Have you ever imagined that one day, the silent personal computer at the foot of your house might find evidence of the existence of extraterrestrial civilization?
The "fifth type of contact" of such extraterrestrial information across time and space is not science fiction, but comes from operational reality.
Back in 1995, David Anderson, then assistant professor at UC Berkeley University in the United States, never thought that the SETI@home project he initiated would cause such intense attention.
SETI is the abbreviation of Search for Extra-terrestrial Intelligence. The Chinese literal translation of SETI@home is "Search for Extra-terrestrial Intelligence at Home".
SETI@home project official website homepage
This is a scientific experiment that uses distributed computing to aggregate global networked computers to jointly search for extraterrestrial civilizations. Volunteers who contribute personal computer computing power on the SETI platform can join the project by running a free program to download and analyze the outer space signal data transmitted from the radio telescope. This program usually runs in screen saver mode or background mode on the user's personal computer. It uses redundant processor resources and does not affect the normal use of computers by volunteers.
A scientific world built on computing power
The SETI@home center platform is set up in the Space Science Laboratory of the University of California, Berkeley, a well-known American institution of higher learning, and is where Professor David Anderson now works. David Anderson led the SETI@home project since May 17, 1999. As of May 2004, nearly 5×10E21 floating-point operations have been carried out, more than 1.3 billion data units have been processed, and 5.43 million users have been attracted. These users have worked on computers for 2.43 million years (this is only in the era when personal computers and the Internet were not so popular), and analyzed a large amount of backlog data.
The total computing power of these idle personal computers surpassed the processing power of any supercomputer in the world at that time. A volunteer once calculated that when more than 1,000 laptops connected to a high-speed local area network are running SETI@home at the same time in a corner of the university, it is possible to enter the world's top 500 supercomputers.
SETI@home is the most successful distributed computing test project so far. Its success is not only reflected in the possibility of finding extraterrestrial life, but its greater value lies in the fact that it has proven for the first time that large-scale grid and distributed computing are realistic, expandable, and effective.
As a matter of course, SETI began to accept gravitational wave verification, AIDS physiological principles and drug research (FightAIDS@home), particle accelerator design (DPAD project), protein internal structure research (Folding@home project), cancer breakthrough, etc. that require huge data analysis. Cutting-edge scientific research projects provide them with cheap and top-value computing power.
In 2016, Russian billionaire Yuri Milner invested 100 million U.S. dollars in a large-scale search operation of alien intelligent life initiated by the late British astrophysicist Stephen Hawking-breakthrough listening program, Also mainly promoted by SETI.
Every research here is enough to have a realistic impact on the survival of human society. What is very dramatic is that all the prerequisites for realizing all these computing power are actually scattered around the world, those "not worth mentioning", and even "forgotten" by the owner.
The birth of BOINC
The distributed computing performance brought by SETI has appeared, and more and more projects have found SETI for computing power services. In order to meet these needs more and create a better computing experience. SETI began to iteratively upgrade, and in 2003, the BOINC platform was born.
The founder who led the development of the BOINC distributed computing platform was also the assistant professor David Anderson. It's just that his current identity has become the founder of the BOINC project and a scientist in the Space Science Laboratory of UC Berkeley University.
SETI@home, initiator of the BOINC project, David Anderson, scientist at UC Berkeley University Space Science Laboratory
BOINC (Berkeley Open Infrastructure for Network Computing, Berkeley Open Infrastructure for Network Computing) is a distributed computing platform upgraded from SETI. One of the reasons why BOINC development is very important is to attract more users to join this meaningful distributed computing project.
In the same vein as SETI, the BOINC network acts as a platform for aggregating and sharing the computing resources of global C-end users, and provides researchers from all over the world with powerful computing capabilities that gather a large number of personal computers from all over the world through crowdsourced distribution and matching. In the 16 years since the birth of BOINC, it has gradually gained grounded practice in the fields of mathematics, medicine, astronomy, and meteorology. The platform is distributed in 118 countries around the world, accounting for 50% of the computing power of the world's scientific volunteer computing projects. It is the world's earliest distributed computing network, earlier than all blockchain distributed networks including Bitcoin.
