Across technology. Учебное пособие

overload – перегрузка

cellular phone – сотовый телефон dial up – набирать номер, звонить answering machine – автоответчик handle, v – обращаться, иметь дело с amount – количество

fiber optic cables – оптоволоконный кабель opposed – зд. вместо

binary code – бинарный код, двоичный код amount of bandwidth – пропускная способность coaxial cable – коаксильный кабель

increase 1,000 fold, v – увеличить(ся) в 1000 раз lane – линии движения

clog – засорение, препятствие

The Future of the Internet

Everywhere we go, we hear about the Internet. It's on television, in magazines, newspapers, and in schools. One might think that this network of millions of computers around the globe is as fast and captivating as television, but with more and more users logging on every day and staying on longer and longer, this «Information Superhighway» could be perhaps more correctly referred to as an expressway of big city centre at rush hour.

It is estimated that thirty five to forty million users currently are on the Internet. According to a recent statistics, an average Internet call lasts five times as longer as the average regular telephone call. 10 percent of the Internet calls last 6 hours or longer. This can cause an overload and, in turn, cause telephone network to fail.

The local network was designed for short calls which you make and then hang up, but Internet calls often occupy a line for hours. With so many users in the Internet and their number is growing by 200 percent annually, it certainly provides new challenges for the telephone companies. The Internet, up to the beginning of the 90s, was used only to read different texts. Then in the early 90's, a way was made to see pictures and listen to a sound on the Internet. This breakthrough made the Internet to be most demanded means of communication, data saving and transporting.

However, today's net is much more than just pictures, text, and sound. The Internet is now filled with voice massages, video conferencing and video games. With voice massages, users can talk over the Internet for the price of the local phone call.

Nowadays we no longer have to own a computer to access the Internet. Now,-devices such as Web TV allow our television to browse the Web and use Electronic Mail. Cellular phones are now also dialing up the Internet to provide E- mail and answering machine services. The telephone network was not designed and built to handle these sorts of things. Many telephone companies are spending enormous amounts of money to upgrade the telephone lines.

41

K. Kao and G. Hockman were the first to come up with the idea of using fiber optic cables, as opposed to copper wire, to carry telephone signals. Fiber optics uses pulses of light to transmit binary code, such as that used in computers and other electronic devices. As a result the amount of bandwidth is incredibly raised. Another solution for the problem is fast modems which satisfy the need for speed.

By accessing the Net through the coaxial cable that provides television to our homes, the speed can be increased 1,000 fold. However, the cable system was built to only send information one way. In other words, they can send stuff to us, but we can't send anything back, if there is no modem available.

Yet another way is being introduced to access the Internet, and that is through the use of a satellite dish just like the TV dishes currently used to deliver television from satellites in space to your home. However, like cable connection, the information can only be sent one way.

Faster ways of connecting to the Internet may sound like a solution to the problem, but, just as new lanes on highways attract more cars, a faster Internet could attract many times more users, making it even slower than before.

To help solve the problem of Internet clogs, Internet providers are trying new ways of pricing for customers. So, in business time any connection to Net cost more than your connection in the night.

In conclusion, I should add that if we want to keep the Internet usable and fairly fast, we must not only improve the telephone lines and means of access, but also be reasonable in usage.

Answer the questions after reading the text:

1.How many users are currently in the Internet?

2.How long does an average Internet call last?

3.What can cause the overload of the telephone system?

4.What was the main purpose of the Internet up to the 90s?

5.Do we need to have a computer to get access to the Internet today?

6.Who was the first to come up with the idea of using fiber optic cables?

7.What is the alternative way to get access to the Internet today?

42

Приложение

Cтатья 1

The Great Transmogrification of Atoms to Bits

Libraries move to curating digital rather than physical collections

By Paul McFedries

As I mentioned in my February 2013 column, “Balancing Act,” the belief that our life offline is separate from our life online has been denounced as digital dualism. But there’s less of a debate when it comes to differentiating between analog objects and digital data. Yes, the print and electronic copies of the same book contain the same words, but it’s obvious to most people (and, increasingly, to researchers) that the two reading experiences are quite different.

We need to understand such differences because the world is going to see a lot more digital data in the near future. This includes born-digital [pdf] data, which is originally created in an electronic format, as well as born-analog data, which starts life as a physical object and then is reborn digital. A great example of this digitization came earlier this year when the New York Public Library announced that it was making more than 180,000 digitized items available to anyone with an Internet connection, no questions asked.

