COMPUTER WIZARD: Dr Philip Emeagwali
Inventor Of The Worlds Most Fastest Computer
Quietly working in a U-M lab, grad student Philip Emeagwali has pulled off a prize-winning feat in supercomputing. Steve Eisenberg reports for the April 26, 1990 issue of the Ann Arbor News (Michigan, USA)
The high school physics teacher was stumped. He stood at the front of the class, staring at the blackboard, not sure how to finish the problem. Within moments, Philip Emeagwali walked forward, grabbed the piece of chalk from the teacher's hand and figured it out.
"He could always challenge the instructor," says fellow classmate Peter Ozoh, 36, now a chemical engineer at Hercules Aerospace in Desoto, Kan.
About 20 years after that day in Nigeria, Emeagwali is still amazing people. The 35-year-old graduate student at the University of Michigan has won one of the nation's most prestigious awards in the computing field, the Gordon Bell Prize.
Emeagwali has programmed a computer to work faster than any other computer --- at a rate of 3.1 billion calculations each second --- to solve one of the nation's 20 most difficult computing problems: understanding how oil flows underground so companies could extract the most"Texas gold."
Typically, oil is trapped within rocks --- like water in a drenched sponge --- and oil companies can remove only 5 percent to 50 percent. Now, simulations of oil fields, which help track the viscous stuff, will take seconds, instead of hours, to produce on a computer. In addition, the increased accuracy may boost the available amount of oil by a few percentage points, Emeagwali says.
"He has made a significant accomplishments in a computer science sense," says Alvis E. McDonald, a research scientist who simulates oil fields at Mobil Research and Development Corp., in Dallas.
Emeagwali not only was the award's first solo winner --- usually, teams from corporations and national laboratories capture the $1,000 check --- but he's also considered a novice in the field. When he won the award in late February, only eight months had passed since he first used a relatively new type of computer that helped him clinch the award.
Like other supercomputers, the Connection Machine, which is made by Thinking Machines Corp., in Cambridge Mass., solves problems in minutes that would take years on a desktop personal computer. The Connection Machine, however, operates differently from its competitors --- most notably, those made by Minneapolis-based Cray Research Inc., which commands 70 percent of the market.
Many researchers believe the Connection Machine is more difficult to use and can solve only specialized problems. In fact, when Emeagwali told his colleagues he had attained his prize-winning speed, they doubted him.
"When he told me about the results, I thought he had made a mistake," says William Martin, director of the U-M's Laboratory for Scientific Computation, where Emeagwali spends about 13 hours daily, including weekends.
The prevailing skepticism forced Emeagwali to enter the contest. "I wanted to see if the judges agreed with me," he says in a soft voice. "Now, all of a sudden, the award gives me credibility --- scientific credibility."
As a result, some one-time doubters are changing their mind about the Connection Machine and wonder whether it should be used more widely.
"Accomplishments like Philip's are going to make other researcher's realize Connection Machines are not a gimmick and not a toy and are useful in real-life problems," says Rick Kufrin, an applications programmer at the National Center for Supercomputing Applications at the University of Illinois in Urbana.
Emeagwali, who will receive a doctoral degree in civil engineering and scientific computing this June, always has been a math whiz.
When growing up in Onitsha, Nigeria, he was tutored by his father, James, who worked as a nurse.
The elder Emeagwali loved math because one of his high school roommates, Chike Obi, became the nation's premier mathematician, receiving his doctoral degree from Cambridge University in England and soon becoming a household name.
"My father admired him and talked very fondly of him," the younger Emeagwali recalls. "He was sort of like our Einstein."
Until the sixth grade, the elder Emeagwali taught his son mathematics in the evenings. "He gave up because he said I knew more than he did," the younger Emeagwali recalls.
During junior high school, his classmates considered him a math genius, says Ozoh, the aerospace engineer from Kansas.
He helped his friends solve problems and soon picked up the nickname, "Calculus," although none of the students knew anything about the college-level math. Emeagwali, meanwhile, had mastered the subject by age 14, with the help of library textbooks.
"They considered me a genius, but that's a relative term," he now says.
After finishing eighth grade, Emeagwali, then 14 and the oldest of nine children, dropped out because his father no longer could afford to send all of them to school. He continued studying on his own, however, and passed a college entrance exam, deciding to attend college in the United States. When he was 17, he landed a scholarship at Oregon State University.
After receiving an undergraduate degree in mathematics --- usually working two jobs along the way --- Emeagwali earned a master's degree degree in civil engineering at George Washington University. There, he studied how to design dams because water supply problems were plentiful in the Third World, where he was thinking about working.
