HAL 9000

From Wikipedia, the free encyclopedia

HAL 9000
Space Odyssey character
Artist's rendering of one of HAL 9000's interfaces.
First appearance	Novel: 2001: A Space Odyssey (1968) Film: 2001: A Space Odyssey (1968)
Last appearance	Novel: 3001: The Final Odyssey (1997) Film: 2010: The Year We Make Contact (1984)
Created by	Arthur C. Clarke Stanley Kubrick
Voiced by	Douglas Rain
Information
Nickname	HAL
Species	Artificial intelligence Computer
Gender	N/A (male vocals)
Relatives	SAL 9000

HAL 9000 is a fictional character and the main antagonist in Arthur C. Clarke's Space Odyssey series. First appearing in the 1968 film 2001: A Space Odyssey, HAL (Heuristically programmed ALgorithmic computer) is a sentient computer (or artificial general intelligence) that controls the systems of the Discovery One spacecraft and interacts with the ship's astronaut crew. Part of HAL's hardware is shown towards the end of the film, but he is mostly depicted as a camera lens containing a red or yellow dot, instances of which are located throughout the ship. HAL 9000 is voiced by Douglas Rain in the two feature film adaptations of the Space Odyssey series. HAL speaks in a soft, calm voice and a conversational manner, in contrast to the crewmen, David Bowman and Frank Poole.

In the film, HAL became operational on 12 January 1992 at the HAL Laboratories in Urbana, Illinois as production number 3. The activation year was 1991 in earlier screenplays and changed to 1997 in Clarke's novel written and released in conjunction with the movie. In addition to maintaining the Discovery One spacecraft systems during the interplanetary mission to Jupiter (or Saturn in the novel), HAL is capable of speech, speech recognition, facial recognition, natural language processing, lip reading, art appreciation, interpreting emotional behaviours, automated reasoning, and playing chess.

Appearances

2001: A Space Odyssey (film/novel)

HAL became operational in Urbana, Illinois, at the HAL Plant (the University of Illinois' Coordinated Science Laboratory, where the ILLIAC computers were built). The film says this occurred in 1992, while the book gives 1997 as HAL's birth year.

In 2001: A Space Odyssey (1968), HAL is initially considered a dependable member of the crew, maintaining ship functions and engaging genially with its human crew-mates on an equal footing. As a recreational activity, Frank Poole plays against HAL in a game of chess. In the film the artificial intelligence is shown to triumph easily. However, as time progresses, HAL begins to malfunction in subtle ways and, as a result, the decision is made to shut down HAL in order to prevent more serious malfunctions. The sequence of events and manner in which HAL is shut down differs between the novel and film versions of the story. In the aforementioned game of chess HAL makes minor and undetected mistakes in his analysis, a possible foreshadowing to HAL's malfunctioning. 

In the film, astronauts David Bowman and Frank Poole consider disconnecting HAL's cognitive circuits when he appears to be mistaken in reporting the presence of a fault in the spacecraft's communications antenna. They attempt to conceal what they are saying, but are unaware that HAL can read their lips. Faced with the prospect of disconnection, HAL decides to kill the astronauts in order to protect and continue its programmed directives. HAL uses one of the Discovery's EVA pods to kill Poole while he is repairing the ship. When Bowman, sans space helmet, uses another pod to attempt to rescue Poole, HAL locks him out of the ship, then disconnects the life support systems of the other hibernating crew members. Bowman circumvents HAL's control, entering the ship by manually opening an emergency airlock with his service pod's clamps, detaching the pod door via its explosive bolts. Bowman jumps across empty space, reenters Discovery, and quickly re-pressurizes the airlock. 

While HAL's motivations are ambiguous in the film, the novel explains that the computer is unable to resolve a conflict between his general mission to relay information accurately, and orders specific to the mission requiring that he withhold from Bowman and Poole the true purpose of the mission. (This withholding is considered essential after the findings of a psychological experiment, "Project Barsoom", where humans were made to believe that there had been alien contact. In every person tested, a deep-seated xenophobia was revealed, which was unknowingly replicated in HAL's constructed personality. Mission Control did not want the crew of Discovery to have their thinking compromised by the knowledge that alien contact was already real.) With the crew dead, HAL reasons, he would not need to lie to them. 

In the novel, the orders to disconnect HAL come from Dave and Frank's superiors on Earth. After Frank is killed while attempting to repair the communications antenna he is pulled away into deep space using the safety tether which is still attached to both the pod and Frank Poole's spacesuit. Dave begins to revive his hibernating crew mates, but is foiled when HAL vents the ship's atmosphere into the vacuum of space, killing the awakening crew members and almost killing Bowman, who is only narrowly saved when he finds his way to an emergency chamber which has its own oxygen supply and a spare space suit inside. 

In both versions, Bowman then proceeds to shut down the machine. In the film, HAL's central core is depicted as a crawlspace full of brightly lit computer modules mounted in arrays from which they can be inserted or removed. Bowman shuts down HAL by removing modules from service one by one; as he does so, HAL's consciousness degrades. HAL finally reverts to material that was programmed into him early in his memory, including announcing the date he became operational as 12 January 1992 (in the novel, 1997). When HAL's logic is completely gone, he begins singing the song "Daisy Bell" (in actuality, the first song sung by a computer, which Clarke had earlier observed at a text-to-speech demonstration).^[4][5][6] HAL's final act of any significance is to prematurely play a prerecorded message from Mission Control which reveals the true reasons for the mission to Jupiter.

2010: Odyssey Two (novel) and 2010: The Year We Make Contact (film)

In the 1982 novel 2010: Odyssey Two, HAL is restarted by his creator, Dr. Chandra, who arrives on the Soviet spaceship Leonov. 

