BODY OF SECRETS -- ANATOMY OF THE ULTRA-SECRET NATIONAL SECURITY AGENCY

CHAPTER FOURTEEN: BRAIN

WDLDXTDKS'B AFSWDX GSADB GSKKYTQ YG CDKSZDC WYQD RJEPBZYPZA QWXPK QWZLX OXLZ QJB KOXWAAWZR YWNBJKJQA IBRUITRUL TEF HTHWEF BRTINRXK NTHXKF RU MRLM BRUIF OHSQSHYJB LGADM-DYJBSL ZDPSW MV DYQS DGK ZPLASLW UABCHPC QTMQ EJHBC QJDDBH UPJAW MBVAW EGPXGVQQ

It would be one of the most delicate operations ever performed. The doctors and technicians would gather early and work late into the night. Any mistake could be extremely serious. The patient's memory could be lost forever, or the ability to function severely damaged. Crypto City was about to undergo its first brain transplant. According to the director, nothing less than "the continued success of the agency's Sigint mission largely depended on this." The' planning had taken years. NSA would create the largest, most powerful, and most secret electronic brain on earth.

But first it would have to build a specialized facility to house the new center. Then it would need to carefully transplant tons of massive and delicate supercomputers -- more than 150 -- from the cavernous basement of OPS 1 to their new home, out of sight in a wooded corner of the secret city nearly a mile away. Whereas most government offices and large corporations measure in square feet the space taken up by their computers, NSA measures it in acres. "I had five and a half acres of computers when I was there," said Marshall Carter, director in the late 1960s. "We didn't Count them by numbers; it was five and a half acres." Even though modern computers have more capacity and smaller footprints, one NSA employee more than a decade later commented, "It's double that today."

Once in place, the computers would be brought back to life and linked by a secure fiber optic spinal cord to the Headquarters/Operations Building complex-all without disrupting NSA's critical operations. When it was finally completed, in 1996, NSA's Supercomputer Facility held the most powerful collection of thinking machines on the planet.

Standing in front of the new building on the afternoon of October 29, 1996, Kenneth Minihan held a pair of scissors up to a thin ribbon of red, white, and blue. No press releases had been issued, and even the invitations to the event gave no hint where the ceremony would actually take place. But then, that was precisely how the man in whose name the Tordella Supercomputer Facility was about to be dedicated would have wanted it. This would be the first NSA building to be named for a person. As the scissors sliced through the colorful ribbon, a handheld machine of elegant simplicity opened the way to a building of infinite complexity.

***

The history of modern codebreaking and the history of computers are, to a large degree, coterminous. Yet because of its "policy of anonymity," NSA's role has been almost totally hidden. When the Association for Computing Machinery sponsored a commemoration of the twenty-fifth anniversary of its founding, NSA simply stayed away. Likewise, when computing pioneers gathered at the quarter-century anniversary meeting of the Institute of Electrical and Electronic Engineers' Computer Society, NSA again exhibited an advanced case of shyness.

But NSA's role in computer development has been, and continues to be, enormous. The man responsible for much of that work -- as well as for the thick shroud of secrecy that still surrounds it -- was Dr. Louis W. Tordella, NSA's keeper of the secrets.

By the outbreak of the Second World War, the importance of machines to aid in codebreaking was known but their use was limited. At that time the Signal Security Agency had only fifteen machines and twenty-one operators. But by the spring of 1945, the SSA was employing 1,275 operators and supervisors to work on 407 keypunch machines.

Besides its off-the-shelf tabulating machines, the agency had specialized machines custom built for codebreaking. Known as Rapid Analytical Machines (RAMs), they employed vacuum tubes, relays, high-speed electronic circuits, and photoelectrical principles. They were the forerunners of the modern computer, but they were expensive and overspecialized. A number of them were built to attack a specific code or cipher, so if a cipher system was changed or abandoned, the machine was of little value.

The Navy's Op-20-G contracted with Eastman Kodak, National Cash Register, and several other firms to design and build its RAMs. The Army's Signal Security Agency, on the other hand, worked closely with Bell Laboratories. Another major contractor during the war was IBM, which built a specialized attachment for its IBM tabulator, thereby increasing the power of the standard punch-card systems by several orders of magnitude.

Two of the SSA's cryptanalytic machines were immense. Costing a million dollars apiece, an extraordinary sum at the time, they were capable of performing operations which, if done by hand in the old Black Chamber, would have required over 200,000 people. By the end of 1945 another monster machine was nearing completion; it had power equivalent to 5 million cryptanalysts.

Tordella hoped the development by outside contractors of new, sophisticated cryptologic equipment would continue. But with no war to fight he found the contractors less willing to undertake the research. The rigorous security clearances, the oppressive physical security, and the limited usefulness of the equipment in the marketplace made many companies shy away from the field. Because of this, a group of former Navy officers, familiar with cryptography and signals intelligence, banded together to form Engineering Research Associates, which took on some of the Naval Security Group's most complex assignments.

At about the same time, a group of engineers and mathematicians at the University of Pennsylvania's Moore School of Electrical Engineering completed an electronic marvel named ENIAC (for "electronic numerical integrator computer"), and thus gave birth to the computer era. ENIAC was an ungainly giant whose body was a good deal larger han its brain. Its total storage capacity was only twenty numbers, yet its 18,000 electron tubes took up the better part of a room thirty feet by fifty. Nevertheless, the machine offered tremendous possibilities in speed.

The development of ENIAC led to a series of lectures on the theoryof computers, presented at the Moore School and sponsored jointly by the Office of Naval Research and the Army's Ordnance Department. Among those attending the lectures, given between July 8 and August 31, 1946, was Lieutenant Commander James T. Pendergrass, a colleague of Tordella's in the Naval Security Group, whose assignment was to assess the potential of computers in cryptography and signals intelligence.

Pendergrass came away from the lectures excited. Computers appeared to offer the flexibility that RAMs lacked. Whereas many of the RAMs were designed to handle one particular problem, such as breaking one foreign cipher system, computers could handle a whole range of problems. "The author believes that the general purpose mathematical computer, now in the design stage, is a general purpose cryptanalytic machine," wrote Pendergrass. "A computer could do everything that any analytic machine in Building 4 can do, and do a good percentage of these problems more rapidly."

Soon after Pendergrass submitted his favorable report, negotiations began between the Security Group and Engineering Research Associates for the design and construction of the signals intelligence community's first computer. But what to name it? A yeoman overheard Tordella and his colleagues discussing ideas and suggested "Atlas," after the mental giant in the comic strip "Barnaby." Atlas lived up to its namesake. "When it was delivered to the Security Group in December 1950, Atlas had an impressive capacity of 16,384 words; it was the first parallel electronic computer in the United States with a drum memory. A second, identical computer was delivered to NSA in March 1953.

A key component of the machine was the vacuum tube. "We had the biggest collection of vacuum tube circuitry anyplace in the world there at one time," said former NSA research chief Howard Campaigne. "And we knew more about the life of vacuum tubes and the kinds of vacuum tubes that were used and how they should be maintained than just about anybody else." The vacuum tubes, he said, were as big as lightbulbs. "And then you get a lot of lightbulbs together and you have to have air-conditioning to cool them off. And so we were having fifteen tons of air-conditioning per machine."

Tordella was not the only one impressed by the Pendergrass report. About the same time that he received it, a copy also landed on the desk, of Sam Snyder at Arlington Hall, headquarters of the Army Security Agency. "A copy of this report hit my desk in November 1946," Snyder later recalled, "and my reaction was explosive. I immediately ran into the office of Dr. Solomon Kullback, my boss, and said something like, 'We have to get a machine like this. Think what it could do for us!' " Kullback assigned Snyder to investigate the possibilities; Snyder spent the next year meeting with experts such as John von Neumann, at Princeton University's Institute for Advanced Study, and visiting institutions and private companies involved in computer research. "In the agency at that time," Snyder said, "money was no object; we could get whatever we wanted."