"The main goal of BOINC was to create a source of cheap computing power for scientists." David Anderson, the founder of the BOINC project and a scientist at the UC Berkeley University Space Science Laboratory, introduced to Linkde.
Using voluntary home computers, each scientific experiment project was able to obtain computing power equivalent to that of the largest supercomputer at the time. For example, one of the early projects using BOINC is Climateprediction.net from Oxford University. The project uses BOINC computing power to predict global climate change in the next 100 years. Prior to this, due to lack of computing power, the project had never been completed.
"Barbarians" who strayed into the cloud computing battlefield
The founder of Softbank, Sun Zhengyi, recalled that in the past 30 years, three things were crucial to the Internet technology industry: CPU computing power, memory size, and communication speed. These three things have improved a million times, and each has improved a million times. It has a huge impact on the entire human society.
In 2020, more than 50 billion machines and equipment will be interconnected worldwide, and more than 200 billion connected sensors will generate massive amounts of data. In this context, computing power will make artificial intelligence a reality.
Big data applications, the Internet of Things, 5G, blockchain, chip research and development, and intelligent infrastructure construction all rely on the basic support provided by the large-scale growth and utilization of computing power. For companies with data processing needs in the future, the basic scope covered by the “new infrastructure” of cloud computing will inevitably result in the former's dependence on the latter's computing power services. Therefore, cloud computing services have exploded in huge commercial efficiency and value in the past five years. Almost all technology giants in the world have not ignored this major layout.
According to the IDC report, 90% of global information data are generated in recent years. Also in 2020, about 40% of the information will be stored by cloud computing service providers, and one third of the data will be valuable. Amazon AWS, which accounts for 51.8% of global cloud computing services, as the world's largest cloud service provider, has laid the most solid prerequisite for Amazon's transition to a great enterprise. Microsoft regained its vitality for innovation and returned to surpass Apple in market capitalization. The biggest change occurred in the cloud service transformation advocated by CEO Nadella. From the existing city of traditional Windows software services, the balance of resources has been gradually transferred to Microsoft Azure (cloud service) business like nails. Microsoft is continuing to catch up with the goal of global computing engine by using B-side services of computing power.
Among Chinese companies, Alibaba Cloud has close to 20% of the Asia-Pacific cloud computing market, which directly opens a new window for the imagination of Alibaba Group. Whenever I see Alibaba Cloud's slogan on a huge airport poster: "For incalculable value." Everything is in it.
One of the cores of cloud computing competition is computing power. Whoever has the strongest, most stable, cheapest, and easiest-to-use computing power service can take the lead in this data enclosure movement. Therefore, all cloud service providers are building and deploying their own servers, looking for areas with cheap energy and land, in order to reduce the cost of computing power as much as possible.
Compared with the centralized cloud services of large Internet technology giants building deep cities, distributed networks have visible space in the subdivision field. Like the impact of edge computing on cloud computing, BOINC follows a set of "distributed thinking" of overtaking in curves.
From virtual machines to long-term data storage, the centralized cloud relies on servers in tall IDC computer rooms, which are particularly delicate and have special requirements for temperature, humidity, and static electricity. In addition to the huge expenses for server equipment, space, energy, buildings, etc., a large number of permanent maintenance personnel is a cost that ordinary enterprises cannot afford. As a result, cloud computing services, or computing power services in particular, require high fees, which discourages most companies and technology research projects, leading to delays in project progress and even failure.
Good quality and low price, when the volunteers contribute, the computing power of the supercomputer can be used accurately, which is the inherent service advantage of BOINC. Soon, electronic consumer devices will have processors, sensors and Internet connections. They are also low-power and inexpensive, and can be shared and utilized by the BOINC platform. Cloud computing is very expensive, and locally available computing resources are nearby and free.
