwhile to make a comprehensive summary. I very much hope that what I write can penetrate the surface of the facts, catch the essence hidden behind, and become a part of historical records
EOS's ICO uses Ethereum's smart contract to complete the automatic subscription and allocation of funds. There is no middleman, and there is no role of investment bank and underwriter of traditional securities listing. This is a primary market. The total number of EOS coins is 1 billion, of which 900 million are distributed to subscribers and 100 million are reserved by block.one. All the funds obtained from the subscription belong to block.one company. Meanwhile, according to the subscription agreement, these funds directly become the profits of the company, and there is no other responsibility and obligation relationship with the subscribers, nor the responsibility and obligation to maintain and develop the EOS software and network. The main network of EOS will be launched by the loosely organized technology forum community, The network operation funds are automatically generated by the code rules, and new EOS tokens are issued to pay to the block procer nodes
in the process of ICO, the secondary market (exchange) is fully opened at the same time. After the completion of the subscription in the primary market, the general financing will enter the secondary market for a period of time. However, the financing cycle is as long as 350 days. The obtained Ethereum EOS token can enter the secondary market for circulation on the same day, resulting in a large number of turnover transactions, The prices of the two markets fluctuated simultaneously
after the end of ICO, the total market value of EOS was US $11 billion, ranking fifth in the list of digital currencies. BCH (US $17 billion), which is closer to fourth place, is more than twice as big as ripple (US $24 billion), which is third place. Compared with the total funding of $4.3 billion, according to the popular view of ICO, it is only three times as much
3. Xpot is a coin ring artifact jointly launched by win GC group of Thailand and pundix of Indonesia. Xpot is a powerful combination of xpot and ipchain. All applications of xpot are developed based on IP chain of intellectual property chain. It will open up the last kilometer of digital asset circulation. It is matched by a card similar to a bank card - xpass
4. If you have spare money, you can go to the okex platform to open an account, smash it in and do a long-term job. You don't have to care about the rise and fall in the middle. This requires you to have a good attitude or make a fixed investment every month.
5. There is nothing wrong with the proct, but the fault lies in the people's heart. The reality is like this. There are those who make a lot of money, and there are also those who have nothing to lose
6. [interpretation of new technology of auto home]? What is the most important component of the automatic driving system? Is it a sensor? Is it control software? Or processing chip? Personally, I think at the present stage, the processing chip is the most critical component, and its performance directly affects the quality of the automatic driving system. In the past, the top chip technology has been monopolized by foreign enterprises, but with the rapid catch-up of Chinese chip enterprises in recent years, the situation has changed. Today, let's talk about the level of China's automatic driving chip
● what is the autopilot chip for
although L3 level conditional autonomous vehicles have not yet landed in China, from some vehicles with high-level L2 driving assistance system, we can find that these vehicles are equipped with a large number of sensors to detect obstacles around the vehicle, so as to provide data support for control system decision-making. These sensors include millimeter wave radar, ultrasonic radar, camera and so on. These sensors will generate several GB (1GB = 1024MB = 10242kb) of data per second. The autopilot chip needs to process these data smoothly to ensure that the system can make correct decisions in time, so as to ensure the driving safety of the vehicle
maybe you don't have the concept of gigabytes per second. Here's an example from life. The peak reading speed of a common USB 3.0 interface U disk is close to 200MB / s. It takes about 5 seconds to read 1GB files from this U disk, which shows that the amount of data in GB / S is quite large
automatic driving system needs not only to solve the problem of large flow data transmission, but also to solve the problem of how to deal with these huge amounts of data quickly, and the powerful automatic driving chip is the right key.