"I like to work on problems that are important to society because you get satisfaction," he says. "Research is hard work, so you might as well work on important research."
Before coming to the U-M in 1987, Emeagwali chalked up two other master's degrees --- one in mathematics and another in ocean, coastal and marine engineering --- and then worked two years in Casper, Wyo., as a civil engineer. Emeagwali always wanted to get a doctoral degree.
"I was taught by my father that it was impressive to have. But my father taught it was so impressive that I wouldn't get one," he says. "So when I told him I would get one, he told me to shut up because he thought I was bragging too much."
At U-M, Emeagwali had one goal: to increase computer speed so it would be possible to determine how fluid --- water or oil --- flows underground.
During the next few years, Emeagwali, who never has had an official academic adviser, floated among several departments, picking up tidbits of knowledge along the way.
"To be real honest, I thought he was getting caught between the cracks," says Trevor Mudge, director of the U-M's advanced computer architecture laboratory.
Soon, Emeagwali was trying several types of computers and decided to give the Connection Machine a shot.
"I'm more confident in pursuing ideas that are not very well accepted," he says. "I'm one of the first persons crazy enough to try it."
At the time, the Connection Machine, which was developed in 1986, had a poor reputation. Typically, researchers needed a year to learn how to use it, and its software was scarce.
Says Emeagwali: "People were skeptical and underestimated what the machine could do. I didn't want to be discouraged so I decided to work alone," using telephone lines to tie into four different machines around the country.
The Connection Machine operates completely differently than conventional supercomputers do. Rather than sending reams of data, such as thousands of number pairs, through a few high-power computer units, or processors, to be added one pair at a time, the Connection Machine assigns each pair to more than 65,000 less advanced processors --- each comparable to a desktop computer --- and the program instructs the calculations to occur simultaneously.
Just picture the conventional supercomputer as eight oxen pulling a cart and the Connection Machine as about 65,000 chicken pulling the same cart.
"The old thinking is that the oxen will do a better job, but if the chicken coordinate their efforts, then they'll do a better job," Emeagwali says.
For his doctoral dissertation, Emeagwali wanted to simulate an oil field, which isn't just a huge, underground cave. Instead, oil is found in pores within rocks, and oil companies must pump gas or water into fields to force the oil to nearby wells.
It then gets trickier. If oil is sucked out too quickly at one well, then oil elsewhere may not flow naturally to the same well and is virtually trapped --- that is, until another well is drilled, at considerable expense.
That's why it's important to understand flow within oil fields. In addition, the models help determine how many barrels are removable.
For his computer program, Emeagwali modified mathematical equations, first derived in 1938, and then divided the oil field into 8 million points, assigning 128 points to each of the Connection Machine's 65,536 processors --- figuring that more points provide better results.
Then, the computer program instructed each each point to talk with six neighbors simultaneously, which was a key in Emeagwali's success, and determined the oil's amount, direction of flow and speed at each point. For the entire oil field --- all eight million points --- the calculations took one-sixth of a second, which mimics a few hours of actual oil flow.
Other computer programs for simulating oil fields take longer, Emeagwali says, and they're not as accurate because the field points aren't distributed equally and are fewer in number.
Emeagwali's accomplishments are important, says McDonald of Mobil Research and Development Corp. During the time saved, "we could do more engineering studies on other oil fields, which saves both time and money," he says.
While the Connection Machine performed 3.1 billion calculations each second in Emeagwali's problem, one of the Cray supercomputers has produced only 1 billion calculations with similar problems.
By the mid-1990's, Emeagwali, who wants to pursue an academic career, expects to achieve 1 trillion calculations each second.
Emeagwali works hard --- typically 13 hours a day --- but takes time out for his favorite sport, tennis. He's an avid player, and last year, he was runner-up in the finals of Ann Arbor's Men's B-class tennis tournament, beginning as an unseeded player.