That librarians would turn themselves into digital curators is no surprise, since as analog curators for the past few centuries they have been constantly bumping into the physical constraints of storage space and material decay. One approach is to get rid of stuff, and librarians and archivists employ a pleasing variety of terms related to the removal of unwanted or duplicate material from their collections: Weeding and culling generally refer to the removal of individual items, while purging, screening, and stripping are most often used for the removal of multiple related items. But the main problem with physical materials is that they possess what archivists call, poetically, inherent vice: the tendency for something to deteriorate over time because of some fault in the material itself (for example, the presence of lignin in cheap paper, which causes the paper to yellow) or the way the material reacts with its surroundings (for instance, the fact that bugs eat some books because they’re attracted to the mold that grows in damp paper).

The digitization of analog materials can solve these problems, and engineers are constantly trying to find faster ways to turn atoms into bits. For now, though, we mostly have to rely on the skills of scanops (scanner operators) to generate those bits, although on their less skilled days those operators end up scanning their own body parts, such as fingers and hands, a phenomenon known as Google hands. Some companies are applying the principles of crowdsourcing and gamification to the digitizing realm, creating leisure activities that let users contribute to the process. (I would be remiss if I didn’t mention the opposite process: turning digital Web documents and data into books and zines, a genre called the printed Web.)

Ideally, digitized data is online (readily available), but it might end up either offline (not available) or nearline (only indirectly available). It can also end up in dark archives (which are inaccessible to the public), dim archives (which are usually inaccessible but can be made accessible), or light archives (another term for those that are fully accessible).

43

Having digitized some data, the archivist now faces a new problem: the eventual obsolescence of the data structures or media used to store the data, necessitating a format migration (or a media migration) to something newer. Copying the data without changing the format or media type is called refreshing.

There is a large cottage industry of life coaches and self-appointed gurus who recommend, with varying degrees of urgency and stridency, that we become digital dualists and spend less time online. Fulminations against digitization are harder to find, and that’s just as well, since, with enlightened institutions such as the New York Public Library leading the way, having digital access to books, photos, and other analog materials can only be a good thing. Try to ignore the fingers.

This article appears in the April 2016 print issue as “Curating the Digital

Age.”

Статья 2

The Nervana SystemsChip That Will Let Intel Advance Its Deep Learning

By Jeremy Hsu

Deep-learning artificial intelligence has mostly relied upon the generalpurpose GPU hardware used in many other computing tasks. But Intel’s recent acquisition of the startup Nervana Systems will give the tech giant ownership of a specialized chip designed specifically for deep learning AI applications. That could give Intel a huge lead in the race to develop next-generation artificial intelligence capable of swiftly finding patterns in huge data sets and learning through imitation.

Nervana has leaned heavily on GPU hardware to build its own portfolio of deep-learning AI services for both companies and independent developers. But the startup has also been developing its own specialized deep-learning hardware, called Nervana Engine, that includes only the components necessary for running deep-learning algorithms and eliminates the extra components used for generalpurpose GPU tasks. Nervana claims that when the Engine chip comes out in 2017, it will deliver around 10 times as much computing power for deep learning as the best of today’s GPUs.

“Nervana’s AI expertise combined with Intel’s capabilities and huge market reach will allow us to realize our vision and create something truly special,” said Naveen Rao, CEO and cofounder of Nervana, in a blog post.

Software algorithms known as artificial neural networks are the heart of deep-learning AI. Such algorithms learn how to perform certain tasks through imitation and by observing correctly labeled examples as they sift through huge amounts of data. To accommodate deep learning’s voracious appetite for data, Nervana’s Engine hardware design includes High Bandwidth Memory technology that has stacked memory and densely packed data channels to swiftly move around large amounts of data.

The end result: 32 gigabytes of on-chip storage and up to 8 terabits per second of memory access speed. By comparison, the GDDR5 memory technology used in GPUs has memory access speeds of just 224 gigabits per second.

44

Intel clearly saw value in acquiring both Nervana and its deep-learning chip. Investors familiar with the acquisition deal pegged the startup’s purchase price at somewhere in the range of $408 million, according to Recode.

The purchase gives Intel a possible edge in the market for deep-learning hardware while sidestepping the general-purpose GPUs produced by rival tech giant Nvidia, said Karl Freund, senior analyst for deep learning and HPC at Moor Insights & Strategy, in an interview with EE Times. Intel currently produces multicore Xeon and Xeon Phi processors and other hardware, but has had no equivalent to the GPUs that currently dominate deep learning.