Eventually the ASA built its own codebreaking computer, which they named Abner. "We chose the name from Li'l Abner Yokum, the comic strip character who was a big brute, but not very smart," said Snyder, a longtime NSA computer expert, "because we believed that computers, which can be big and do brute-force operations, aren't very bright either; they can only follow simple instructions but can't think for themselves." Abner was originally given only fifteen simple programs, or "instructions" (later doubled to thirty). Nevertheless, when it was secretly completed in April 1952 it was the "most sophisticated computer of its time. One could enter or extract information not only with the standard keypunched computer card but also with punched paper tape, magnetic tape, a parallel printer, a typewriter, or a console.

At NSA Tordella became chief of NSA-70, which was responsible for high-level cryptanalysis. He and the others who were pushing for ever-increasing computer power got a boost in 1954. James Killian, a Harvard professor exploring US. vulnerability to another surprise attack, concluded that 90 percent of war warnings would inevitably come from signals intelligence. But, he pointed out, since nuclear attack could come in a matter of minutes, it would be necessary to speed up the timeline on eavesdropping and codebreaking to beat the clock. "From then on," said one former NSAer, "the focus of the Sigint process was on speed."

Several years later, in July 1956, one of the most costly as well as far-reaching research programs ever undertaken by NSA was born. Its birthplace, however, was not a chalk-covered blackboard in Research and Engineering but a cocktail party. Over drinks, several high-level NSA equipment planners began discussing with Director Canine a number of the agency's perennial problems. At the top of the list was the battle between the codebreakers, always looking to attack ever-increasing volumes of data, and the engineers, constantly attempting to design and build bigger and faster computers to meet those needs. No matter how powerful the new equipment, the engineers never seemed to catch up. Tordella began pushing for research into second-generation computer technology.

At the time, NSA was using the PACE 10, the first analog desktop computer used at the agency. It was self-contained, to the extent that the logic was in the interior. The output was a printing device. The plug-in units had a wire associated with them and each panel was set up to do a different mathematical function. For a· fairly complex mathematical problem, one would plug in all the appropriate panels and hand-wire them together. The computer's operations manual boasted that once it was set up, a problem could be completed in fifteen to sixty seconds.

On the drawing board was a second-generation computer known as Harvest. It was designed to be an estimated hundredfold improvement in speed over the best current computers, but a completion date was still several years away. Exasperated by this situation, Canine exploded: "Dammit, I want you fellows to get the jump on those guys [computer companies]! Build me a thousand-megacycle machine! I'll get the money!"

The head of NSA's REMP (Research, Engineering, Math and Physics) Branch at the time was Howard Campaigne, who had helped uncover the high-level Russian "Fish" cipher system as part of TICOM. ''After the ideas of Harvest were started," he said, "we in research tried to think of other things; and one of the suggestions that came up was that we ought to have a big program. We ought to attack it like the Manhattan Project. We ought to really go after it. And so we dreamed up this 'Project Lightning.'"

***

It was a time, according to Campaigne, when anything was possible. "We were always surprised. We had an idea which looked expensive and we'd go ahead and they'd always be encouraging -- 'Do it,' " he said. "During most of my career, we always had encouragement from above to do things. If you can see something to do, do it. We made some mistakes, but by and large, most of the things we attacked were at least partially successful." Among the successes was developing the first solid-state computer by replacing vacuum tubes with transistors. Then the transistors were replaced by magnetic cores in a computer named Bogart.

But by the late 19605, said Campaigne, things began to change. "In the late sixties we weren't getting encouragement. We were being told the budget had to be cut. We had to do without ... I used to argue that it [the research-and-development percentage of the overall NSA budget] should be more than five percent. It ought to be up in the seven and eight percent [range] ... During the Lightning program, my budget had been as high as nine million dollars a year. And when I left in '69, that was my last full fiscal year, our budget was three million. It had been cut to a third ... And we had been pretty much cut down in contract work. All the contracts were much smaller than they had been. So when I became eligible to retire, I figured, Well, gee, no point in staying around here to cut budgets. So I went out." By the late 1990s, the research-and-development portion of the overall NSA budget had dropped even further than during Campaigne's time, to less than 4 percent.

Part of NSA's early success, said Campaigne, was a willingness to take chances. "What the research-and-development people are doing is just trying things out," he said. "They're doing experiments. And so you'd expect them to have a lot of failures and a few successes. Historically, as a matter of fact, they had many more successes than they should have." Later on, as NSA grew, the experiments became less bold. "The reason is they're so damned cautious. See, they're more cautious that we were. At least, more cautious than we should have been.... I guess it's because the researchers like to look good. They don't like-to have a failure, even though they're there just to experiment. They like to succeed. But, in fact, somebody who was administering a research-and-development activity ought to say, 'You know, you guys are too damn cautious. Get out there and do some experimenting.'"

Campaigne's optimistic push-ahead-at-all-costs philosophy derived from his belief that every cipher machine, no matter how difficult, could eventually be broken. "There is no such thing as an unbreakable cipher/' he said, "and it irritates me when people talk about such things without realizing it's nonsense.... But people keep thinking there might be such a thing as an unbreakable cipher."

Secrecy was always NSA's best ally when attempting to get money from Congress. "All those committee chairs were very friendly in those days, and secrecy impressed them," said Arthur Levenson, in charge of Russian codebreaking at NSA and also a veteran of TICOM. "We got most of what we wanted, and a free hand in how we used it." Another former official said, of congressional oversight: "We didn't have any in those days." When General Canine was asked a question during a closed budget appropriations committee hearing, his favorite answer was, "Congressman, you don't really want to know the answer to that. You wouldn't be able to sleep at night." Said one former official, "And the members would look at each other and they were content with that."

***

Awarded $25 million by Congress, and okayed by Eisenhower, NSA's five-year race to develop "thousand-megacycle electronics" was on.

Lightning research began in June 1957. Contractors on the project, the largest government-supported computer research program in history up until then, included Sperry Rand, RCA, IBM, Philco, General Electric, MIT, the University of Kansas, and Ohio State. Though the project's primary goal was to increase circuitry capability by 1,000 percent, the end results went even further, extending the state of the art of computer science well beyond expectations. Research was conducted on cryogenic components, subminiaturization of components, and superfast switching devices, called tunnel diodes.

One of the most rewarding by-products of Lightning was the boost it gave the development of NSA's mammoth Harvest complex, which was designed to be NSA's largest general-purpose computer. For years computers were designed to attack specific codebreaking machines, such as the complex, Swiss-made Hagelin, which was used by many countries around the world. "We had in the past, before that time, we had built a special device for every problem," said Howard Campaigne. "And we'd gotten some very effective devices. But it always took a long time to build it. We had to formulate the problem and design the equipment, and get it constructed, and debugged, and all that had to take place when we ought to be operating."

But a superpowerful computer like Harvest, it was hoped, would be able to attack not only the Hagelin machine but also a variety of cipher machines and systems from multiple countries. "As the computers became more sophisticated," said Solomon Kullback, one of William F. Friedman's original pioneers, it became possible to "program one of, these high- speed general purpose computers so that it could simulate the action of the Hagelin and use them for the Hagelin problem." However, the computer would not be limited to the Hagelin machine.

The original name for the computer was to be Plantation, but it was discovered that the White House had already taken the name to use as a codeword for emergency relocation. "The idea was to have a modular computer set up in which you'd have things which resembled barns and stables and that the plantation [would be] a center or central thing," recalled Howard Campaigne. "So they called it Harvest as part of this plantation group of things."

Ironically, Solomon Kullback, who headed NSA's research-and-development office for a "decade, never had any real enthusiasm for computers until they started proving their worth. "He didn't interfere with us," said Campaigne. "He didn't try to stop us or anything like that, but he just had no personal enthusiasm for it at all. And later on he was willing to spend plenty of money on them. And there were a lot of people like that."

In 1955 IBM began planning its most ambitious computer, the Stretch. So huge was Stretch that IBM designers believed the market contained only two possible customers; NSA and the Atomic Energy Commission. The AEC signed up for the computer primarily because of its advantages in high-speed multiplication. But NSA, looking for more flexibility as well as the manipulation of great volumes of data, sent the engineers back to the drawing board for a more customized version. In April 1958 a final design was approved, and in February 1962 the agency took delivery of its long-awaited Stretch, now modified and considerably faster. "IBM regarded it as a bad experience because the Stretch as a whole they lost money on," said Howard Campaigne. "And since then, they've been very careful about getting into big computers. They just let Seymour Cray build them."