Prior to leaving Earth, Dr. Chandra has also had a discussion with HAL's twin, the SAL 9000. Like HAL, SAL was created by Dr. Chandra. Whereas HAL was characterized as being "male", SAL is characterized as being "female" (voiced by Candice Bergen) and is represented by a blue camera eye instead of a red one. 

Dr. Chandra discovers that HAL's crisis was caused by a programming contradiction: he was constructed for "the accurate processing of information without distortion or concealment", yet his orders, directly from Dr. Heywood Floyd at the National Council on Astronautics, required him to keep the discovery of the Monolith TMA-1 a secret for reasons of national security. This contradiction created a "Hofstadter-Moebius loop", reducing HAL to paranoia. Therefore, HAL made the decision to kill the crew, thereby allowing him to obey both his hardwired instructions to report data truthfully and in full, and his orders to keep the monolith a secret. In essence: if the crew were dead, he would no longer have to keep the information secret.

The alien intelligence initiates a terraforming scheme, placing the Leonov, and everybody in it, in danger. Its human crew devises an escape plan which unfortunately requires leaving the Discovery and HAL behind to be destroyed. Dr. Chandra explains the danger, and HAL willingly sacrifices himself so that the astronauts may escape safely. In the moment of his destruction the monolith-makers transform HAL into a non-corporeal being so that David Bowman's avatar may have a companion. 

The details in the novel and the 1984 film 2010: The Year We Make Contact are nominally the same, with a few exceptions. First, in contradiction to the book (and events described in both book and film versions of 2001: A Space Odyssey), Heywood Floyd is absolved of responsibility for HAL's condition; it is asserted that the decision to program HAL with information concerning TMA-1 came directly from the White House. In the film, HAL functions normally after being reactivated, while in the book it is revealed that his mind was damaged during the shutdown, forcing him to begin communication through screen text. Also, in the film the Leonov crew initially lies to HAL about the dangers that he faced (suspecting that if he knew he would be destroyed he would not initiate the engine-burn necessary to get the Leonov back home), whereas in the novel he is told at the outset. However, in both cases the suspense comes from the question of what HAL will do when he knows that he may be destroyed by his actions. 

In the novel, the basic reboot sequence initiated by Dr. Chandra is quite long, while the movie uses a shorter sequence voiced from HAL as: "HELLO_DOCTOR_NAME_CONTINUE_YESTERDAY_TOMORROW".

While Curnow tells Floyd that Dr. Chandra has begun designing the HAL 10000, the 10000 has not been mentioned in subsequent novels.

2061: Odyssey Three and 3001: The Final Odyssey

In Clarke's 1987 novel 2061: Odyssey Three, Heywood Floyd is surprised to encounter HAL, now stored alongside Dave Bowman in the Europa monolith. 

In Clarke's 1997 novel 3001: The Final Odyssey, Frank Poole is introduced to the merged form of Dave Bowman and HAL, the two merging into one entity called "Halman" after Bowman rescued HAL from the dying Discovery One spaceship towards the end of 2010: Odyssey Two.

Concept and creation

Clarke noted that the first film was criticized for not having any characters, except for HAL and that a great deal of the establishing story on Earth was cut from the film (and even from Clarke's novel). Clarke stated that he had considered Autonomous Mobile Explorer–5 as a name for the computer, then decided on Socrates when writing early drafts, switching in later drafts to Athena, a computer with a female personality, before settling on HAL 9000. The Socrates name was later used in Clarke and Stephen Baxter's A Time Odyssey novel series. 

The earliest draft depicted Socrates as a roughly humanoid robot, and is introduced as overseeing Project Morpheus, which studied prolonged hibernation in preparation for long term space flight. As a demonstration to Senator Floyd, Socrates' designer, Dr. Bruno Forster, asks Socrates to turn off the oxygen to hibernating subjects Kaminski and Whitehead, which Socrates refuses, citing Asimov's First Law of Robotics.

In a later version, in which Bowman and Whitehead are the non-hibernating crew of Discovery, Whitehead dies outside the spacecraft after his pod collides with the main antenna, tearing it free. This triggers the need for Bowman to revive Poole, but the revival does not go according to plan, and after briefly awakening, Poole dies. The computer, named Athena in this draft, announces "All systems of Poole now No–Go. It will be necessary to replace him with a spare unit." After this, Bowman decides to go out in a pod and retrieve the antenna, which is moving away from the ship. Athena refuses to allow him to leave the ship, citing "Directive 15" which prevents it from being left unattended, forcing him to make program modifications during which time the antenna drifts further.

During rehearsals Kubrick asked Stefanie Powers to supply the voice of HAL 9000 while searching for a suitably androgynous voice so the actors had something to react to. On the set, British actor Nigel Davenport played HAL. When it came to dubbing HAL in post-production, Kubrick had originally cast Martin Balsam, but as he felt Balsam "just sounded a little bit too colloquially American", he was replaced with Douglas Rain, who "had the kind of bland mid-Atlantic accent we felt was right for the part". Rain was only handed HAL's lines instead of the full script, and recorded them across a day and a half.