“[Nervana’s] IP and expertise in accelerating deep-learning algorithms will expand Intel’s capabilities in the field of AI,” said Diane Bryant, executive vice president and general manager of the Data Center Group at Intel, in a blog post.

The 48-person Nervana team will remain at its San Diego headquarters and maintain a “startup mentality” as part of the deal. Nervana will also continue developing the Engine deep-learning hardware alongside other existing products such as its Neon deep-learning framework, a programming language and set of libraries intended to help outsiders create deep-learning models.

Intel’s big bet on Nervana signifies the growing importance of deep learning and the broader field of machine learning. The purchase is the latest in a string of deep-learning startup acquisitions by major tech companies such as Google, IBM, and Amazon. To learn more about the race between startups and tech titans to develop deep learning services, see the IEEE Spectrum article “Now You Too Can Buy Cloud-Based Deep Learning.”

Статья 3

Machine Learning Tools Help Google Science Fair Finalists Find Lost Objects,

Predict Breast Cancer Risk

By Tekla S. Perry

Anika Cheerla's submission to the Google Science Fair used machine learning to improve the accuracy of breast cancer risk prediction

This week, 16 teams of teens from around the world assembled in Mountain View to demonstrate the results of research projects at the Google Science Fair.

I’ve been attending these finals for several years now and am always impressed with how creatively the teens use the technologies of today. And this year was no exception: machine learning is hot in the tech world, and the teens are embracing it.

Consider 14-year-old Anika Cheerla’s submission. A Silicon Valley girl from Cupertino, Calif., Cheerla was curious about the current state of breast cancer prediction, and discovered that prediction methods using digital mammograms are just 64 percent effective, typically simply considering the percentage of dense tissue in a breast. She developed software that considers a broader range of features, including dense and non-dense regions, and, using a database of digital mammograms from Stanford University, built and began training classifiers to use in predicting risk. She discovered that the area closest to the nipple has the highest predictive power, and her system can take that into account. Right now her system

45

is about 84 percent effective. She is hoping to improve her system by training it on more images and adding additional machine learning capabilities.

CreditGoogle Science Fair finalist Shriank Kanapurti went through several iterations of his system designed to keep track of objects around you–and help you find your lost keys. He tested the system on his grandfather.

Shriank Kanapurti, a 16- year-old from Bangalore, India, turned to machine learning to help the forgetful find misplaced objects. His approach, called KeepTab, involves a wearable camera constantly recording images of what’s in front of you, and designed software that extracts the objects from the images and figures out what they are in relation to other objects in your environment. To date he says he has trained the software on 600,000 images. He uses Google-Now’s natural language software to communicate with his system–you can say “Locate my keys,” and it will respond, “Your keys are on the television”. He’d like to eventually see his software run with less obtrusive wearables, like a future version of Google Glass.

Статья 4

Software, the Invisible Technology

We used to go to stores to buy it. Now software is so ubiquitous, we don’t even notice it

By G. Pascal Zachary

Five years ago Marc Andreessen, the Web pioneer and celebrated tech investor, predicted software would eat the world.

He turned out to be right. Too right. Software is eating the world, and also eating itself.

The cannibalization of software defies easy explanation. Software is the motor of the world’s digital economy. Code is the ground of our computationally rich existence. Software applications and platforms are the source of vast wealth for Apple, Google, Facebook, Amazon, and many other tech titans. Yet even as software grows in importance, code becomes less visible, less tangible, less understood, and–perhaps most paradoxically–less valuable in monetary terms.

How has this great shift happened? Software originally coevolved with computers themselves. The IBM System/360, the business computer of choice in the 1960s, came bundled with code, and if customers needed more or different programs, they asked IBM. In 1980, IBM chose to rely on outsiders, notably Microsoft, for PC code, igniting a ferocious race to sell programs as distinct products.

Over the past 20 years, the program as artifact has vanished. Consumers download new versions, patches, and feature improvements as easily as switching on their devices. They rarely pay for this. Software battles now occur over platforms, which define an experience such as shopping (Amazon), searching (Google), or networking (LinkedIn). Competitive advantage is now achieved through superior software, but software supremacy is neither the aim nor the result of the new game.