In this sense, BOINC can be regarded as a type of global cross-industry data exchange and collaborative computing infrastructure, providing new edge computing solutions.
Like human history, the history of computing has always been a conflict between centralization and decentralization. There used to be large computers, now there is a central cloud. They are centralized and controlled by one organization. In the long run, distributed methods tend to win.
Just as Zuckerberg, who once stood at the hearing of the U.S. House of Representatives, still makes many organizations and providers and owners of personal privacy data anxious. The security of core and confidential data has become the primary reason why many companies do not accept Alibaba Cloud and Tencent Cloud services.
In the process of thinking about these factors, the team of BOINC initiator Professor Anderson began to pay attention to an idea: how to use the BOINC network to provide a subset of AWS (Amazon Cloud Service) at a lower cost. It's not easy, Professor Anderson said frankly to the chain. Because BOINC host can disappear at any time, but there are some algorithms that can solve this problem.
"Distributed thinking" is back
The core thinking of BOINC system is distributed.
Distributed thinking is not born with hot words such as blockchain and edge computing. It originated from the most basic fundamental ideas of Internet technology and is one of the core principles of the Internet world.
In the fall of 1988, Robert Morris, a first-year graduate student at Cornell University, successfully released the "Morris worm", the worm that caused widespread panic and harm in the past with only 99 lines of programs. But when it reached the peak of infection, this "worm" did another thing. It infects thousands of machines, and what is unexpected is that these machines operate in concert at the same time. According to later calculations, Morris' "worm" became the most powerful parallel computer in history during an unforgettable 48-hour survival period. When such parallelism reached its peak, it successfully achieved a processing speed of 400 billion operations per second, twice the speed of the most expensive supercomputer at present.
To a certain extent, worms are the pioneers of distributed computing. In that era of Internet rash, they used "distributed thinking" to create a peak of computing power with an unethical method.
According to the chain, the last thing Zou Sheng, deputy general manager of the original Shenzhen Stock Exchange, completed before leaving the Shenzhen Stock Exchange was to completely transform the data architecture and background processing methods of the Shenzhen Stock Exchange into a distributed structure.
The new generation trading system adopts a distributed large-scale parallel processing architecture based on a high-speed message bus. This architecture has the characteristics of high availability and strong scalability, and is consistent with the mainstream architecture of pioneering systems such as Nasdaq and German Exchange. The architecture adopts an open platform and open source technology, and its basic hardware equipment no longer needs a mainframe with a price of more than 100 million yuan per unit. It can run with cheap domestically produced ordinary servers, and the technical cost is only one-third of the traditional closed platform.
Due to such a distributed architecture, the processing speed of the new-generation trading system of the Shenzhen Stock Exchange is significantly faster than that of the previous generation, and is the first to solve the low-latency transactions in the large-scale retail market. Even users in the northwest region far away from the Shenzhen Stock Exchange system have received a huge increase in network processing speed.
The thinking origin of BOINC and blockchain
Whether Satoshi Nakamoto used BOINC as the source of thought for the POW consensus mechanism when he created the Bitcoin network, we do not know. Just inferring from the time sequence, the BOINC platform, which has been popular in the network technology circle in 2003, is too much in line with the Bitcoin white paper released on November 1, 2008 in terms of distributed computing ideas and computing power contribution incentive system.
Both BOINC and blockchain are composed of distributed computer networks, and both have a large amount of computing power. The difference is that the blockchain is completely decentralized and uses P2P communication. In BOINC, although calculation is distributed, control is not. The “jobs” obtained by the volunteers are sent from the central server (operated by the project), and the results are returned to those servers. All communication is from client to server.
"But this connection (BOINC and blockchain) will expand."
Professor Anderson, the founder of the BOINC project, is promoting the further development of BOINC. In his view, BOINC is a general framework for aggregating computing resources. The current work is centralized scheduling, he plans to convert BOINC into a P2P network with decentralized communication, computing and storage. In this case, a blockchain that distributes computing tasks (completed using smart contracts) and records all transaction information in the BOINC network is very suitable for this scenario.