Once in place as the heart -- or, more appropriately, brain -- of NSA's enormous Harvest complex, even Stretch began to look somewhat diminutive. Attached was a variety of unusual, complex accessories that more than doubled the computer's original size. One was the Stream Processing Unit, which was able to take over a number of the more tedious and time-consuming cryptanalytic functions. A key to codebreaking is the ability to quickly test encrypted text against every conceivable combination of letters in an alphabet. Because it may take millions of tries before the right combination of letters is found which breaks the cipher, speed is essential. "It was clear to us that one way of getting high capacity was to go fast," said Campaigne. An evaluation conducted by an NSA team concluded that Harvest was more powerful than the best commercially available computer by a factor of 50 to 200, depending on the particular application.

During World War II, the U.S. Navy's codebreaking machine, known as the bombe, was able to perform tests on 1,300 characters per second. In other words, it was able to try 1,300 separate keys in the German lock every second, looking for the right one to pop it open. With the new Stream Processing Unit, that speed was increased to some 3 million characters each second-a 230;000 percent increase. Thus, to pick the lock, NSA could now try 3 million new keys every second until the right one was found-truly lightning speed.

From one foreign cipher system alone, Harvest was able to process 7,075,315 intercepted messages of about 500 characters each, examining every possible combination, to see if they contained any 7,000 different target words or phrases on a watch list. The watch list might include such words as "submarine" or "battalion," or the names of key leaders. It was all done in just three hours and fifty minutes: an average of over 30,000 intercepted messages per minute.

Like misers hoarding every last penny in a rusted treasure chest, NSA computer scientists hoard microseconds. "You save enough microseconds and 10 and behold you've got a tremendously fast machine," recalled Solomon Kullback.

Harvest not only increased NSA's speed, it also enlarged its memory, with a specially designed system that permitted the storage and retrieval of data at nearly 10 million characters per second.

Still another area advanced by Harvest was information retrieval, which used a unit known as Tractor. Tractor was capable of automatically locating desired information from a magnetic tape library of 480 reels, each capable of storing some 90 million characters. The machines would then mount, position, and thread the correct tape, and transfer the information at a then mind-boggling 1,128,000 characters per second -- " a rate," said a secret NSA document at the time, "that is still beyond present computer tape technology." Whereas most magnetic tape contained 100 bits to the inch, NSA managed to pack 3,000 bits in the same space, and then whisk them past the reading heads at 235 inches per second.

Feeding streams of intercepts from the worldwide listening posts to the analysts at NSA is a special highly secure Sigint Communications System. First opened on the eve of Pearl Harbor, the system carried over 25 million words a day by the mid-1960s. Analysts using Harvest would then further process the encrypted traffic.

Another system bears critically important intelligence from an intercept operator at a listening post in a distant part of the world straight to the president of the United States at breakneck speed. The surprise launch by the Soviet Union of Sputnik in 1957 caused an earthquake within the intelligence community. At the time, it took an average of 8 hours and 35 minutes for a message containing critical intelligence to reach the White House. President Eisenhower demanded that the time be reduced to minutes. At a National Security Council meeting on August 27, 1958, attended by Eisenhower, CIA director Allen Dulles agreed that "there was little purpose in developing critical intelligence overseas unless we had the communications means to insure its rapid transmission to Washington."

A month later, in a meeting in the Oval Office with Eisenhower, Tordella proposed a system known as CRITICOMM. After Tordella outlined the costs and benefits, Eisenhower turned to the deputy secretary of defense and said, "Do it." Within six months NSA was able to reduce transmission time from more than 8 hours to 52 minutes. In another six months the agency was able to have a CRITIC, or critical intelligence message, on Eisenhower's desk within a brief thirteen minutes, regardless of where it had originated. Eventually the time shrunk to between three and five minutes.

Finally, a system codenamed Rye provided remote access to Harvest, thus permitting analysts throughout NSA to access the main computer via several dozen distant terminals. "RYE has made it possible for the Agency to locate many more potentially exploitable cryptographic systems and 'bust' situations," said one secret report at the time. "Many messages that would have taken hours or days to read by hand methods, if indeed the process were feasible at all, can now be 'set' and machine decrypted in a matter of minutes.... Decrypting a large batch of messages in a solved system [is] also being routinely handled by this system."

Few could have foreseen Harvest's bright future when the machine was first built. Because the complexity of the system baffled even many of the best analysts, it was originally considered a white elephant. During employee tours, the huge, boxy machine was pointed to and mocked. "It's beautiful, but it doesn't work," officials would scoff But once the machine was fully understood, Harvest became so successful that it was used continuously for fourteen years. The agency finally switched to a more advanced system only in 1976.

As computers more and more became essential in both codemaking and codebreaking, worries developed over the progress the Soviet Union was making in the field, especially given its early lead in space exploration. In 1959 a top secret panel was created to investigate where the United States stood in its computer race with Russia. The results were encouraging. By then the U.S. government had about 3,000 computers, of which about 300 were high-performance machines valued at more than $1 million each. Russia, however, had fewer than 400 computers, of which 'only about 50 were large machines.

Although for a time both countries attained comparable speed -- the Soviet M-20 was about as fast as the IBM 709 -- the United States had left Russian computer scientists in the dust with the development of the transistor. Nevertheless, the secret panel's report advised against overconfidence. "The Soviet Union could achieve a computer production capability equivalent to that of the U.S. in 2-3 years, if they place the highest possible priority on the effort." But, the report added, "There is no evidence that they intend to establish such a priority." Nor, the report said, was the Soviet Union engaged in anything equivalent to Project Lightning.

Following Harvest, NSA's brain, like that of a human, was divided into right and left hemispheres, codenamed Carillon and Lodestone. Carillon was at one time made up of IBM 3605, and later of four enormous IBM 30335 linked together and attached to three IBM 22,000-line-per-minute page printers.

Even more powerful, however, was Lodestone. Dominating the center of a yellow-walled, gold-carpeted hall of computers, front-end interfaces, and mass storage units, was a decorative, 4-1/2-foot-wide, 6-1/2-foot- high semicircle of narrow gold and deep green panels surrounded by a black vinyl-upholstered bench-type seat. It appeared to be an ideal resting place for lunch or a mid-morning coffee break. It was, however, the fastest, most powerful, and most expensive computer of its time.

Built by Cray Research at its plant in Chippewa Falls, Wisconsin, a town also known for its Leinenkugel's beer and Chippewa Springs water, the $15 million CRAY-1 may be the ultimate testimony to the old proposition that looks are deceiving. Housed within what one wag once called "the world's most expensive love seat" were more than 200,000 integrated circuits, each the size of a thumbnail, 3,400 printed circuit boards, and 60 miles of wire. So compact was the five-ton, seventy-square- foot unit that enough heat was generated per cubic inch to reduce the machine to a molten mass in seconds had it not been for a unique Freon cooling system using vertical aluminum-and-stainless-steel cooling bars that lined the wall of the computer chassis.

The supercomputer was the brainchild of Seymour Gray, a shy, enigmatic engineer who rarely allowed interviews or pictures but was one of the most influential figures in computer science. The founder of Gray Research, Inc., Gray "is to supercomputers what Edison was to light bulbs," said Time in 1988, "or Bell to the telephone." When not in his laboratories, Gray could likely be found deep in the earth beneath his Wisconsin home, slowly tunneling toward the nearby woods. Eight feet high and four feet wide, the tunnel was lined with four-by-four cedar boards. When a tree once crashed through the roof of the tunnel, Cray turned the hole into a lookout with the installation of a periscope.

To Cray, the tunnel was both inspiration and recreation. "I work when I'm at home," he once told a visiting scientist. "I work for three hours and then I get stumped, and I'm not making progress. So I quit, and I go to work in the tunnel. It takes me an hour or so to dig four inches and put in the four-by-fours." Half kidding, Cray continued: "Now, as you can see, I'm up in the Wisconsin woods, and there are elves in the woods. So when they see me leave, they come into my office and solve all the problems I'm having. Then I go back up and work some more." According to John Rollwagen, then chairman of Cray Research, "The real work happens when Seymour is in the tunnel."