HAL's point of view shots were created with a Cinerama Fairchild-Curtis wide-angle lens with a 160° angle of view. This lens is about 8 inches (20 cm) in diameter, while HAL's on set prop eye lens is about 3 inches (7.6 cm) in diameter. Stanley Kubrick chose to use the large Fairchild-Curtis lens to shoot the HAL 9000 POV shots because he needed a wide-angle fisheye lens that would fit onto his shooting camera, and this was the only lens at the time that would work. The Fairchild-Curtis lens has a focal length of 23 mm (0.9 in) with a maximum aperture of f/2.0 and a weight of approximately 30 lb (14 kg); it was originally designed by Felix Bednarz with a maximum aperture of f/2.2 for the first Cinerama 360 film, Journey to the Stars, shown at the 1962 Seattle World's Fair. Bednarz adapted the lens design from an earlier lens he had designed for military training to simulate human peripheral vision coverage. The lens was later recomputed for the second Cinerama 360 film To the Moon and Beyond, which had a slightly different film format. To the Moon and Beyond was produced by Graphic Films and shown at the 1964/1965 New York World's Fair, where Kubrick watched it; afterwards, he was so impressed that he hired the same creative team from Graphic Films (consisting of Douglas Trumbull, Lester Novros, and Con Pederson) to work on 2001.

A HAL 9000 face plate, without lens (not the same as the hero face plates seen in the film), was discovered in a junk shop in Paddington, London, in the early 1970s by Chris Randall. This was found along with the key to HAL's Brain Room. Both items were purchased for ten shillings (£0.50). Research revealed that the original lens was a Fish-eye-Nikkor 8 mm f/8. The collection was sold at a Christies auction in 2010 for £17,500 to film director Peter Jackson.

Origin of name

HAL's name, according to writer Arthur C. Clarke, is derived from Heuristically programmed ALgorithmic computer. After the film was released fans noticed HAL was a one-letter shift from the name IBM and there has been much speculation since that this was a dig at the large computer company, something that has been denied by both Clarke and 2001 director Stanley Kubrick. Clarke addressed the issue in his book The Lost Worlds of 2001:

...about once a week some character spots the fact that HAL is one letter ahead of IBM, and promptly assumes that Stanley and I were taking a crack at the estimable institution ... As it happened, IBM had given us a good deal of help, so we were quite embarrassed by this, and would have changed the name had we spotted the coincidence.

IBM was consulted during the making of the film and their logo can be seen on props in the film including Pan Am Clipper's cockpit instrument panel and on the lower arm keypad on Poole's space suit. During production it was brought to IBM's attention that the film's plot included a homicidal computer but they approved association with the film if it was clear any "equipment failure" was not related to their products.

HAL Communications Corporation is a real corporation, with facilities located in Urbana, Illinois, which is where HAL identifies himself as being activated: "I am a HAL 9000 computer. I became operational at the H-A-L plant in Urbana Illinois on the 12th of January 1992." 

The former president of HAL Communications, Bill Henry, has stated that this is a coincidence: "There was not and never has been any connection to "Hal", Arthur Clarke's intelligent computer in the screen play "2001" - later published as a book. We were very surprised when the movie hit the Coed Theatre on campus and discovered that the movie's computer had our name. We never had any problems with that similarity - "Hal" for the movie and "HAL" (all caps) for our small company. But, from time-to-time, we did have issues with others trying to use "HAL". That resulted in us paying lawyers. The offenders folded or eventually went out of business." 

Influences

The scene in which HAL's consciousness degrades was inspired by Clarke's memory of a speech synthesis demonstration by physicist John Larry Kelly, Jr., who used an IBM 704 computer to synthesize speech. Kelly's voice recorder synthesizer vocoder recreated the song "Daisy Bell", with musical accompaniment from Max Mathews.

HAL's capabilities, like all the technology in 2001, were based on the speculation of respected scientists. Marvin Minsky, director of the MIT Computer Science and Artificial Intelligence Laboratory (CSAIL) and one of the most influential researchers in the field, was an adviser on the film set. In the mid-1960s, many computer scientists in the field of artificial intelligence were optimistic that machines with HAL's capabilities would exist within a few decades. For example, AI pioneer Herbert A. Simon at Carnegie Mellon University, had predicted in 1965 that "machines will be capable, within twenty years, of doing any work a man can do", the overarching premise being that the issue was one of computational speed (which was predicted to increase) rather than principle.

Cultural impact

HAL is listed as the 13th-greatest film villain in the AFI's 100 Years...100 Heroes & Villains.

The 9000th of the asteroids in the asteroid belt, 9000 Hal discovered on May 3, 1981 by E. Bowell, at Anderson Mesa Station, is named after HAL 9000.

HAL was featured in a guest role in the game LEGO Dimensions, where he is summoned by the player in the Portal 2 level to distract GLaDOS. 

Nintendo collaborator and development studio HAL Laboratory, developer of the Kirby video game series and previous Super Smash Bros video game titles, presumably adopted its name as a reference to HAL 9000; however, according to Satoru Iwata in 2012, the name was chosen because each letter precedes the letters of "IBM" in the alphabet.

Hal is referenced in the popular webcomic Homestuck as "A.R./Auto-Responder". A.R., who later renames himself Lil'Hal, is an artificial intelligence (AI), and program created to mimic the character Dirk Strider, and talk in his stead when he is indisposed.

Supercomputer

From Wikipedia, the free encyclopedia

The IBM Blue Gene/P supercomputer "Intrepid" at Argonne National Laboratory runs 164,000 processor cores using normal data center air conditioning, grouped in 40 racks/cabinets connected by a high-speed 3-D torus network.

 

A supercomputer is a computer with a high level of performance compared to a general-purpose computer. The performance of a supercomputer is commonly measured in floating-point operations per second (FLOPS) instead of million instructions per second (MIPS). Since 2017, there are supercomputers which can perform up to nearly a hundred quadrillion FLOPS. Since November 2017, all of the world's fastest 500 supercomputers run Linux-based operating systems. Additional research is being conducted in China, the United States, the European Union, Taiwan and Japan to build even faster, more powerful and more technologically superior exascale supercomputers.