● what is the level of foreign autopilot chips
although this article mainly talks about the development of China's autopilot chip, we should know ourselves well and win every battle. Before we examine the local situation, we should first understand the situation abroad. There are only three foreign autopilot chips that can really enter the mass proction car market on a large scale: NVIDIA, Mobileye (now acquired by Intel) and Tesla
among them, Mobileye, which takes a practical route, currently has a market share of more than 70%. The procts on the market are mainly eyeq3 chips (0.256tops, "tops" means trillions of operations per second), which are used in L2 driving assistance system, Unless otherwise specified, the computing power indicated in this paper refers to 8-bit integer computing power) and eyeq4 chip (computing power 2.5tops) with L3 level automatic driving ability. For example, Xiaopeng G3, Weilai ES6 / es8 and GAC new energy aion LX adopt eyeq4 chip as the core of their driving assistance system
compared with the computing power of drive PX Pegasus 320tops, the flagship of NVIDIA's previous generation automatic driving platform, the flagship configuration of the new drive AgX Orin platform has achieved multiple performance growth. In addition, the degree of flexibility of the drive AgX Orin platform is further improved compared with the previous platform, which can meet the requirements of different levels of vehicles from general driving assistance to L5 level fully automatic driving through the increase or decrease of hardware configuration
Tesla autopilot 1.0 system uses one NVIDIA tegra3 chip and one Mobileye eyeq3 chip; Autopilot 2.0 system uses one NVIDIA Tegra Parker chip and one Pascal architecture GPU chip; Autopilot 2.5 system uses two NVIDIA Tegra Parker chips and one Pascal architecture GPU chip
the self-developed FSD chip has been carried on the latest off-line Tesla models. The computing power of a single chip is 72tops. The full self driving computer integrates two FSD chips that work independently. One "hangs" and the other "goes up" immediately, which improves the security and stability of the whole system
of course, in addition to the above three enterprises, there are also many enterprises coveting the cake of autonomous driving chips, including Qualcomm, Xilinx, NXP, etc., but the autonomous driving chips that these enterprises are really going to mass proce cars are not large-scale. Limited to space, I will not introce them here
● the rapid rise of China's automatic driving chip Enterprises. Auto driving chip market is booming, and foreign technology giants rush to land. How about the strength of Chinese enterprises? Now let's take a look< The Cambrian China Science and Technology Co., Ltd. (hereinafter referred to as "Cambrian") is a research group led by Chen Yunji and Chen Tianshi under the Institute of computing technology, Chinese Academy of Sciences. The research group began to study neural network algorithm and chip in 2008, and began to publish research results in 2012
in 2016, diannaoyu, the instruction set of deep learning processor proposed by the above research group, was accepted by isca2016. Experiments show that the chip with this instruction set has two orders of magnitude performance advantages in neural network computing compared with the chip with traditional x86 instruction set. As the research results of the research group tend to mature, China Science and technology Cambrian science and Technology Co., Ltd. was officially established, and began to transform its chip and instruction set to the commercial field. Also in 2016, Cambrian released its first commercial deep learning processor, Cambrian 1a
after talking about the company's life experience, let's take a look at its procts. At present, there are two latest AI chip IP licenses in Cambrian, namely cambrian-1m and cambrian-1h. Cambricon-1m-4k, with the strongest performance index, has 8tops computing power at 1GHz clock frequency; Cambricon-1h8mini, with the weakest performance index, has 0.5tops computing power at 1GHz clock frequency. The detailed force calculation parameters of all models can be seen in the table below