Cray began his career by building codebreaking machines in the 1950s with Engineering Research Associates, then headed by future NSA research chief and deputy director Howard Engstrom. Cray's dream was to build a number cruncher capable of 150 to 200 million calculations per second. It would have between 20 and 100 times the capacity of then current general-purpose computers-the equivalent of half a dozen IBM 370/195s.

In the spring of 1976 the first CRAY-1 rolled out of the firm's production plant in Chippewa Falls and directly into the basement of NSA. A second was quietly delivered to NSA's secret think tank, the Communications Research Division of the Institute for Defense Analysis at Princeton University.

The CRAY had a random access semiconductor memory capable of transferring up to 320 million words per second, or the equivalent of about 2,500 300-page books; NSA could not have been disappointed. And when it was hooked up to the computer's specialized input-output subsystem, the machine could accommodate up to forty-eight disk storage units, which could hold a total of almost 30 billion words, each no farther away than .80 millionths of a second.

In a field where time is measured in nanoseconds -- billionths of a second -- seven years is an eternity. Thus it was with tremendous excitement that in June 1983 the agency made space in its basement for a new arrival from Chippewa Falls, the CRAY X-MP. Serial number 102 stamped on its side, the machine was the first X-MP to be delivered to a customer; NSA thus had the most powerful computer in the world at the time. The six-ton brain, which contained forty-five miles of wiring and required a fifty-ton refrigeration unit to keep it cool, was revolutionary. Rather than achieving its gains in speed simply by using a faster processor, the X-MP used two processors, working in parallel. Two separate jobs could be run at the same time, or one job could run on both processors. This capability made the X-MP five times faster than even the most advanced CRAY-1, the CRAY-1S/1000.

To NSA, parallel processing was the wave of the future. Among the projects the agency was closely involved with was the Butterfly processor, which linked 148 microprocessors. Developed by the Defense Advanced Research Projects Agency's (DARPA's) Strategic Computing Program, Butterfly could have been scaled up to combine 256 or 512 or even 1,000 linked processors. Future testing included plans to link about 1 million' processors.

The X-MP arrived just in time. That same year NSA secretly put into operation an enormous worldwide computer network codenamed Platform. The system tied together, into a single cyber-web, listening posts belonging to NSA, GCHQ, and other Sigint agencies around the world, with NSA as the central brain.

Two years later, in 1985, NSA's basement complex became even more crowded with the long-awaited arrival of the CRAY-2. With its bright red Naugahyde base and transparent, blue-tinted towers of bubbling liquid coolant, Seymour Gray's latest masterpiece looked more like bordello furniture than a super number cruncher in a codebreaking factory. Nicknamed Bubbles, the $17.6 million computer was almost human, with cool, bubbling Fluorinert, also used as an artificial blood plasma, running through its system. The liquid was necessary to keep the enormous heat generated by electrons flowing through the tightly packed circuit boards from causing a meltdown.

The unit of speed used in assessing supercomputers is the "flop," "floating point operations per second." Whereas it may take the average person several minutes to calculate with a pencil the correct answer to a single multiplication problem, such as 0.0572 x 8762639.8765, supercomputers are measured by how many times per second they can solve such problems. If it takes one second to come up with the answer, including where to place the "floating" decimal point, then the computer is said to operate at one flop per second. Bubbles, on the other hand, was able to perform at an astonishing 1.2 gigaflops, or 1.2 billion mathematical calculations a second. This made it up to twelve times faster than its predecessor and 40,000 to 50,000 times faster than a personal computer of that time.

By 1988 workers were laying wires and arranging power for still another new product from the backwoods of'" Wisconsin, the CRAY Y-MP. So dense were the chips on the new machine that engineers were now able to squeeze eight processors into a space originally designed for only one. Working together, and under ideal conditions, the processors were capable of performing between 2 billion and 4 billion operations a second.

In the mid to late 1980s, the pace of supercomputer development was so fast that NSA barely had enough time to boot up each new megamachine before a newer one was wheeled into its basement "flop house."

The race to build the fastest supercomputer began to resemble a mainframe Grand Prix. Sleek, shiny, and ever more powerful new machines were continuously zooming to the starting line while engineers worked on ever more powerful and speedy designs. Nobody wanted to be left in the dust. In September 1987, Steve Chen, the Chinese-born computer superstar who lead the Cray Research design team on the X-MP and Y-MP projects, left Cray after his machines became too expensive and risky. He was quickly hired by IBM. "Five, years from now," boasted an IBM executive, "we should be at 100 billion gigaflops. A problem that takes three months to do now, we want to do in a day."

Off in the shadows, the Sandia National Laboratory, in Albuquerque, was tweaking a chunky little blue box. Three feet on a side and known as the Ncube, or hypercube, the computer was "massively" parallel, with 1,024 processors, each as powerful as a traditional minicomputer. In a test, Sandia asked the computer to calculate the stresses inside a building beam supported only at one end. A powerful minicomputer working twenty-four hours a day would have taken twenty years to arrive at an answer, but the lightning-fast Ncube accomplished it in a week.

At ETA, a subsidiary of Control Data Corporation, a dark, bubble-topped box known as the ETA 10 was unveiled. An eight-processor powerhouse, it used computer chips that were smaller and denser than those used by Cray Research. Liquid nitrogen carried away the excess heat. And by using only one circuit board, the engineers were able to reduce the space that electrons have to travel during calculations. The end result was a $30 million black box designed to operate at a peak rate of 10 billion calculations per second, 30 times faster than previous supercomputers.

Not to be outdone, Los Alamos National Laboratories, by stringing together an array of supercomputers and associated networks, was able to perform more computing work in a twenty-four-hour period than had been done by all of humanity before the year 1962. And that estimate was considered conservative by other researchers, who suggested that a date in the late 1970s might be more accurate.

The speed of electrons, however, was not NSA's most immediate problem; the agency was also worried about the speed of the Japanese. Japan was the only other nation aggressively pursuing supercomputer development. In the summer of 1988, a gathering of leading computer science experts, among them NSA's director of supercomputer research, met to assess Japan's progress in supercomputers. If they felt confident when they walked into the meeting, they were more than a little nervous when they left. Starting only six years earlier, Japan had already matched or surpassed the United States in a field the United States invented and had been advancing for twenty years.

The main problem for the American supercomputer industry was dependence on Japanese computer companies-their arch-competitors in a cutthroat business-for critical parts, such as computer chips, for their machines. This was a result of the gradual abandonment of semiconductor manufacturing in the United States during the mid to late 1980s. In 1986, for example, NSA was virtually dependent on a Japanese company, Kyocera, for critical components that went into 171 of its 196 different computer chips, according to the minutes of a Department of Defense study group. When, without warning, Kyocera stopped making a component known as a ceramic package, used in a key chip, NSA began to shudder.

In a worst-case scenario, Japanese computer manufacturers could slow down or cut off the supply of essential components to their American supercomputer competitors -- and NSA. This fear led the panel to conclude that within a few years, "U.S. firms would be most fortunate if they found themselves only a generation or so behind."

As a result of such worries, NSA, with the help of National Semiconductor, built its own $85 million microelectronics production and laboratory plant, known as the Special Processing Laboratory. Located in Crypto City, the ultra-modern, windowless, 60,000-square-foot building first began producing chips in 1991. Today it employs several hundred people. The building contains 20,000 square feet of "class 10" clean rooms -- rooms whose air is 10,000 times cleaner than normal air. The water must also be ultra-pure because the particles in the water can destroy a transistor.

Building its own plant also solved another problem for NSA: the need for supersecrecy in producing highly customized parts for use in the agency's unique codebreaking machines. These components, "applications specific integrated circuits" (ASICs), are often the "brain" of a codebreaking system, thus making outside procurement "a nightmare," said one NSAer. With the ability to squeeze 1 million or more transistors on a single piece of silicon, designers can now build entire algorithms on a chip-a complete crypto system on a piece of material many times smaller than a dime. For such a chip to fall into the wrong hands would be disastrous.