In addition, because blockchain and BOINC have a lot of computing power. Then people can have a blockchain ledger whose proof of work is based on BOINC calculations. In this case, the network can reward contributors.
In contrast to the blockchain industry’s pursuit of network rewards, since the birth of the Bitcoin POW consensus mechanism, there have been endless questions about the waste of resources, idle computing power, and energy consumption caused by the “mining” behavior of computing power. In August 2017, BCH forked BTC on the grounds of capacity expansion, and the power parties based on the control of computing power established a different interest pattern; the computing power war that broke out in 2018 when BCH itself forked, called a large number of calculations in other currencies Struggling to maintain the benefit of the city became one of the incentives for the currency market to continue to bottom out; in 2019, the National Development and Reform Commission issued the "Industrial Structure Adjustment Guidance Catalog (Opinion Draft)", which classified virtual currency mining activities as a series of eliminating backward production capacity. Prior to this, government policies, including major mining towns in Sichuan, Yunnan, Xinjiang, and Inner Mongolia, have gradually eliminated virtual currency mines. Computing power, as an infrastructure that maintains the operation of industry systems and exchanges of value, faces five types of threats such as power, commercial inefficiency, simplification of functions, narrowing of revenue, and regulatory policies.
In the current mainstream public chains that are based or partly based on the POW consensus mechanism, the power of computing power always affects the survival of the entire network and the balance of transaction prices. On the one hand, this kind of computing power promotes the expansion and development of the industry, and on the other hand, due to changes in the interest structure, a "computing power war" may break out at any time, thereby engulfing the industry and users as cannon fodder. At present, the power in the mainstream computing power system of the blockchain is roughly divided into three levels: (1) large mining pool alliances and large mining machine manufacturer groups; (2) Internet giants and large technology companies with large IDC centers; (3) mainstream Blockchain companies and fragmented mining communities.
At the beginning of 2018, public opinion focused on a short-term question. Has the mining alliance of Bitmain approached or possessed the ability to launch a 51% "double-spending" attack on the BTC network? Regardless of whether the cost of launching the attack makes the attack itself a reality, even if you don’t do so, in a classic decentralized network, because of the formation of computing power hegemony, it will develop into such an attack worry, which breaks the distribution itself. The basic consensus of the existence of the type network. Using BTC expansion as an excuse, BCH was born through an interest bifurcation based on computing power hegemony, including the subsequent BCH bifurcation, which confirmed that a group with a clear interest orientation was easily slipped into the abyss of moral centralization in computing power wars.
Due to the rise of "mining" in the blockchain industry, the concept and role of computing power has been further popularized. Many of the voices of blockchain distributed networks come from the mining industry, but not only that. A reasonable game network first requires the participation of several benchmark stakeholders.
Imagine a possibility: if there is a community that can obtain mining revenue by contributing computing power, and also introduce these computing power into the real economy and scientific research community. Preventing the idle consumption of computing power used for "mining" can efficiently motivate community contributors. So what will be the effect?
It seems that it also guides the commercialization direction of a blockchain reality combined with the physical industry: shared computing power commercial and supercomputing applications.
"Edge" idle computing power can also be realized
Currently, about 40 scientific projects around the world are using BOINC. The entire network contains more than 150,000 volunteers and 650,000 computers. These computers generate 30 "PetaFLOPS" computing power per second, or 3 billion floating-point arithmetic operations, which is equivalent to the fifth-ranked supercomputer in the world. If the computing power of the Folding@home project previously run by SETI is included, all the volunteers in the community will have more computing power than any supercomputer.
According to a Reuters report on July 30, CSCNT Inc. (California Super Computing Network Technology Inc), a California-based start-up company, is signing an agreement with the University of California, Berkeley, deciding to cooperate in the development of cloud computing based on idle computing resources. Service software system. It is planned to upgrade the BOINC system and introduce blockchain technology and a token economic model. This project is called "BOINC Computing Power Earth Project".