cambricon-1m and cambricon-1h are defined as terminal intelligent processor IP. The applications such as face recognition, fingerprint recognition, obstacle recognition, road sign recognition and so on, which appear on mobile phones or cars, can be accelerated by integrating the above processor IP in the chip
the word "edge" mentioned above comes from "edge computing". Edge computing refers to providing network, storage, computing, application and other capabilities at the end close to intelligent devices (terminals) or data sources (cloud), so as to achieve faster network service response and more secure local data transmission. Edge computing can meet the requirements of real-time business, intelligent applications, security and privacy protection, and provide local intelligent services for users. Siyuan 220 plays the role of improving data security, recing processing delay and optimizing bandwidth utilization in edge computing
at present, Cambrian high computing power chip procts are defined as smart accelerators, which can be used to accelerate AI operations in servers. I believe you've heard that Google's alphago AI robot beat South Korea's world go champion Li Shishi. Behind the alphago AI robot is Google's self-developed TPU chip. The features and applications of Cambrian high computing power chip procts are similar to Google TPU. Of course, they can also be regarded as competitors
the difference is that Siyuan 270-s4 adopts a passive cooling design, with a maximum thermal design power consumption of 70W, and is positioned as a data center accelerator card with higher energy efficiency than that designed by artificial intelligence reasoning. This also means that the card will have a "power wall" setting, that is, when the power consumption of the accelerator card reaches the upper limit of the threshold, the computing power will be reced to ensure lower power consumption and heating
Siyuan 270-f4 is equivalent to "full blood version" Siyuan 270-s4, with a maximum thermal design power consumption of 150W and active cooling by turbofan. Good heat dissipation and sufficient power supply enable Siyuan 270-f4 to give full play to the full performance of Siyuan 270 chip. The positioning of the card is to provide data center level AI computing power for desktop environment. In short, it is a high-performance AI accelerator card for desktop
although Siyuan 270 only uses TSMC's 16nm process, the overall energy consumption ratio is quite good. Although the computing power of a single card is less than that of the latest NVIDIA flagship computing card, the peak computing power of five Siyuan 270-s4 / Siyuan 270-f4 parallel cards can also reach the level of NVIDIA A100. However, the more advanced process of NVIDIA A100 should have certain advantages in energy consumption
among them, Siyuan 100-c is equipped with video and image decoding units and adopts passive cooling mode, with the maximum thermal design power consumption of 110W; Siyuan 100-d is not equipped with video and image decoding unit, and adopts passive cooling mode, with the maximum thermal design power consumption of 75W. At present, Siyuan 100 series procts have been applied in didi cloud and Jinshan cloud in 2019. Didi cloud uses Siyuan 100 card to accelerate the elastic reasoning service, which is used for deep learning reasoning task; Jinshanyun uses Siyuan 100 card to accelerate the application of artificial intelligence such as voice, image and video
what I mentioned above is all about server level computing cards. Does this deviate from the topic of autopilot chip that we should talk about? Not really. As mentioned earlier, Cambrian is currently an enterprise focusing on the development of artificial intelligence chips, and the field of automatic driving is not deeply involved, but there are still some achievements through the cooperation with other domestic partners< The br / > WiseADCU CN1 automatic driving domain controller provides the computing power and the number of sensor connections required for L3 or above level automatic driving system, and realizes the self control of seven key control points of simulation, model, system, architecture, encoding, acceleration, and algorithm.