So NSA added another new feature: a secret self-destruct mechanism. Developed by Lawrence Livermore and Sandia National Laboratories, NSA's chips are shielded by special self-destructing coatings. "If a hostile agent tries to take off the lid," said one knowledgeable source, "the coating literally rips out the top [circuit] layer."

Six months after the 1988 computer science panel meeting, fear over Japan's rapid push into the supercomputer industry once again surfaced. On December 6, 1988, Japan's Fujitsu-a key supplier of critical chips to Cray -- announced a major new advance: a blisteringly fast computer with a theoretical top speed of 4 billion operations per second. This equaled and perhaps beat Cray's top-of-the-line machine, the Y-MP, which had been on the market for less than a year. The problem for NSA was that the Japanese company could easily sell the superfast computer to other nations, which might then use it to develop encryption systems far too fast for NSA's codebreaking computers to conquer.

But while Japanese companies were catching up and maybe even passing their American competitors in speed, the U.S. supercomputer industry was far ahead in both software development and the use of parallel processing. As fast as the Fujitsu computer was, it had only two processors. Cray and ETA had both developed machines with eight processors -- eight brains, in a sense -- which could simultaneously attack separate parts of a problem.

To Seymour Cray, sixteen brains were better than eight, and for several years he had been trying to prove it by building a sixteen processor CRAY-3. It was an expensive and time-consuming endeavor -- too much so, it turned out, for Cray Research, the company he had founded but no longer owned. In May 1989, the two split, Seymour Cray took ',200 employees and $100 million and moved to Colorado Springs to found Cray Computer, Inc., as a wholly owned subsidiary of Cray Research. Eventually, it was planned, Cray Computer would be come independent.

Like a race-car driver with his foot stuck to the accelerator, Cray continued to push for more and more speed; he hoped to break sixteen billion operations a second. The secret would be to make the hundreds of thousands of chips that would constitute the soul of the new computer not out of conventional silicon but out of a radical new material: gallium arsenide. Although it was more difficult and costly to work with, electrons could travel up to ten times as fast through the new compound as through silicon.

But as "the Hermit of Chippewa Falls," as Cray was affectionately known, quietly pushed ahead in his new laboratory in Colorado Springs, the world around him began shifting and turning. The Cold War had ended and weapons designers were no longer shopping for supercomputers. The fat Reagan years of Star Wars were giving way to the Clinton era of cutbacks and deficit reduction. And industry was turning away from the diamond-encrusted CRAYs, made of a small number of superpowerful processors, and toward less pricey massively parallel computers made up of thousands of inexpensive microprocessors. The enormously expensive, hand-built Formula One racers were being forced off the track by cheap stock cars packed with store-bought superchargers and sixteen-barrel carburetors.

At ETA Systems, which had pushed the supercomputer speed envelope with its ETA 10, 800 employees showed up for work on a spring Monday in 1989 to find the doors locked shut. The company had developed a super debt of $400 million.

Four years later, Steve Chen folded up his new company, Super-computer Systems, when IBM finally cut off funding for his 55-1. Partly funded by NSA, Chen had spent half a decade attempting to build a computer a hundred times faster than anything on earth. But in the end, the innovations were overtaken by excessive costs and endless missed deadlines. A few months after the company's doors closed, one of its former engineers driving past a farm spotted a strange but familiar column of metal. A closer look confirmed his worst fears: it was the outer frame for the SS-1, and it had been sold for scrap.

In 1991, Thinking Machines Corporation delivered to NSA its first massively parallel computer -- the Connection Machine CM-5, which the agency named Frostberg. Used until 1997, the futuristic black cube with long panels of blinking red lights looked like something left over from a Star Wars set. Just two years after the $25 million machine was installed, NSA doubled its size by adding 256 additional processor units. This allowed Frostberg to take a job and break it into 512 pieces and work on each piece simultaneously, at 65.5 billion operations a second. Equally impressive was the Frostberg's memory, capable of storing up to 500 billion words.

By the time the. CRAY-3 at last made its debut in 1993 -- clocking in at roughly 4 billion operations a second -- there were no takers. Nearly out of money, the company spent a year looking for customers and finally landed a deal with its old partner, NSA. In August of 1994, the agency awarded Cray $4.2 million to build a highly specialized version of the CRAY-3 for sign~ processing and pattern recognition -- in other words, eavesdropping and codebreaking. Named the CRAY-3/Super Scalable System, the machine would become the brain of what has been dubbed "the world's ultimate spying machine." It would link two supercomputer processors with a massively parallel array of chips con taining more than half a million inexpensive processors designed by NSA's Supercomputer Research Center.

But while hoping for Cray to succeed, NSA scientists were also working in-house on new ideas. One was a processor called Splash 2, which, when attached to a general-purpose computing platform, was able to accelerate the machine's performance to super-Cray levels at only a fraction of the Cray cost.

As Seymour Cray struggled to complete his CRAY-5, he was also in a race with his old parent company, Cray Research, which was building a successor to its Y-MP called the C-90. The company was also near completion on a powerhouse known as the T-90, which would operate at up to 60 billion operations per second. Meanwhile, Seymour Cray hoped to leapfrog his competitors once again with his CRAY-4, due out in 1996.

By the fall of 1994, work on the CRAY-4 was going surprisingly well. Cray Computer in Colorado Springs was predicting a completion date in early 1995 with a machine with twice the power of the CRAY~3 at one-fifth the cost. There was even talk of a CRAY-5 and CRAY-6 before the planned retirement of Seymour Cray. Which was why the yellow tape came as such a shock. When employees came to work on the morning of March 24, 1995, they were first confused to see the yellow police tape sealing the doors. But when they saw the white flag that had been run up the flagpole, they did not need a supercomputer to conclude that the end had finally arrived. The man with the unlimited ideas that reached to the stars had tumbled to the bottom of his finite bank account.

Ever optimistic, Seymour Cray pulled together a few of his most loyal followers, scraped together some money from their own bank accounts, and formed SRC (Seymour Roger Cray) Computers. Cray felt almost liberated at this chance to "start from a clean sheet of paper." It was also, he believed, a chance to finally break the speed barrier by building the first teraflop supercomputer, capable of a trillion mathematical operations a second-12,000 times more than his CRAY-1.

But the enemy had landed. In the spring of 1996, even the U.S. government had turned its back on all the Cray companies and awarded a $35 million Contract to the Japanese computer giant NEC for its 12B-processor SX-4 supercomputer. The SX-4 would go to the National Center for Atmospheric Research. The agency was worried because meteorological centers in Australia, Canada, England, and elsewhere were installing systems that by January 1998 would be capable of between 20 and 80 billion operations a second. And Cray Research, the agency concluded, was just not producing computers fast enough. "Simply put," said William Buzbee, head of the weather center, "Cray Research lost this procurement because their offer had unacceptable technical risk."

Others, too, knew that despite the never-say-die bravado and the endless promises of millions of flops, the luster was at last disappearing from Cray's blinding star. "The rules changed when it became clear that Cray Computer Corp. wasn't going to make it," said John Mashey, director of systems technology at Silicon Graphics. "It's like watching your favorite quarterback, who won the Super Bowl many times. But it's not 1976 anymore -- his knees are gone and those three- hundred-pound defensive tackles are fierce. While he keeps getting up, it's agonizing to watch and you really wish he could have quit on a high."

A few months later, while returning from a brief trip to a software store, Cray was seriously injured when his black Grand Cherokee was struck by another car and rolled over three times. Two weeks later, on October 5, 1996, the shy maverick who hand-built the fastest machines on earth, with the meticulous care and fine craftsmanship of a Swiss watchmaker, died, never having regained consciousness. His ashes were scattered among the cragged peaks and somber valleys of the Colorado mountains. They had served as his inspiration, and as silent comforters, during his last years. "In the days before PCs brought megaflops to the masses," said one computer expert, "Cray was the computer industry's closest equivalent to a rock star."

Sadly, only months before Cray's death, the daring company he had given birth to in Chippewa Falls, Wisconsin, decades earlier, also died. Following the worst financial year of its life, in which it was forced to cut nearly a quarter of its employees, and facing an uncertain future, Cray Research called it quits. It was acquired by Silicon Graphics, Incorporated -- later known simply as SGI -- a Mountain View, California, manufacturer of high-performance workstations, the son of machines that became Cray's greatest competitor. "Cray represents the last of the 1980s pure plays in the supercomputer market," one market analyst said wistfully. "There are no other major players left standing from the supercomputer battles of the 1980s and 1990s."