Supercomputers play an important role in the field of computational science, and are used for a wide range of computationally intensive tasks in various fields, including quantum mechanics, weather forecasting, climate research, oil and gas exploration, molecular modeling (computing the structures and properties of chemical compounds, biological macromolecules, polymers, and crystals), and physical simulations (such as simulations of the early moments of the universe, airplane and spacecraft aerodynamics, the detonation of nuclear weapons, and nuclear fusion). Throughout their history, they have been essential in the field of cryptanalysis.

Supercomputers were introduced in the 1960s, and for several decades the fastest were made by Seymour Cray at Control Data Corporation (CDC), Cray Research and subsequent companies bearing his name or monogram. The first such machines were highly tuned conventional designs that ran faster than their more general-purpose contemporaries. Through the 1960s, they began to add increasing amounts of parallelism with one to four processors being typical. From the 1970s, vector processors operating on large arrays of data came to dominate. A notable example is the highly successful Cray-1 of 1976. Vector computers remained the dominant design into the 1990s. From then until today, massively parallel supercomputers with tens of thousands of off-the-shelf processors became the norm.

The US has long been the leader in the supercomputer field, first through Cray's almost uninterrupted dominance of the field, and later through a variety of technology companies. Japan made major strides in the field in the 1980s and 90s, but since then China has become increasingly active in the field. As of November 2018, the fastest supercomputer on the TOP500 supercomputer list is the Summit, in the United States, with a LINPACK benchmark score of 143.5 PFLOPS, followed by, Sierra, by around 48.860 PFLOPS. The US has five of the top 10 and China has two. In June 2018, all supercomputers on the list combined have broken the 1 exabyte mark.

History

A circuit board from the IBM 7030

 

The CDC 6600. Behind the system console are two of the "arms" of the plus-sign shaped cabinet with the covers opened. Each arm of the machine had up to four such racks. On the right is the cooling system.

 

A Cray-1 preserved at the Deutsches Museum

 

In 1960 Sperry Rand built the Livermore Atomic Research Computer (LARC), today considered among the first supercomputers, for the US Navy Research and Development Centre. It still used high-speed drum memory, rather than the newly emerging disk drive technology. Also among the first supercomputers was the IBM 7030 Stretch. The IBM 7030 was built by IBM for the Los Alamos National Laboratory, which in 1955 had requested a computer 100 times faster than any existing computer. The IBM 7030 used transistors, magnetic core memory, pipelined instructions, prefetched data through a memory controller and included pioneering random access disk drives. The IBM 7030 was completed in 1961 and despite not meeting the challenge of a hundredfold increase in performance, it was purchased by the Los Alamos National Laboratory. Customers in England and France also bought the computer and it became the basis for the IBM 7950 Harvest, a supercomputer built for cryptanalysis.

The third pioneering supercomputer project in the early 1960s was the Atlas at the University of Manchester, built by a team led by Tom Kilburn. He designed the Atlas to have memory space for up to a million words of 48 bits, but because magnetic storage with such a capacity was unaffordable, the actual core memory of Atlas was only 16,000 words, with a drum providing memory for a further 96,000 words. The Atlas operating system swapped data in the form of pages between the magnetic core and the drum. The Atlas operating system also introduced time-sharing to supercomputing, so that more than one programe could be executed on the supercomputer at any one time. Atlas was a joint venture between Ferranti and the Manchester University and was designed to operate at processing speeds approaching one microsecond per instruction, about one million instructions per second.

The CDC 6600, designed by Seymour Cray, was finished in 1964 and marked the transition from germanium to silicon transistors. Silicon transistors could run faster and the overheating problem was solved by introducing refrigeration to the supercomputer design. Thus the CDC6600 became the fastest computer in the world. Given that the 6600 outperformed all the other contemporary computers by about 10 times, it was dubbed a supercomputer and defined the supercomputing market, when one hundred computers were sold at $8 million each.

Cray left CDC in 1972 to form his own company, Cray Research. Four years after leaving CDC, Cray delivered the 80 MHz Cray-1 in 1976, which became one of the most successful supercomputers in history. The Cray-2 was released in 1985. It had eight central processing units (CPUs), liquid cooling and the electronics coolant liquid fluorinert was pumped through the supercomputer architecture. It performed at 1.9 gigaFLOPS and was the world's second fastest after M-13 supercomputer in Moscow.

Massively parallel designs

The only computer to seriously challenge the Cray-1's performance in the 1970s was the ILLIAC IV. This machine was the first realized example of a true massively parallel computer, in which many processors worked together to solve different parts of a single larger problem. In contrast with the vector systems, which were designed to run a single stream of data as quickly as possible, in this concept, the computer instead feeds separate parts of the data to entirely different processors and then recombines the results. The ILLIAC's design was finalized in 1966 with 256 processors and offer speed up to 1 GFLOPS, compared to the 1970s Cray-1's peak of 250 MFLOPS. However, development problems led to only 64 processors being built, and the system could never operate faster than about 200 MFLOPS while being much larger and more complex than the Cray. Another problem was that writing software for the system was difficult, and getting peak performance from it was a matter of serious effort. 

But the partial success of the ILLIAC IV was widely seen as pointing the way to the future of supercomputing. Cray argued against this, famously quipping that "If you were plowing a field, which would you rather use? Two strong oxen or 1024 chickens?" But by the early 1980s, several teams were working on parallel designs with thousands of processors, notably the Connection Machine (CM) that developed from research at MIT. The CM-1 used as many as 65,536 simplified custom microprocessors connected together in a network to share data. Several updated versions followed; the CM-5 supercomputer is a massively parallel processing computer capable of many billions of arithmetic operations per second.