in fact, e to the weak performance competitiveness of processor procts, Weisheng group has long withdrawn from the competition in the mainstream x86 processor market, leaving Intel and AMD competing in the market. Since the establishment of megacore, it has mastered Viagra's x86 technology, made comprehensive improvement and Optimization on the basis of the latest processor architecture of Viagra at that time, and successively launched zx-a, zx-c, zx-c + and other processor procts
on June 2, the municipal Party committee of science and Technology Innovation Board announced the results of the 33rd review meeting in 2020, and the listing of Cambrian was approved. From acceptance to approval, it took only 68 days for Cambrian to refresh the review speed of science and technology innovation board. Cambrian has become the only AI chip company in A-share market since its listing. The market space in this field is expected to exceed 50 billion US dollars in 2022, with huge development potential. Can Cambrian, which has opened up A-share financing channels, further develop and grow with its unique technical advantages? No one can say for sure, but it is certain that the successful listing of the Cambrian period has won the confidence and hope of many companies engaged in this field, which may usher in the era of Chinese artificial intelligence chips
> horizon robot
OK, after talking about the Cambrian, let's talk about another AI chip enterprise, horizon robot technology research and Development Co., Ltd. (hereinafter referred to as "Horizon"). Horizon is a company founded in 2015 by Yu Kai, former executive vice president of the online Deep Learning Institute, focusing on automatic driving and artificial intelligence chips. Yu Kai is also the originator of Internet autopilot
the horizon established by Yu Kai has always adhered to the direction of combining software and hardware. He believes that algorithms, chips and cloud computing will constitute the three core fulcrums of automatic driving. Compared with the Cambrian period, which focused on building high-performance hardware chips, horizon's business model focused on "algorithm + chip"
● what is the autopilot chip for
although L3 level conditional autonomous vehicles have not yet landed in China, from some vehicles with high-level L2 driving assistance system, we can find that these vehicles are equipped with a large number of sensors to detect obstacles around the vehicle, so as to provide data support for control system decision-making. These sensors include millimeter wave radar, ultrasonic radar, camera and so on. These sensors will generate several GB (1GB = 1024MB = 10242kb) of data per second. The autopilot chip needs to process these data smoothly to ensure that the system can make correct decisions in time, so as to ensure the driving safety of the vehicle
maybe you don't have the concept of gigabytes per second. Here's an example from life. The peak reading speed of a common USB 3.0 interface U disk is close to 200MB / s. It takes about 5 seconds to read 1GB files from this U disk, which shows that the amount of data in GB / S is quite large
automatic driving system needs not only to solve the problem of large flow data transmission, but also to solve the problem of how to deal with these huge amounts of data quickly, and the powerful automatic driving chip is the right key.
● what is the level of foreign autopilot chips
although this article mainly talks about the development of China's autopilot chip, we should know ourselves well and win every battle. Before we examine the local situation, we should first understand the situation abroad. There are only three foreign autopilot chips that can really enter the mass proction car market on a large scale: NVIDIA, Mobileye (now acquired by Intel) and Tesla
among them, Mobileye, which takes a practical route, currently has a market share of more than 70%. The procts on the market are mainly eyeq3 chips (0.256tops, "tops" means trillions of operations per second), which are used in L2 driving assistance system, Unless otherwise specified, the computing power indicated in this paper refers to 8-bit integer computing power) and eyeq4 chip (computing power 2.5tops) with L3 level automatic driving ability. For example, Xiaopeng G3, Weilai ES6 / es8 and GAC new energy aion LX adopt eyeq4 chip as the core of their driving assistance system
compared with the computing power of drive PX Pegasus 320tops, the flagship of NVIDIA's previous generation automatic driving platform, the flagship configuration of the new drive AgX Orin platform has achieved multiple performance growth. In addition, the degree of flexibility of the drive AgX Orin platform is further improved compared with the previous platform, which can meet the requirements of different levels of vehicles from general driving assistance to L5 level fully automatic driving through the increase or decrease of hardware configuration
Tesla autopilot 1.0 system uses one NVIDIA tegra3 chip and one Mobileye eyeq3 chip; Autopilot 2.0 system uses one NVIDIA Tegra Parker chip and one Pascal architecture GPU chip; Autopilot 2.5 system uses two NVIDIA Tegra Parker chips and one Pascal architecture GPU chip
the self-developed FSD chip has been carried on the latest off-line Tesla models. The computing power of a single chip is 72tops. The full self driving computer integrates two FSD chips that work independently. One "hangs" and the other "goes up" immediately, which improves the security and stability of the whole system
of course, in addition to the above three enterprises, there are also many enterprises coveting the cake of autonomous driving chips, including Qualcomm, Xilinx, NXP, etc., but the autonomous driving chips that these enterprises are really going to mass proce cars are not large-scale. Limited to space, I will not introce them here