In fact, there was one. The shakeout and the death of Seymour Cray left a single independent to fight the army of "killer micros," the massively parallel microprocessors that turned the budget-draining, high-performance supercomputer into an endangered species. The large, rumpled man with the Don Quixote dream was Burton Smith, whose company, Tera Computer, stunned many in the field by building a machine that in 1997 set a world speed record for sorting integers. Burton's idea was to increase speed by decreasing the waiting time it took for processors to be sent new data on which to work. This, Burton believed, would overcome the Achilles' heel of powerful computing -- the gap between a computer's short-term theoretical "peak" speed and its long-term "sustained" speed.

Smith no doubt had his eye on NSA as a key future customer for his machines, which would cost as much as $40 million. He spent three years working for NSA's Supercomputer Research Center before leaving in 1988 to found Tera. Much of his early money, in fact, came from NSA's partner, DARPA.

Encouraged by Smith's research, a "senior intelligence official" approached Sid Karin, the director of the San Diego Supercomputer Center, and asked him to help support Tera. "We don't have a lot of innovative architects like Burton Smith and Seymour Cray," the intelligence official told Karlin, "and they need to be nurtured and supported." So, in 1998, Smith installed his first system in the San Diego center.

Nevertheless, Smith still has his skeptics. One well-known computer designer fondly refers to the Tera system as "Burton's folly." And even Smith acknowledges the long odds: "Most people think we're out of our mind." Still, noted one observer, ((Burton Smith is the last man standing."

***

As the supercomputer business began crashing, worries increased at NSA. For decades the agency had quietly underwritten a large portion of the industry; the massive number crunchers were the engines that powered its codebreaking machines. Now agency officials watched SGT, following its takeover of Cray, like spectators at a slow- motion automobile ccident. Within a year and a half of the acquisition the company was in turmoil. SGI posted a loss of Over $50 million, a major layoff was announced, and the longtime chief executive officer resigned. Noting that only three years earlier the company had produced the graphics that made the motion picture Jurassic Park possible, one reporter quipped, "The question was whether the company was in danger of going the way of the dinosaur."

By 1999, SGI looked like a boxer struggling to rise before the final count. Its stock had plunged more than 20 percent, another chief executive officer had called it quits, and the firm said it would cut as many as 3,000 jobs and spin off its Cray supercomputer division. NSA was worried: it had contracted with the company to build its newest supercomputer, the CRAY SV2. [1]The decision was made to open the drawer of the cash register. "The United States is committed to maintaining and building on its long-held position as the global leader in supercomputing," said NSA's chief scientist, George Cotter. "These powerful computers are absolutely essential to U.S. national security interests. To that end, the U.S. government is committing significant support to SGI's CRAY SV2 program."

Cotter also noted the critical need for NSA to continue similar joint supercomputer projects. "The government support reflects a continuing need for government-industry Cooperative development of critical technologies for high-end computing," he said. "The SV2 will include technology jointly developed with the U.S. government. This will considerably extend the combination of custom-designed high-end processors with the high-speed memory access that Current Cray supercomputers offer." The new system was expected to dramatically extend the capability of NSA's supercomputers with exceptional memory bandwidth, interconnections, and vector-processing capabilities. Its peak speed was estimated to be in the tens of teraflops, faster than any supercomputer in existence.

In 2000 the Supercomputer business came full circle. Like two broke gamblers at a racetrack putting their change together for one last bet, Burton Smith's Ten Computer acquired Seymour Cray's former Cray Research from SGI. Thus was reborn Cray, Inc., once again an independent company. It was good news for NSA. One report said the agency was involved in the deal "because it wants at least one U.S. company to build state-of-the-art supercomputers with capabilities beyond the needs of most business customers." Work would continue on NSA's SV2, with delivery scheduled for 2002.

At the same time, Cray began work on a new Department of Defense contract, one to upgrade a CRAY T3E-1200 supercomputer. With the addition of 816 processors to its existing 272 processors, the new machine will be the largest Cray system ever built, with 1,088 processors and a record speed of 1.3 teraflops -- 1.3 trillion calculations per second. Four years after Seymour Cray died, a machine bearing his name would at last break the tera barrier.

But despite the encouraging signs, the supercomputer shakeout had convinced many at NSA of the need to move away from the insecurity of the outside world and to return to the black computer laboratories of Crypto City.

***

The massive brain transplant began in February 1997, as the first supercomputer began its slow trip from the basement of OPS 1. Its destination was the top floor of the Tordella Supercomputer Facility, hidden away in a wooded corner of Crypto City. More than a year later, the final supercomputer was carefully nudged into place and connected by a spinal cord of secure fiber optic nerves to the main body of the agency, a mile away. Once the operation was completed, NSA possessed the most powerful electronic brain on earth.

Surrounded by thick woods and protected by guard posts, double fences, and concrete barriers, the Tordella Supercomputer Facility, is located on Crypto City's Ream Road, a street named after NSA's fourth deputy director. The nearly windowless outside walls of the 183,000-square-foot facility are decorated with attractive, light-colored enameled metal panels. The life-support equipment is housed on the first floor -- an 8,000-ton chilled water plant, mechanical and electrical support facilities, and 29-megavolt-amperes of electrical power, enough to supply half of Annapolis.

The top floor's five rooms contain, among other things, the Computer Operations Command Center and approximately 150,000 magnetic tapes moved there from storage in "silo-farms" back in the main part of Crypto City. Supercomputers, such as the GRAY Y-MP EL and the Silicon Graphics Power Challenge, occupy the rest of the floor. Also installed in 1999 was the new IBM RS/6000 SFP This is a faster version of the system that powered the company's supercomputer "Deep Blue," which won a grueling six-game chess match against virtuoso Garry Kasparov in 1997. The extra power and speed come from IBM's new Power 3 microprocessor, which is capable of crunching through 2 billion instructions per second-more than double the power of the Power2 Super Chip. The computer is the centerpiece for a system IBM called Deep Computing. One of its primary uses is "data mining," searching through enormous quantities of data, such as intercepted communications or complex cipher systems, and coming up with the answer. The RS/6000 SP, said IBM executive David Turek, is "supercomputing at your fingertips."

Moving the tremendous amounts of information into and out of the supercomputers, like the ultimate jukebox, is the massive dodecagonal Automated Cartridge System. As big as a small room, and weighing more than four tons, this high- peed storage device can hold 6,000 cartridges containing a total of 300 terabytes of information -- the equivalent of more than 150 billion pages of text. According to NSA, this is the equivalent of one and a half million years of the wall Street Journal; it is also enough pieces of paper to circle the globe 3,000 times, or to fill a wall of books stacked eleven deep and running from New York City to Los Angeles.

The robotic arm has two cameras and a "hand"; the cameras find the bar code of the requested cartridge, and the hand moves it to the retrieval area, where the needed cartridge can be extracted. The arm can move cartridges in and out of the computers at the rate of 450 an hour.

Such a system is necessary when one considers NSA's information storage capabilities. To store the massive amounts of data flowing in from its worldwide listening posts, NSA a few years ago turned to E-Systems, long a key contractor on secret projects for the agency. The solution was to link several computers the size of telephone booth's. When completed the system was capable of storing 5 trillion pages of text-a stack of paper 150 miles high. Included was a new retrieval system that permitted the access of any piece of information almost instantly.

***

As the supercomputer industry began crumbling around it, NSA turned inward, creating a top secret facility for developing its own classified computers. Known as the Supercomputer Research Center (SRC), it was built in 1984 in order to leapfrog over the rest of the world in computer power, as Project Lightning had thirty years earlier. Only this time, the work would be done in total secrecy. According to Lieutenant General Lincoln D. Faurer, the NSA's director at the time, a principal goal of the SRC was to build a new generation of computers that would be 10,000 times faster than the current generation.