In 1982, Osaka University's LINKS-1 Computer Graphics System used a massively parallel processing architecture, with 514 microprocessors, including 257 Zilog Z8001 control processors and 257 iAPX 86/20 floating-point processors. It was mainly used for rendering realistic 3D computer graphics. Fujitsu's Numerical Wind Tunnel supercomputer used 166 vector processors to gain the top spot in 1994 with a peak speed of 1.7 gigaFLOPS (GFLOPS) per processor. The Hitachi SR2201 obtained a peak performance of 600 GFLOPS in 1996 by using 2048 processors connected via a fast three-dimensional crossbar network. The Intel Paragon could have 1000 to 4000 Intel i860 processors in various configurations and was ranked the fastest in the world in 1993. The Paragon was a MIMD machine which connected processors via a high speed two dimensional mesh, allowing processes to execute on separate nodes, communicating via the Message Passing Interface.

Software development remained a problem, but the CM series sparked off considerable research into this issue. Similar designs using custom hardware were made by many companies, including the Evans & Sutherland ES-1, MasPar, nCUBE, Intel iPSC and the Goodyear MPP. But by the mid-1990s, general-purpose CPU performance had improved so much in that a supercomputer could be built using them as the individual processing units, instead of using custom chips. By the turn of the 21st century, designs featuring tens of thousands of commodity CPUs were the norm, with later machines adding graphic units to the mix.

The CPU share of TOP500

 

Systems with a massive number of processors generally take one of two paths. In the grid computing approach, the processing power of many computers, organised as distributed, diverse administrative domains, is opportunistically used whenever a computer is available. In another approach, a large number of processors are used in proximity to each other, e.g. in a computer cluster. In such a centralized massively parallel system the speed and flexibility of the interconnect becomes very important and modern supercomputers have used various approaches ranging from enhanced Infiniband systems to three-dimensional torus interconnects. The use of multi-core processors combined with centralization is an emerging direction, e.g. as in the Cyclops64 system.

As the price, performance and energy efficiency of general purpose graphic processors (GPGPUs) have improved, a number of petaFLOPS supercomputers such as Tianhe-I and Nebulae have started to rely on them. However, other systems such as the K computer continue to use conventional processors such as SPARC-based designs and the overall applicability of GPGPUs in general-purpose high-performance computing applications has been the subject of debate, in that while a GPGPU may be tuned to score well on specific benchmarks, its overall applicability to everyday algorithms may be limited unless significant effort is spent to tune the application towards it. However, GPUs are gaining ground and in 2012 the Jaguar supercomputer was transformed into Titan by retrofitting CPUs with GPUs.

High-performance computers have an expected life cycle of about three years before requiring an upgrade.

Special purpose supercomputers

A number of "special-purpose" systems have been designed, dedicated to a single problem. This allows the use of specially programmed FPGA chips or even custom ASICs, allowing better price/performance ratios by sacrificing generality. Examples of special-purpose supercomputers include Belle, Deep Blue, and Hydra, for playing chess, Gravity Pipe for astrophysics, MDGRAPE-3 for protein structure computation molecular dynamics and Deep Crack, for breaking the DES cipher.

Energy usage and heat management

A Blue Gene/L cabinet showing the stacked blades, each holding many processors

 

Throughout the decades, the management of heat density has remained a key issue for most centralized supercomputers. The large amount of heat generated by a system may also have other effects, e.g. reducing the lifetime of other system components. There have been diverse approaches to heat management, from pumping Fluorinert through the system, to a hybrid liquid-air cooling system or air cooling with normal air conditioning temperatures. A typical supercomputer consumes large amounts of electrical power, almost all of which is converted into heat, requiring cooling. For example, Tianhe-1A consumes 4.04 megawatts (MW) of electricity. The cost to power and cool the system can be significant, e.g. 4 MW at $0.10/kWh is $400 an hour or about $3.5 million per year. 

An IBM HS20 blade

 

Heat management is a major issue in complex electronic devices and affects powerful computer systems in various ways. The thermal design power and CPU power dissipation issues in supercomputing surpass those of traditional computer cooling technologies. The supercomputing awards for green computing reflect this issue.

The packing of thousands of processors together inevitably generates significant amounts of heat density that need to be dealt with. The Cray 2 was liquid cooled, and used a Fluorinert "cooling waterfall" which was forced through the modules under pressure. However, the submerged liquid cooling approach was not practical for the multi-cabinet systems based on off-the-shelf processors, and in System X a special cooling system that combined air conditioning with liquid cooling was developed in conjunction with the Liebert company.

In the Blue Gene system, IBM deliberately used low power processors to deal with heat density. The IBM Power 775, released in 2011, has closely packed elements that require water cooling. The IBM Aquasar system uses hot water cooling to achieve energy efficiency, the water being used to heat buildings as well.

The energy efficiency of computer systems is generally measured in terms of "FLOPS per watt". In 2008, IBM's Roadrunner operated at 3.76 MFLOPS/W. In November 2010, the Blue Gene/Q reached 1,684 MFLOPS/W. In June 2011 the top 2 spots on the Green 500 list were occupied by Blue Gene machines in New York (one achieving 2097 MFLOPS/W) with the DEGIMA cluster in Nagasaki placing third with 1375 MFLOPS/W.