● the rapid rise of China's automatic driving chip Enterprises. Auto driving chip market is booming, and foreign technology giants rush to land. How about the strength of Chinese enterprises? Now let's take a look< The Cambrian China Science and Technology Co., Ltd. (hereinafter referred to as "Cambrian") is a research group led by Chen Yunji and Chen Tianshi under the Institute of computing technology, Chinese Academy of Sciences. The research group began to study neural network algorithm and chip in 2008, and began to publish research results in 2012
in 2016, diannaoyu, the instruction set of deep learning processor proposed by the above research group, was accepted by isca2016. Experiments show that the chip with this instruction set has two orders of magnitude performance advantages in neural network computing compared with the chip with traditional x86 instruction set. As the research results of the research group tend to mature, China Science and technology Cambrian science and Technology Co., Ltd. was officially established, and began to transform its chip and instruction set to the commercial field. Also in 2016, Cambrian released its first commercial deep learning processor, Cambrian 1a
after talking about the company's life experience, let's take a look at its procts. At present, there are two latest AI chip IP licenses in Cambrian, namely cambrian-1m and cambrian-1h. Cambricon-1m-4k, with the strongest performance index, has 8tops computing power at 1GHz clock frequency; Cambricon-1h8mini, with the weakest performance index, has 0.5tops computing power at 1GHz clock frequency. The detailed force calculation parameters of all models can be seen in the table below
cambricon-1m and cambricon-1h are defined as terminal intelligent processor IP. The applications such as face recognition, fingerprint recognition, obstacle recognition, road sign recognition and so on, which appear on mobile phones or cars, can be accelerated by integrating the above processor IP in the chip
the word "edge" mentioned above comes from "edge computing". Edge computing refers to providing network, storage, computing, application and other capabilities at the end close to intelligent devices (terminals) or data sources (cloud), so as to achieve faster network service response and more secure local data transmission. Edge computing can meet the requirements of real-time business, intelligent applications, security and privacy protection, and provide local intelligent services for users. Siyuan 220 plays the role of improving data security, recing processing delay and optimizing bandwidth utilization in edge computing
at present, Cambrian high computing power chip procts are defined as smart accelerators, which can be used to accelerate AI operations in servers. I believe you've heard that Google's alphago AI robot beat South Korea's world go champion Li Shishi. Behind the alphago AI robot is Google's self-developed TPU chip. The features and applications of Cambrian high computing power chip procts are similar to Google TPU. Of course, they can also be regarded as competitors
the difference is that Siyuan 270-s4 adopts a passive cooling design, with a maximum thermal design power consumption of 70W, and is positioned as a data center accelerator card with higher energy efficiency than that designed by artificial intelligence reasoning. This also means that the card will have a "power wall" setting, that is, when the power consumption of the accelerator card reaches the upper limit of the threshold, the computing power will be reced to ensure lower power consumption and heating
Siyuan 270-f4 is equivalent to "full blood version" Siyuan 270-s4, with a maximum thermal design power consumption of 150W and active cooling by turbofan. Good heat dissipation and sufficient power supply enable Siyuan 270-f4 to give full play to the full performance of Siyuan 270 chip. The positioning of the card is to provide data center level AI computing power for desktop environment. In short, it is a high-performance AI accelerator card for desktop
although Siyuan 270 only uses TSMC's 16nm process, the overall energy consumption ratio is quite good. Although the computing power of a single card is less than that of the latest NVIDIA flagship computing card, the peak computing power of five Siyuan 270-s4 / Siyuan 270-f4 parallel cards can also reach the level of NVIDIA A100. However, the more advanced process of NVIDIA A100 should have certain advantages in energy consumption
among them, Siyuan 100-c is equipped with video and image decoding units and adopts passive cooling mode, with the maximum thermal design power consumption of 110W; Siyuan 100-d is not equipped with video and image decoding unit, and adopts passive cooling mode, with the maximum thermal design power consumption of 75W. At present, Siyuan 100 series procts have been applied in didi cloud and Jinshan cloud in 2019. Didi cloud uses Siyuan 100 card to accelerate the elastic reasoning service, which is used for deep learning reasoning task; Jinshanyun uses Siyuan 100 card to accelerate the application of artificial intelligence such as voice, image and video
what I mentioned above is all about server level computing cards. Does this deviate from the topic of autopilot chip that we should talk about? Not really. As mentioned earlier, Cambrian is currently an enterprise focusing on the development of artificial intelligence chips, and the field of automatic driving is not deeply involved, but there are still some achievements through the cooperation with other domestic partners< The br / > WiseADCU CN1 automatic driving domain controller provides the computing power and the number of sensor connections required for L3 or above level automatic driving system, and realizes the self control of seven key control points of simulation, model, system, architecture, encoding, acceleration, and algorithm.