46

“Software on wheels,” for instance, now defines the car of the future, and is the reason why Google and Apple, the reigning kings of code, each are pushing for a big role in next-gen autos. No surprise that Toyota, Ford, and Daimler want their own software expertise, not to peddle programs to drivers but to enhance their new models. Without cool software, these venerable automakers might not even survive (see Tesla’s and Google’s self-driving cars).

View Uber in this same light. The company is best understood not as a taxi service on steroids but as a software-management system for personal transportation.

The bottom line: No one gets rich making software anymore. The days of Bill Gates building a fortune on the strength of shrink-wrapped programs sold like disposable diapers is gone. Software today can make you rich only by enabling you to do something else that people pay for.

Netflix, Facebook, and Google don’t get a dime from selling software, yet their revenue-producing services depend on continuously and seamlessly improving their code. Similar examples are legion.

The new logic of software has different implications for those who make it, sell it, and use it. For makers, code no longer spawns tycoons and celebrities. The last person to get famous from writing software was Linus Torvalds; and this year he celebrates the 25th anniversary of his seminal achievement. Today’s top coders are largely if not wholly unknown by the wider public; at best, they are cult figures, revered in underground communities.

Another paradox: As software becomes ever more essential to creating the digital experience, the invisibility of software is a victory for the apostles of “the free.” Hippies, misfits, and dropouts improbably created fabulous wealth through pricey apps in the last two decades of the 20th century. Some among these characters, notably Richard Stallman, promoted a counter-ethos that conceived of software as a public good, available without charge to all.

That free software sits at the heart of an explosion of profitable digital services is the latest, greatest riddle of global capitalism–and a sobering message for the poets of programming. Code writers are essential and well paid but increasingly interchangeable and anonymous.

For this, we should not pity the programmer, and we should remember: Hope springs from the unseen.

This article appears in the November 2016 print issue as “The Invisible Technology.”

Статья 5

China's New Supercomputer Is World's Most Powerful

By Rachel Courtland

For the last three years, China has topped the Top500 list of the most powerful supercomputers with its massive Tianhe-2. But today, the Top500 group announced that Tianhe-2 has been ousted by another Chinese supercomputer, the Sunway TaihuLight. The new machine, which is based at the National Supercomputing Center in Wuxi, can perform a key benchmark test called Linpack

47

at 93 petaflops (a thousand trillion floating point operations per second)–nearly three times the speed of the Tianhe-2.

The new rankings further solidify China’s status as a supercomputing force to be reckoned with. In addition to this new machine, the United States has, for the first time, lost its status as the country with the most systems on the list; China now has 167 systems to the U.S.’s 165.

Unlike the Tianhe-2, which used Intel Xeon chips to take the top spot, the processors inside the Sunway TaihuLight are homegrown. At each of the machine’s 40,960 nodes, the supercomputer uses a new 260-core chip, designed by the Shanghai High Performance IC Design Center.

According to the Top500 site, Sunway TaihuLight will be used for research and engineering work, including weather modeling and advanced manufacturing.

Although supercomputing progress has slowed in recent years, there are still-more-powerful machines on the horizon. The United States, for one, has a batch of new machines in the works. According to a report on Sunway TaihuLight by Top500 team member Jack Dongarra, 2018 could see the arrival of three new U.S. Department of Energy machines, the speediest of which will be Summit, a 200-petaflop supercomputer to be installed at Oak Ridge National Laboratory in Tennessee.

The top 10 supercomputers from the June 2016 Top500.org list.

Статья 6

Use a GPU to Turn a PC Into a Supercomputer

By Mark Anderson

As Moore’s Law slows for CPUs, dedicated graphics co-processors are picking up some of the slack. Just as GPUs are changing the game in deep learning and autonomous cars, the GPU-powered desktop PC might even begin to keep pace with the conventional supercomputer for a portion of supercomputer applications.

For instance, a group of Russian scientists are reporting this month that they’ve been able to solve computational problems in nuclear physics using an off- the-shelf, high-end PC containing a GPU. And, they say, after fine-tuning their algorithm for GPUs, they were able to run their calculations faster than the

48

traditional, CPU-powered supercomputer their colleagues use. Bonus: they ran those calculations for free as opposed to their colleagues, who must pay for access to the supercomputer to run their computations.

“I have no doubts that many research groups over the world can reach the similar results in their own fields such as geophysics, seismology, plasma physics, [medical] diagnostics, etc.” says Vladimir Kukulin, professor of theoretical physics at Lomonosov Moscow State University. “But only by combining two above ingredients: Reformulation of the whole problem, and then by inventing some effective way how to parallelize the whole execution in thousands or even millions of independent threads.”