Professor David Anderson, the founder of the BOINC platform and a famous UC Berkeley grid computing scientist, will lead part of the research and development of the "BOINC Computing Power Earth Project".
As a computing platform, BOINC provides a unified client program for volunteers. The client itself does not perform actual computing work, but only provides management functions. Volunteers have joined computing projects on the BOINC platform After that, the client program will automatically download the new task unit and call the calculation program of the corresponding project for calculation. If you participate in multiple projects, it will automatically allocate computing resources between each project according to the user's settings. After the calculation is completed At the same time, it will also automatically upload the calculation results and obtain a new calculation unit at the same time.
"Everything is done automatically by BOINC." Professor Anderson, the founder of the BOINC project, said to the chain.
BOINC regularly communicates with the project server to obtain jobs, download programs and input files, and upload output files of completed jobs. It runs jobs in the background with low priority, so volunteer users will not notice anything. It mobilizes its GPU when volunteer users are not using it. Volunteers can configure BOINC functions as needed, for example, limit the amount of disk space or memory it uses.
In general, BOINC can be regarded as a system for (1) using computer capacity (computation, storage, network bandwidth) as much as possible without the user’s attention; (2) fairness among multiple competing projects Divide resources; (3) accurately calculate the amount of resources used by each project.
The "monetization" of widely idle computing power is a feature that BOINC was born close to commercial demand. Using a distributed sharing method with low cost and high computing power, the redundant, ineffective, and decentralized computing power can be used to concentrate the value of physical support. Fifteen years have passed since its inception, and the demand for computing power in the whole society is far more hungry than in scientific research. Although BOINC's original intention is to support scientific computing, its urgent need to expand the commercial and industrial value space is obviously beyond the original scope. Any form of large-scale distributed computing can be used to achieve significant advancement, such as AI, IOT training and computing load.
Throughout the many branches of AI, the most promising form is considered to be "deep learning." It involves the use of data to train neural networks, which will take up a lot of computing resources as the amount of data and network scale grow.
"When you add the goal of automated network topology design, I think this is a great task that is very suitable for BOINC." Anderson said that many methods in artificial intelligence involve "biologically inspired" algorithms that can simulate animal Evolution and plant behavior. The entire region is in its infancy, and BOINC can help it grow.
The three motivational motivations of the sharing system: interest, feelings and benefits
As mentioned above, BOINC can manage programs that use its GPU under the authorization of volunteers. Therefore, in the previous years, people have been using BOINC to mine Bitcoin. Professor Anderson did not reject this, but deepened his understanding and interest in Tokenlization, the most innovative invention since the birth of the blockchain.
"In the concept of cryptocurrency, the proof-of-work mechanism (POW consensus) is a useful calculation. Such as scientific research, rather than a meaningless hash function. There is already a point incentive mechanism for Gridcoin. I think this is A great way to reward volunteers (who pay for electricity) and attract new volunteers. This can increase the volunteer base by 100 or 1,000 times, thereby creating a new kind of computing power that can revolutionize many fields of science.” Anderson clearly sees The practical direction of expanding the community.
At the same time, it has the actual value of points export, commercial landing and monetization channels, a large number of C-end user bases, shared circulation of global communities, easy-to-operate low-threshold computing power contribution channels, autonomous operations of decentralized communities, and easy-to-incentivize potential community expansion With the demand and expectation of changing hands with flexibility and high transaction frequency, BOINC naturally has the basic conditions for the transformation of all token systems.