Over the years millions of dollars would go into research on subjects such as specialized parallel processing algorithms, which would give computers the superspeed needed to break increasingly powerful foreign encryption systems. At the same time, SRC would develop ways to push American cryptographic systems beyond the reach of hostile codebreakers. Little, if any, of the research done by the SRC would ever see the light of day outside of Crypto City, so NSA would be far ahead in the race for the fastest and most powerful computers on earth.

Constructed at a cost of $12 million on a twenty-acre site at the University of Maryland's Science and Technology Center in Bowie, the SRC is actually operated by the Communications Research Division (CRD), part of the Institute for Defense Analysis. For more than four decades the CRD has run NSA's own highly secret think tank. Originally known as the NSA Research Institute, it was first approved by President Eisenhower in 1958. Its purpose was to carry out long-range, theoretical, and advanced research in mathematical and statistical problems related to NSA's codebreaking and eavesdropping missions. The institute also conducted a special summer program that brought together members of the academic community and introduced them to members of the cryptologic community.

At one point, in 1965, the institute developed a unique piece of codebreaking machinery that proved enormously successful. "That one piece of equipment," said a secret 1965 NSA report, "by itself, has been judged to be well worth the total cost of the Institute thus far."

Among the early directors of the institute was Dr. J. Barkley Rosser, a professor of mathematics at the University of Wisconsin, noted for his work in symbolic logic and number theory. Dr. A. Adrian Albert, dean of the division of physical sciences at the University of Chicago and an expert in linear algebra and number theory, followed him in 1961.

Originally, the NSA Research Institute was located behind a high wall on the campus of Princeton University. But as a result of the antiwar protests of the 1960s, NSA, fearing for the continued secrecy and-security of the institute, moved it to a boxy, three-story brick building virtually hidden in an isolated wooded area off campus. Windowless except for the third floor, the mysterious building has no signs to indicate the name or nature of the occupant. Eventually, to further hide its connection to NSA, the Research Institute's name was changed to the Communications and Computing Center. Specializing in such esoteric codebreaking and eavesdropping disciplines as cryptomathematics, cryptocomputing, speech research, and special signals processing techniques, the IDA-C3I, as it is sometimes known, received $34 million in funding in 1994 and employed a technical staff of 149.

In addition to the Supercomputer Research Center, NSA also has a Laboratory for Physical Sciences (LPS), which is part of the agency's Directorate of Technology. Like the NSA Research Institute, LPS was born in the 1950s, when the NSA's Scientific Advisory Board recommended that the agency establish a "window on the world of academia and academic research in the physical sciences." As a result, the agency collaborated with the University of Maryland to create the LPS, with quarters built adjacent to the school's College Park campus.

In 1992 the LPS moved into a new, nondescript 63,500-square-foot building on Greenmead Drive in College Park. Leased from the university for $480,000 a year, the facility, near a Moose lodge, draws little attention and does not appear in the campus telephone directory. "We don't know what they do there," said the administrator of the veterinary center next door.

The lab was built at a cost of $10.9 million; its ultra-advanced technology is designed to fast-forward NSA's ability to eavesdrop. Using magnetic microscopy, scientists are able to study the minute tracks on magnetic tape and greatly increase data density, thus enabling intercept operators to pack ever more conversations into their recorders. Increasing computer speed is also critical. To achieve this acceleration, the LPS ontains a state-of-the-art molecular beam epitaxy (MBE) facility to develop miniature lasers, optical amplifiers, and other components made out of gallium arsenide.

But speed equals heat. Thus the LPS is also pushing the limits on such technologies as the development of synthetic diamonds, which are many times more efficient for heat conduction than copper and far less expensive than real diamonds. For example, an integrated circuit mounted on ordinary ceramic will turn a very warm 87 degrees centigrade when its surroundings are at room temperature. One mounted on synthetic diamonds, however, will reach only 54 degrees centigrade, allowing NSA's codebreaking machines to be relatively cool as well as fast.

Speed not only equals heat, it also equals massive demand for data storage. Increasing use of space-eating multimedia files compounds the problem, as does the need to make the information available to an ever larger group of customers. One solution was Project Oceanarium, which for the first time automated the storage of NSA's masses of multimedia Sigint reports.

At the same time, Oceanarium modernized the way in which reports were retrieved and distributed. Where once each spy agency jealously guarded its individual intelligence files behind thick fortresses, today the buzz phrase is "sharable databases." Through Oceanarium, NSA's dark secrets can now be retrieved not only over its own internal intranet, Webworld, but throughout the intelligence community via highly classified programs such as Intelink.

Because the breadth and depth of NSA's data storage sea is finite, scientists are turning to newer ways to narrow the rivers of information emptying into it. Among the most promising are microscopic magnets, only one molecule in size. Scientists at Xerox believe that such a magnet, made of a special combination of manganese, oxygen, carbon, and hydrogen, may be able to pack data thousands or even millions of times more densely than today's systems of memory storage. Using these molecule-sized magnets, experts believe, it may someday be possible to store hundreds of gigabytes of data -- millions of typed pages -- on an area no larger than the head of a pin.

By 2001, NSA's tape and disk storage capacity approached a density of ten gigabytes per square inch-the equivalent of more than half a million typed, double-spaced pages. But the closer data are packed, the harder they are to erase and the more chance that telltale secrets will remain behind on reused media. Therefore, another key area of research at NSA's LPS is exploring the microscopic properties of data storage and erasure to find more effective ways to rid used tapes and hard drives of all their old secrets. According to computer expert Simson Garfinkel, tiny pieces of a hard drive can still contain sizable amounts of information. For instance, a 1/16-inch-square piece of a six-gigabyte hard drive can hold 750,000 bytes -- enough to fill a 300-page book. "A spy could remove a hard disk, grind it up, and smuggle out the data in little pieces like pocket lint," said Garfinkel. To solve the problem, NSA developed a drive-controlled disk sanitization device, which attaches to the head disk assembly and can completely eradicate the sensitive information used on disks and drives.

Inside NSA's Supercomputer Research Center, the secret race for the fastest computer seems almost unworldly. In 1994 and 1995 NSA scientists participated in a series of meetings devoted to exploring the feasibility of a great leap forward in computer technology. The goal was to advance from billions, past trillions, to more than a quadrillion operations a second -- pentaflop speed -- within two decades.

Among the ideas developed by NSA for achieving speeds of over a quadrillion (10¹⁵) mathematical operations a second was the placement of processors in the middle of memory chips. Processor-in-memory chips, or PIMs, have the advantage of reducing the time it normally takes for electronic signals to travel from the processors to the separate memory chips. These PIM chips are now among the products manufactured by the agency's Special Processing Lab.

By 2001, the SRC had long since broken the teraflop barrier and was approaching petaflop speeds -- at which point time is measured in femtoseconds, the shortest events known to science. With such extraordinary speed, a machine would be capable of pounding a stream of intercepted, enciphered text with a quadrillion -- a million billion-possible solutions in the time it takes to wink. Original estimates by scientists were that the outside world would reach that point sometime around 2010, but IBM intends to cut the wait in half with a megasupercomputer dubbed Blue Gene.

Over five years, between 2000 and 2005, the company plans to build the fastest computer on earth -- 500 times faster than anything currently in existence. "It will suck down every spare watt of electricity and throw off so much heat that a gas turbine the size of a jet engine is required to cool it off," said one report. According to the company, the computer would be about forty times more powerful than the aggregate power of the forty fastest supercomputers in the world today -- or 2 million times more powerful than the fastest desktop in. existence.

The ultimate goal of Blue Gene is to solve a puzzle of a different sort from those at NSA -- although NSA may also secretly be a customer. Blue Gene's singular objective is to try and model the way a human protein folds into a particular shape. Because proteins are the molecular workhorses of the human body, it is essential to discover their molecular properties. In a sense, Blue Gene is like NSA's old RAMs, which were designed to attack one specific encryption system.

When completed, Blue Gene will consist of sixty-four computing towers standing six feet high and covering an area forty feet by forty feet. Inside will be a mind-boggling one million processors. The target speed is a petaflop.