The problem Kukulin’s group was tackling involved the extensive calculations needed to describe scattering problems in their field – such as when one nucleon collides with a particle or another nucleon and produces a spray of particles and daughter nuclei as a result. This nuclear many-body problem, Kukulin says, could require calculations involving matrices containing millions of elements.

Matrix algebra with this many moving parts can stymie even a supercomputer. But, Kukulin says, his group realized, too, that calculations with giant matrices can mean independent threads of instructions to run in parallel with many other similar threads. The parallelizability of his group’s nuclear calculations, in other words, meant it was a prime candidate for running efficiently on a GPU.

The GPU, originally designed to handle matrix-heavy calculations needed to generate real-time graphics, is finding unexpected applications in a number of fields today, including Bitcoin mining, molecular modeling, and the applications noted above. Kukulin says the list of computational tasks the GPU can handle, a list that now includes nuclear physics, will only increase.

Overall, he says the kinds of problems that could lend themselves to GPUsupercomputing on the cheap are those whose individual elements are not interdependent on one another. Because interdependence means that individual elements (i.e. threads in a GPU’s calculations) would have to go through regular ifthen logic gates, checking for each element’s ongoing influence on other elements of the calculation.

And such conditional logic steps are probably going to involve the CPU, which slows an otherwise streamlined calculation down. Instead, to maximize the speedup a GPU can produce, he says it’s best to find a way of expressing one’s problem – or find an approximation that enables the expression – as a system containing many discrete and unconnected elements.

“You should write your problem into a form that allows you to massively parallelize,” he says. “It’s necessary to avoid somehow any conditional.”

So complicated simulations, where every component’s interaction is dependent on other components and their trajectories in time, would be difficult to translate into a GPU-ready problem. By contrast, he says, tsunami early warning systems that predict tidal wave landfalls in faster-than-real-time have sped up when run on GPUs as opposed to traditional CPU supercomputers.

49

Kukulin says 3D ultrasound imaging, a compute-intensive medical diagnostic tool, could be more widely embraced if medical offices needed only a few thousand dollars for a desktop PC as opposed to many thousands of dollars for access to a supercomputer.

Ultimately, then, problems fit for GPU speedup require not just programming finesse but also expertise in the field of application to find the right way through the problem that could produce a GPU speedup.

“This is some art, but not [just] in programming,” Kukulin says.

Статья 7

3D-Printed PlasticBlocks Generate Complex Acoustic Holograms

By Evan Ackerman

The closest thing we have right now to a Star Trek–style tractor beam is a technology based on moving small objects with sound. Last year, researchers from the Public University of Navarre, in Spain, demonstrated how ultrasonic acoustic holograms can be used to manipulate things in midair, using arrays of ultrasonic transducers and some reasonably complicated modeling and programming. The overall complexity of the acoustic hologram–a 3D structure made of sound–that you can create in this way (and consequently what you can do with it) is limited primarily by the characteristics of your transducer array, and because transducers can only get so small, this is a significant limitation.

An acoustic hologram manifests itself as variations in air pressure. You can make pressure structures like vortices and bottles, which can trap small, lightweight things in areas of low pressure inside areas of high pressure. Creating the structures involves the creation of a sound field where a bunch of different sound waves of varying amplitudes constructively and destructively interfere in just the right way to make exactly the structure that you want. One way to do this is with an array of individual transducers, each one emitting a slightly different sound wave.

But in a paper published in Nature this week, a team from the Max Planck Institute for Intelligent Systems, in Germany, describe a new way of easily creating very high resolution acoustic holograms that work in air or water. Rather than relying on a whole bunch of small transducers, they use just one giant transducer instead. It sits underneath a special 3D-printed transmission hologram made out of finely contoured plastic.

In this new research, the single transducer emits one type of sound wave, which means that you can't use it to create a sound field that'll do all that much for you. What the researchers realized, however, is that the only important thing is finding a way to generate all of those different sound waves–which you can do without more than one transducer if you're clever and willing to make a few compromises.

The trick is to use a 3D-printed hunk o’ plastic. Or, to get needlessly technical, a “3D-printed monolithic element.” Or, to get less needlessly technical, “a finely contoured solid plastic block.” The block is attached to the transducer, and when it transduces, the sound wave has to propagate through the block before

50