Because copper wires can transfer energy into a supercomputer with much higher power densities than forced air or circulating refrigerants can remove waste heat, the ability of the cooling systems to remove waste heat is a limiting factor. As of 2015, many existing supercomputers have more infrastructure capacity than the actual peak demand of the machine – designers generally conservatively design the power and cooling infrastructure to handle more than the theoretical peak electrical power consumed by the supercomputer. Designs for future supercomputers are power-limited – the thermal design power of the supercomputer as a whole, the amount that the power and cooling infrastructure can handle, is somewhat more than the expected normal power consumption, but less than the theoretical peak power consumption of the electronic hardware.

Software and system management

Operating systems

Since the end of the 20th century, supercomputer operating systems have undergone major transformations, based on the changes in supercomputer architecture. While early operating systems were custom tailored to each supercomputer to gain speed, the trend has been to move away from in-house operating systems to the adaptation of generic software such as Linux.

Since modern massively parallel supercomputers typically separate computations from other services by using multiple types of nodes, they usually run different operating systems on different nodes, e.g. using a small and efficient lightweight kernel such as CNK or CNL on compute nodes, but a larger system such as a Linux-derivative on server and I/O nodes.

While in a traditional multi-user computer system job scheduling is, in effect, a tasking problem for processing and peripheral resources, in a massively parallel system, the job management system needs to manage the allocation of both computational and communication resources, as well as gracefully deal with inevitable hardware failures when tens of thousands of processors are present.

Although most modern supercomputers use a Linux-based operating system, each manufacturer has its own specific Linux-derivative, and no industry standard exists, partly due to the fact that the differences in hardware architectures require changes to optimize the operating system to each hardware design.

Software tools and message passing

Wide-angle view of the ALMA correlator

 

The parallel architectures of supercomputers often dictate the use of special programming techniques to exploit their speed. Software tools for distributed processing include standard APIs such as MPI and PVM, VTL, and open source-based software solutions such as Beowulf. 

In the most common scenario, environments such as PVM and MPI for loosely connected clusters and OpenMP for tightly coordinated shared memory machines are used. Significant effort is required to optimize an algorithm for the interconnect characteristics of the machine it will be run on; the aim is to prevent any of the CPUs from wasting time waiting on data from other nodes. GPGPUs have hundreds of processor cores and are programmed using programming models such as CUDA or OpenCL. 

Moreover, it is quite difficult to debug and test parallel programs. Special techniques need to be used for testing and debugging such applications.

Distributed supercomputing

Opportunistic approaches

Example architecture of a grid computing system connecting many personal computers over the internet

 

Opportunistic Supercomputing is a form of networked grid computing whereby a "super virtual computer" of many loosely coupled volunteer computing machines performs very large computing tasks. Grid computing has been applied to a number of large-scale embarrassingly parallel problems that require supercomputing performance scales. However, basic grid and cloud computing approaches that rely on volunteer computing cannot handle traditional supercomputing tasks such as fluid dynamic simulations. 

The fastest grid computing system is the distributed computing project Folding@home (F@h). F@h reported 101 PFLOPS of x86 processing power As of October 2016. Of this, over 100 PFLOPS are contributed by clients running on various GPUs, and the rest from various CPU systems.

The Berkeley Open Infrastructure for Network Computing (BOINC) platform hosts a number of distributed computing projects. As of February 2017, BOINC recorded a processing power of over 166 PetaFLOPS through over 762 thousand active Computers (Hosts) on the network.

As of October 2016, Great Internet Mersenne Prime Search's (GIMPS) distributed Mersenne Prime search achieved about 0.313 PFLOPS through over 1.3 million computers. The Internet PrimeNet Server supports GIMPS's grid computing approach, one of the earliest and most successful grid computing projects, since 1997.

Quasi-opportunistic approaches

Quasi-opportunistic supercomputing is a form of distributed computing whereby the "super virtual computer" of many networked geographically disperse computers performs computing tasks that demand huge processing power. Quasi-opportunistic supercomputing aims to provide a higher quality of service than opportunistic grid computing by achieving more control over the assignment of tasks to distributed resources and the use of intelligence about the availability and reliability of individual systems within the supercomputing network. However, quasi-opportunistic distributed execution of demanding parallel computing software in grids should be achieved through implementation of grid-wise allocation agreements, co-allocation subsystems, communication topology-aware allocation mechanisms, fault tolerant message passing libraries and data pre-conditioning.

HPC in the Cloud

Cloud Computing with its recent and rapid expansions and development have grabbed the attention of HPC users and developers in recent years. Cloud Computing attempts to provide HPC-as-a-Service exactly like other forms of services currently available in the Cloud such as Software-as-a-Service, Platform-as-a-Service, and Infrastructure-as-a-Service. HPC users may benefit from the Cloud in different angles such as scalability, resources being on-demand, fast, and inexpensive. On the other hand, moving HPC applications have a set of challenges too. Good examples of such challenges are virtualization overhead in the Cloud, multi-tenancy of resources, and network latency issues. Much research is currently being done to overcome these challenges and make HPC in the cloud a more realistic possibility.

Performance measurement

Capability versus capacity

Supercomputers generally aim for the maximum in capability computing rather than capacity computing. Capability computing is typically thought of as using the maximum computing power to solve a single large problem in the shortest amount of time. Often a capability system is able to solve a problem of a size or complexity that no other computer can, e.g., a very complex weather simulation application.

Capacity computing, in contrast, is typically thought of as using efficient cost-effective computing power to solve a few somewhat large problems or many small problems. Architectures that lend themselves to supporting many users for routine everyday tasks may have a lot of capacity but are not typically considered supercomputers, given that they do not solve a single very complex problem.