When NSA crosses the petaflop threshold, if it hasn't already, it is unlikely that the rest of the world will know. By 2005 the SRC, with years of secret, highly specialized development accumulated, will likely be working with computers operating at exaflop speeds -- a quintillion operations a second-and pushing for zettaflop and even yottaflop machines, capable of a septillion (10²⁴) operations every time a second hand jumps. Beyond yottaflop, numbers have not yet been named. "It is the greatest play box in the world," marveled one agency veteran of the NSA's technology capability. "They've got one of everything."

Operating in the exaflop-and-above world will be almost unimaginable. The key will be miniaturization, an area in which NSA has been pushing the theoretical envelope. By the mid-1990s, NSA's Special Processing Laboratory had reduced the size of a transistor so much that sev enty of them would fit on the cross section of a human hair. NSA is also attempting to develop a new generation of computer chips by bombarding light-sensitive material with ions to etch out microscopic electronic circuit designs. Using ion beams instead of traditional light in the process provides the potential for building far smaller, more complex, more efficient chips.

In the late 19905 NSA reached a breakthrough when it was able to shrink a supercomputer to the size of a home refrigerator-freezer combination. Eventually the machine was pared down to the size of a small suitcase, yet its speed was increased by 10 percent. In 1999, a joint NSA and DARPA program demonstrated that portions of a supercomputer could be engineered to fit into a cube six inches on a side -- small enough to fit into a coat pocket. The circuitry was made of diamond-based multichip modules and cooled by aerosol spray to remove the 2,500 watts of heat from the system.

But to reach exaflop speed, computer parts -- or even computers themselves -- may have to be shrunk to the size of atoms, or even of subatomic particles. At the SRC, scientists looking for new and faster ways to break into encryption systems have turned to quantum computing. This involves studying interactions in the microscopic world of atomic structures and looking for ways to harness individual atoms to perform a variety of different tasks, thereby speeding up computer operations to an unthinkable scale.

NSA has had a strong interest in quantum computing as far back as 1994, when Peter Shor, a mathematician at 'Bell Laboratories, which has long had a close and secret relationship with the agency, discovered the codebreaking advantages of the new science. Since then, NSA has spent about $4 million a year to fund research at various universities, and put additional money into studies at government laboratories.

Operated at top speed, a quantum computer could be used to uncover pairs of enormously large prime numbers, which are the "passwords" for many encryption systems. The largest number that ordinary supercomputers have been able to factor is about 140 digits long. But according to another Bell Labs scientist, Lov K. Grover, using quantum computing, 140-digit-long numbers could be factored a billion times faster than is currently possible. "On paper, at least," said Glover, "the prospects are stunning: ... a search engine that could examine every nook and cranny of the Internet in half an hour; a 'brute-force' decoder that could unscramble a DES [Data Encryption Standard-the encryption standard used by banks and most businesses] transmission in five minutes."

A quantum computer could also be used to speed through unfathomable numbers of intercepted communications -- a "scan" in NSA-speak -- searching for a single keyword, a phrase, or even, with luck, a "bust." Long the secret leading to many of NSA's past codebreaking successes, a bust is an abnormality -- sometimes very subtle -- in a target's cryptographic system. For example, it may be an error in a Russian encryption program, or a faulty piece of hardware, or a sloppy transmission procedure. Once such a hairline crack is discovered, NSA codebreakers, using a massive amount of computer power in what is known as a brute force attack, can sometimes chisel away enough of the system to expose a golden vein of secret communications.

A breakthrough into quantum computing came in April 1998, when researchers at MIT, IBM, the University of California at Berkeley, and the University of Oxford in England announced they had succeeded in building the first working quantum computers. The processor consisted of a witches' brew of hydrogen and chlorine atoms in chloroform. Digital switches were shrunk down to the smallest unit of information, known as a quantum bit, or qubit. Where once a traditional computer bit would have to be either, for example, 0 or 1, a qubit could be both simultaneously. Instead of just black or white, a qubit could become all the colors of the rainbow.

According to John Markoff, who has long followed the issue for the New York Times, another milestone came in July 1999. That was when researchers at Hewlett-Packard and the University of California at Los Angeles announced that they had succeeded in creating rudimentary electronic logic gates -- one of the basic components of computing -- only a single molecule thick. Four months later, scientists at Hewlett-Packard reported they had crossed another key threshold by creating rows of ultramicroscopic conductive wires less than a dozen atoms across.

Translated into practical terms, a quantum computer could thus perform many calculations simultaneously, resulting in a hyperincrease in speed. Now, instead of a supercomputer attempting to open a complex cipher system -- or lock -- by trying a quadrillion different keys one after another, a quantum computer will be able to try all quadrillion keys simultaneously. Physicists speculated that such machines may one day prove thousands or even millions of times faster than the most powerful supercomputer available today.

The discovery was greeted with excitement by the codebreakers in Crypto City. "It looked for a long time like a solution without a problem," said NSA's Keith Miller. At Los Alamos, where NSA is secretly funding research into the new science, quantum team leader Richard J. Hughes added: "This is an important step. What's intriguing is that they've now demonstrated the simplest possible algorithm on a quantum computer."

Also heavily involved in molecular-scale electronics, known as moletronics, is DARPA, long NSA's partner in pushing computing past the threshold. Scientists working on one DARPA program recently speculated that it may soon be possible to fashion tiny switches, or transistors, from tiny clusters of molecules only a single layer deep. Such an advance, they believe, may lead to computers that would be 100 billion times as fast as today's fastest PCs. According to James Tour, a professor of chemistry at Rice University who is working on molecular-scale research, "A single molecular computer could conceivably have more transistors than all of the transistors in all of the computers in the world today."

On the other side of the city, however, the codemakers welcomed the news with considerable apprehension. They were worried about the potential threat to NSA's powerful cipher systems if a foreign nation discovered a way to harness the power and speed of quantum computing before the United States had developed defenses against it. By 1999, for example, Japan's NEC had made considerable progress with the development of a solid-state device that could function as a qubit. "We have made a big step by showing the possibility of integrating quantum gates using solid-state devices," said NEC's Jun'ichi Sone. "It takes one trillion years to factorize a two-hundred-digit number with present supercomputers," he said. "But it would take only one hour or less with a quantum computer."

As intriguing as quantum computing is, perhaps the most interesting idea on how to reach exa-speed and beyond came out of the series of "great leap forward" meetings in which the NSA took part in the mid-1990s. The computer of the future -- already with a circulatory system of cool, bubbling Fluorinert, an artificial blood plasma -- may be constructed partly out of mechanical parts and partly out of living parts.

"I don't think we can really build a machine that fills room after room after room and costs an equivalent number of dollars," said Seymour Gray, one of those at the meetings. "We have to make something roughly the size of our present machines but with a thousand times the components," One answer to scaling down to the nanometer, according to Cray, was to fabricate computing devices out of biological entities. At the same time, other biological processes could be used to manufacture nonbiological devices -- for example, bacteria could be bioengineered to build transistors.

By 2001, researchers at MIT were actively attempting to marry the digital with the biological by altering the common E. coli bacterium to function as an electronic circuit. Such a melding would produce a computer part with the unique ability to continually reproduce itself. Through such a process, enormous numbers of nearly identical processors could be "grown." "We would like to make processors by the wheelbarrow-load," said MIT computer scientist Harold Abelson. Abelson and his colleagues are hoping to someday map circuitry onto biological material, in a process they call amorphous computing, thus turning living cells into digital logic circuits. However, since the cells could compute only while alive, millions or billions of the tiny biocomponents would have to be packed into the smallest spaces possible.

Bell Labs, part of Lucent Technologies, is also perusing the idea of a "living" computer by creating molecular-size "motors" out of DNA -- motors so small that 30 trillion could fit into a single drop of water. According to Bell Labs physicist Bernard Yurke, it might eventually be possible to bind electronic components to DNA. Then, by linking the DNA strands together, a computer could be created with incredible speed and storage capacities.

Eventually NSA may secretly achieve the ultimate in quickness, compatibility, and efficiency -- a computer with petaflop and higher speeds shrunk into a container about a liter in size, and powered by only about ten watts of power: the human brain.

_______________

NOTES:

1. Despite the capabilities of massively parallel computers, supercomputers are still useful for attacking specific codebreaking problems.

Go to Next Page