in fact, e to the weak performance competitiveness of processor procts, Weisheng group has long withdrawn from the competition in the mainstream x86 processor market, leaving Intel and AMD competing in the market. Since the establishment of megacore, it has mastered Viagra's x86 technology, made comprehensive improvement and Optimization on the basis of the latest processor architecture of Viagra at that time, and successively launched zx-a, zx-c, zx-c + and other processor procts
on June 2, the municipal Party committee of science and Technology Innovation Board announced the results of the 33rd review meeting in 2020, and the listing of Cambrian was approved. From acceptance to approval, it took only 68 days for Cambrian to refresh the review speed of science and technology innovation board. Cambrian has become the only AI chip company in A-share market since its listing. The market space in this field is expected to exceed 50 billion US dollars in 2022, with huge development potential. Can Cambrian, which has opened up A-share financing channels, further develop and grow with its unique technical advantages? No one can say for sure, but it is certain that the successful listing of the Cambrian period has won the confidence and hope of many companies engaged in this field, which may usher in the era of Chinese artificial intelligence chips
> horizon robot
OK, after talking about the Cambrian, let's talk about another AI chip enterprise, horizon robot technology research and Development Co., Ltd. (hereinafter referred to as "Horizon"). Horizon is a company founded in 2015 by Yu Kai, former executive vice president of the online Deep Learning Institute, focusing on automatic driving and artificial intelligence chips. Yu Kai is also the originator of Internet autopilot
the horizon established by Yu Kai has always adhered to the direction of combining software and hardware. He believes that algorithms, chips and cloud computing will constitute the three core fulcrums of automatic driving. Compared with the Cambrian period, which focused on building high-performance hardware chips, horizon's business model focused on "algorithm + chip"
7. 1、 After arriving at the assembly center by transmission, walk to the center of the assembly center< After arriving at the center of the assembly center, walk to the people in black
3. When you approach the person in black and click it, the person is the trainer
4. After opening the trainer interface, click the "accept challenge" button to receive the task
the advantage of the skill book is that three people and nine demons surround the demons all the time. If you learn the above routine, you can bully the boss. I'm afraid Xiao doesn't know which combat power is higher than which one, so he may kneel down in the second and third level. Three people divided into output, auxiliary, meat shield, each dry
the first person, the person with the highest output, takes the whole output
the second person is the assistant, Yan Ruyu + Jiang Ruyuan + Luming King / crabs are hot. The output person is the Luming king who is attacked by things and crabs are attacked by methods
the third person is full of spirits.
3. When you approach the person in black and click it, the person is the trainer
4. After opening the trainer interface, click the "accept challenge" button to receive the task
the advantage of the skill book is that three people and nine demons surround the demons all the time. If you learn the above routine, you can bully the boss. I'm afraid Xiao doesn't know which combat power is higher than which one, so he may kneel down in the second and third level. Three people divided into output, auxiliary, meat shield, each dry
the first person, the person with the highest output, takes the whole output
the second person is the assistant, Yan Ruyu + Jiang Ruyuan + Luming King / crabs are hot. The output person is the Luming king who is attacked by things and crabs are attacked by methods
the third person is full of spirits.
Hot content