Performance metrics

Top supercomputer speeds: logscale speed over 60 years

 

In general, the speed of supercomputers is measured and benchmarked in "FLOPS" (FLoating point Operations Per Second), and not in terms of "MIPS" (Million Instructions Per Second), as is the case with general-purpose computers. These measurements are commonly used with an SI prefix such as tera-, combined into the shorthand "TFLOPS" (10¹² FLOPS, pronounced teraflops), or peta-, combined into the shorthand "PFLOPS" (10¹⁵ FLOPS, pronounced petaflops.) "Petascale" supercomputers can process one quadrillion (10¹⁵) (1000 trillion) FLOPS. Exascale is computing performance in the exaFLOPS (EFLOPS) range. An EFLOPS is one quintillion (10¹⁸) FLOPS (one million TFLOPS). 

No single number can reflect the overall performance of a computer system, yet the goal of the Linpack benchmark is to approximate how fast the computer solves numerical problems and it is widely used in the industry. The FLOPS measurement is either quoted based on the theoretical floating point performance of a processor (derived from manufacturer's processor specifications and shown as "Rpeak" in the TOP500 lists), which is generally unachievable when running real workloads, or the achievable throughput, derived from the LINPACK benchmarks and shown as "Rmax" in the TOP500 list. The LINPACK benchmark typically performs LU decomposition of a large matrix. The LINPACK performance gives some indication of performance for some real-world problems, but does not necessarily match the processing requirements of many other supercomputer workloads, which for example may require more memory bandwidth, or may require better integer computing performance, or may need a high performance I/O system to achieve high levels of performance.

The TOP500 list

Distribution of TOP500 supercomputers among different countries, as of November 2015

 

Since 1993, the fastest supercomputers have been ranked on the TOP500 list according to their LINPACK benchmark results. The list does not claim to be unbiased or definitive, but it is a widely cited current definition of the "fastest" supercomputer available at any given time. 

This is a recent list of the computers which appeared at the top of the TOP500 list, and the "Peak speed" is given as the "Rmax" rating. 

Top 20 Supercomputers in the World, as of June 2014

 

Year	Supercomputer	Peak speed (Rmax)	Location
2018	IBM Summit	122.3 PFLOPS	Oak Ridge, U.S.
2016	Sunway TaihuLight	93.01 PFLOPS	Wuxi, China
2013	NUDT Tianhe-2	33.86 PFLOPS	Guangzhou, China
2012	Cray Titan	17.59 PFLOPS	Oak Ridge, U.S.
2012	IBM Sequoia	17.17 PFLOPS	Livermore, U.S.
2011	Fujitsu K computer	10.51 PFLOPS	Kobe, Japan
2010	Tianhe-IA	2.566 PFLOPS	Tianjin, China
2009	Cray Jaguar	1.759 PFLOPS	Oak Ridge, U.S.
2008	IBM Roadrunner	1.026 PFLOPS	Los Alamos, U.S.
		1.105 PFLOPS

Applications

The stages of supercomputer application may be summarized in the following table: 

Decade	Uses and computer involved
1970s	Weather forecasting, aerodynamic research (Cray-1).
1980s	Probabilistic analysis, radiation shielding modeling (CDC Cyber).
1990s	Brute force code breaking (EFF DES cracker).
2000s	3D nuclear test simulations as a substitute for legal conduct Nuclear Non-Proliferation Treaty (ASCI Q).
2010s	Molecular Dynamics Simulation (Tianhe-1A)

The IBM Blue Gene/P computer has been used to simulate a number of artificial neurons equivalent to approximately one percent of a human cerebral cortex, containing 1.6 billion neurons with approximately 9 trillion connections. The same research group also succeeded in using a supercomputer to simulate a number of artificial neurons equivalent to the entirety of a rat's brain.

Modern-day weather forecasting also relies on supercomputers. The National Oceanic and Atmospheric Administration uses supercomputers to crunch hundreds of millions of observations to help make weather forecasts more accurate.

In 2011, the challenges and difficulties in pushing the envelope in supercomputing were underscored by IBM's abandonment of the Blue Waters petascale project.

The Advanced Simulation and Computing Program currently uses supercomputers to maintain and simulate the United States nuclear stockpile.

Development and trends

Diagram of a three-dimensional torus interconnect used by systems such as Blue Gene, Cray XT3, etc.

 

In the 2010s, China, the United States, the European Union, and others competed to be the first to create a 1 exaFLOP (10¹⁸ or one quintillion FLOPS) supercomputer. Erik P. DeBenedictis of Sandia National Laboratories has theorized that a zettaFLOPS (10²¹ or one sextillion FLOPS) computer is required to accomplish full weather modeling, which could cover a two-week time span accurately. Such systems might be built around 2030.

Many Monte Carlo simulations use the same algorithm to process a randomly generated data set; particularly, integro-differential equations describing physical transport processes, the random paths, collisions, and energy and momentum depositions of neutrons, photons, ions, electrons, etc. The next step for microprocessors may be into the third dimension; and specializing to Monte Carlo, the many layers could be identical, simplifying the design and manufacture process.

The cost of operating high performance supercomputers has risen, mainly due to increasing power consumption. In the mid 1990s a top 10 supercomputer required in the range of 100 kilowatt, in 2010 the top 10 supercomputers required between 1 and 2 megawatt. A 2010 study commissioned by DARPA identified power consumption as the most pervasive challenge in achieving Exascale computing. At the time a megawatt per year in energy consumption cost about 1 million dollar. Supercomputing facilities were constructed to efficiently remove the increasing amount of heat produced by modern multi-core central processing units. Based on the energy consumption of the Green 500 list of supercomputers between 2007 and 2011, a supercomputer with 1 exaflops in 2011 would have required nearly 500 megawatt. Operating systems were developed for existing hardware to conserve energy whenever possible. CPU cores not in use during the execution of a parallelised application were put into low-power states, producing energy savings for some supercomputing applications.