Once all the rage in the country, Xunlei's Wangxin Technology launched a shared computing intelligent hardware player cloud for individual users, and it has been practiced in a similar way of thinking. At that time, Xunlei expected countless individual users to share idle bandwidth, computing, storage and other resources at home with the help of Playke Cloud, and eventually be converted into shared computing services. Wankeyun was once sold out, and its crowdfunding share successfully ended with a 100 times over-subscription within ten days. As the circulation token in the play cloud system, the price of over-the-counter transactions was so high that it was difficult to restrain it at that time, which eventually aroused the attention of the supervisory authorities and stopped. But this incident did not extinguish the huge enthusiasm of domestic blockchain entrepreneurs for shared computing and shared CDN business, but instead opened a window of application ideas.
Although it is the shared logic of distributed computing and idle resource revitalization, BOINC is naturally different from Wanke Cloud. First of all, Wanke Cloud is essentially a personal cloud disk without GPU computing capabilities. The goal of shared bandwidth storage is different from BOINC's computing power integration application. At the same time, BOINC, as a software terminal service, is not a hardware sale. It has higher scalability and experience in terms of participation threshold, interactivity, operability, flexibility and basic cost. Secondly, at that time, the value and use of LinkToken were narrow, sharing bandwidth. After providing shared storage and helping to run computing tasks, users received LinkToken rewards, which could be exchanged for limited iQiyi members and the cloud of the Star Controls platform. service. BOINC has supported more than one hundred scientific research project cases in existing cases, and can directly connect with the cost-effective computing power services of commercial projects. There is room for advantage in the usage scenarios and the breadth of consensus.
Finally, BOINC has more than ten years of international community operations, with millions of associated users all over the world. There is no centralized business operation entity and manager, and as an online terminal product, there is no unique geographical user. Attribution, and the user space in the Chinese market has not yet been developed.
According to the statistics of BOINC STATS in 2009, the utilization rate of BOINC in mainland China is at a low position. In Taiwan and Hong Kong, the per capita contribution rate is higher. The country with the highest points is the United States.
In order to encourage and give back to volunteers who have made contributions, BOINC's point system has a long history.
Because the projects that can be run on the BOINC platform are very different, for example, the task unit of project A takes 3 hours to complete in a certain machine, and the calculation package of project B takes 30 hours to complete in this machine, or project A itself The completion time of the task package has a certain floating range. Obviously, it is no longer feasible to use the number of task units to measure the amount of calculation contributed by the user; similarly, there are differences in machine performance, and the CPU time required to complete the task unit is used to measure the work. The amount is not enough.
As a general computing platform, the scoring system of the BOINC platform must pass certain algorithms to more accurately record the amount of calculation actually completed by the user.
In this regard, Professor Anderson said to the chain, “We have put in a lot of effort on the credit system, which basically tracks how many floating-point operations you have completed. We must make it suitable for GPU and CPU, and we must use It prevents deception."
In order to ensure the normal function of the points, the team of Professor Anderson is creating a use case of reward points to stabilize the price. In this regard, it is a very good way to develop AWS (Amazon Cloud Services) style services on BOINC and allow customers to use points to purchase web services. The research is already underway.
In addition, there are many users in the BOINC community who are not concerned about the value of points trading in principle of incentives. They are more scientifically inspired and attracted. Most of the early volunteers who participated in the BOINC network were students from top universities in North America and Europe.
"Such a basic user structure determines that most users do not value price. They are more motivated by scientific goals, such as providing direct support for anticancer drugs or gravitational wave research on the BOINC platform by providing their own computing power. This sense of accomplishment is beyond words." Professor Anderson's pride is beyond words.
From searching for intelligent life to changing the problems that human beings face urgently. The story of BOINC originated from a sci-fi movie-like imagination. This vision has changed infinite possibilities because of the limited free computing power controlled by everyone.
From a commercial point of view, on the side of blockchain, AI, and cloud computing, an original ecological "edge" revolution is probably already on the way.
From the perspective of mission, the scientific research practice that is eager to use computing power to overcome major issues such as AIDS, cancer, malaria, and climate change continues to take root. (This article is exclusively published by Titanium Media Author/Li Feifan)
We only provide valuable dry goods!
Long press the QR code