The increasing cost of operating supercomputers has been a driving factor in a trend towards bundling of resources through a distributed supercomputer infrastructure. National supercomputing centres first emerged in the US, followed by Germany and Japan. The European Union launched the Partnership for Advanced Computing in Europe (PRACE) with the aim of creating a persistent pan-European supercomputer infrastructure with services to support scientists across the European Union in porting, scaling and optimizing supercomputing applications. Iceland built the world's first zero-emission supercomputer. Located at the Thor Data Center in Reykjavik, Iceland, this supercomputer relies on completely renewable sources for its power rather than fossil fuels. The colder climate also reduces the need for active cooling, making it one of the greenest facilities in the world of computers.

Funding supercomputer hardware also became increasingly difficult. In the mid 1990s a top 10 supercomputer cost about 10 Million Euros, while in 2010 the top 10 supercomputers required an investment of between 40 and 50 million Euros. In the 2000s national governments put in place different strategies to fund supercomputers. In the UK the national government funded supercomputers entirely and high performance computing was put under the control of a national funding agency. Germany developed a mixed funding model, pooling local state funding and federal funding.

In fiction

Many science-fiction writers have depicted supercomputers in their works, both before and after the historical construction of such computers. Much of such fiction deals with the relations of humans with the computers they build and with the possibility of conflict eventually developing between them. Some scenarios of this nature appear on the AI-takeover page. 

Examples of supercomputers in fiction include HAL-9000, Multivac, The Machine Stops, GLaDOS, The Evitable Conflict and Vulcan's Hammer.

Trained neural nets perform much like humans on classic psychological tests

Neural networks were inspired by the human brain. Now AI researchers have shown that they perceive the world in similar ways.

• by Emerging Technology from the arXiv
• March 15, 2019
• Original link:  https://www.technologyreview.com/s/613114/trained-neural-nets-perform-much-like-humans-on-classic-psychological-tests/?utm_medium=tr_social&utm_source=facebook&utm_campaign=site_visitor.unpaid.engagement&fbclid=IwAR32B34iO2qx1yysNyvGvzCH-umbqTJ41LOqayU-GySShrl7EcjnmVUsBf0

In the early part of the 20th century, a group of German experimental psychologists began to question how the brain acquires meaningful perceptions of a world that is otherwise chaotic and unpredictable. To answer this question, they developed the notion of the “gestalt effect”—the idea that when it comes to perception, the whole is something other than the parts.

Sine then, psychologists have discovered that the human brain is remarkably good at perceiving complete pictures on the basis of fragmentary information. A good example is the figure shown here. The brain perceives two-dimensional shapes such as a triangle and a square, and even a three-dimensional sphere. But none of these shapes is explicitly drawn. Instead, the brain fills in the gaps.

A natural extension to this work is to ask whether gestalt effects occur in neural networks. These networks are inspired by the human brain. Indeed, researchers studying machine vision say the deep neural networks they have developed turn out to be remarkably similar to the visual system in primate brains and to parts of the human cortex.


That leads to an interesting question: can neural networks perceive a whole object by looking merely at its parts, as humans do?

Today we get an answer thanks to the work of Been Kim and colleagues at Google Brain, the company’s AI research division in Mountain View, California. The researchers have tested various neural networks using the same gestalt experiments designed for humans. And they say they have good evidence that machines can indeed perceive whole objects using observations of the parts.

 

Kim and co’s experiment is based on the triangle illusion shown in the figure. They first create three databases of images for training their neural networks. The first consists of ordinary complete triangles displayed in their entirety.

The next database shows only the corners of the triangles, with lines that must be interpolated to perceive the complete shape. This is the illusory data set. When humans view these types of images, they tend to close the gaps and end up perceiving the triangle as a whole. “We aim to determine whether neural networks exhibit similar closure effects,” say Kim and co.

The final database consists of similar “corners” but randomly oriented so that the lines cannot be interpolated to form triangles. This is the non-illusory data set.

By varying the size and orientation of these shapes, the team created almost 1,000 different images to train their machines.

Their approach is to train a neural network to recognize ordinary complete triangles and then to test whether it classifies the images in the illusory data set as complete triangles (while ignoring the images in the non-illusory data set). In other words, they test whether the machine can fill in the gaps in the images to form a complete picture.

They also compare the behavior of a trained network with the behavior of an untrained network or one trained on random data.

The results make for interesting reading. It turns out that the behavior of trained neural networks shows remarkable similarities to human gestalt effects. “Our findings suggest that neural networks trained with natural images do exhibit closure, in contrast to networks with randomized weights or networks that have been trained on visually random data,” say Kim and co.

That’s a fascinating result. And not just because it shows how neural networks mimic the brain to make sense of the world.

The bigger picture is that the team’s approach opens the door to an entirely new way of studying neural networks using the tools of experimental psychology. “We believe that exploring other Gestalt laws—and more generally, other psychophysical phenomena—in the context of neural networks is a promising area for future research,” say Kim and co.

That looks like a first step into a new field of machine psychology. As the Google team put it: “Understanding where humans and neural networks differ will be helpful for research on interpretability by enlightening the fundamental differences between the two interesting species.” The German experimental psychologists of the early 20th century would surely have been fascinated.

Ref: arxiv.org/abs/1903.01069 : Do Neural Networks Show Gestalt Phenomena? An Exploration of the Law of Closure