Top computing prize goes to a University of Michigan Ph.D. candidate
PHILIP EMEAGWALI, who took on an enormously difficult problem and, like most students working on Ph.D. dissertations, solved it alone, has won computation's top prize, captured in the past only by seasoned research teams. Mr. Emeagwali, a Ph.D. candidate in civil engineering and scientific computing at the University of Michigan, was awarded the $1,000 Gordon Bell Prize from the Institute for Electrical and Electronics Engineers for speeding up a supercomputer to solve an important problem.  Philip Emeagwali and his Gordon Bell Prize plaque His program for the Connection Machine, a massively parallel computer, made it run at 3.1 billion calculations per second --- twice as fast as the speedup achieved by last year's winners, and 24 times as fast as the speedup achieved by the winners from the year before. Mr. Emeagwali, who was the first to win using a Connection Machine, solved a serious problem in the petroleum industry. His program simulates the flow of oil in a petroleum reservoir and enables engineers to determine where best to place wells to capture as much of the trapped oil as possible. Today's technology enables recovery of about 30 per cent of that oil, Mr. Emeagwali says. If his program can squeeze out a few more percentage points, it will help decrease U.S. reliance on foreign oil.
Mr. Emeagwali, 35, was a mathematical whiz as a child in Nigeria, and since coming to the United States has earned three master's degrees. Although the Connection Machine, with more than 65,000 processors, is considered particularly difficult to use, Mr. wrote his program after working on the computer for only eight months.
Mr. Emeagwali hopes to continue his reseach and remain in academe after he gets his Ph.D. this summer.
Grad Student Wins Top Supercomputer Award
U-M grad student Philip Emeagwali has won the 1989 Gordon Bell Prize Competition, which is considered the annual high point of supercomputer research. With his petroleum reservoir simulation, Emeagwali became the first sole investigator to win the prize.
PHILIP EMEAGWALI, a Ph.D. candidate in U-M's Department of Civil Engineering and Program in Scientific Computing, has won first place in the 1989 Gordon Bell Prize Competition, known as the annual "supercomputer olympics." Emeagwali received the award on February 28, 1990, at the Institute for Electrical and Electronics Engineers (IEEE) CompCon conference in San Francisco.
The $1000 prize recognizes outstanding achievement in the use of supercomputers to solve significant scientific and engineering problems. The IEEE competition is considered the annual high point of supercomputer research. Each year, the winning works makes headlines in several publications, and the winning supercomputer is declared the "fastest supercomputer on earth."
Emeagwali conducted his award-winning work in petroleum reservoir simulation on Connection Machine supercomputers at the Los Alamos National Laboratory, the National Center for Supercomputer Applications at the University of Illinois, and the Thinking Machines Corporation. He accessed the supercomputers over the Internet from local workstations.
Philip Emeagwali at the award ceremony All previous IEEE Gordon Bell Prize Competition winners have represented collaborative efforts, involving researchers from industry, academia, and research laboratories. Previous participants have included multidisciplinary teams from Mobile Oil, Cray Research, IBM, the California Institute of Technology, MIT, the NASA Ames Research Center, the National Center for Atmospheric Research, AT&T Bell Labs, and other institutions. This is the first time the prize has been given to a sole investigator and the first time it has been given to an individual from a university.
Emeagwali's primary research interests are parallel computation and large- scale problems in computational fluid dynamics. His award-winning research focuses on underground petroleum recovery through high-speed supercomputer simulations.
Background of the Study
In 1988, the cost of imported oil accounted for 29 percent of the United States trade deficit. This fact, coupled with the drive toward less dependence on imported oil, makes it important to maximize the amount of oil recovered from petroleum production wells. Currently, engineers can only recover about 30 percent of the oil in a petroleum reservoir. Engineers can recover more oil by using petroleum reservoir simulation models to manage a group of oil wells economically.
Because supercomputers are often used to solve the equations used in petroleum reservoir simulations, it is not surprising that 10 percent of supercomputers in existence have been purchased by the petroleum industry to conduct simulations aimed towards increasing the total amount of recoverable oil. Given the huge economic benefits to be derived and the fact that more powerful supercomputers are needed for accurate reservoir simulation, such simulations have been designated by the U.S. government as one of the 20 national Grand Challenges in science and engineering.
Connection Machine Supercomputer
The Connection Machine, one of the fastest supercomputers ever built, consists of a collection of more than 65,000 separate processors cooperating simultaneously to solve single, complex problems. The Connection Machine is ideal for applications that require the simultaneous performance of thousands and even millions of simple arithmetical operations. For such computation-intensive applications, the processing power of the Connection Machine actually increases as the amount of data increases.
Challenges
The Connection Machine simulation of petroleum reservoirs poses several mathematical and programming challenges:
Currently, few algorithms are suitable for the architecture of supercomputers like the Connection Machine. This means that only a few complex, real-life problems can be solved on such machines. Many other problems would be potentially solvable if appropriate algorithms could be developed. To use the newer supercomputers effectively, researchers must rewrite and reformulate many computation-intensive applications. New architectures will stimulate the development of new problem-solving approaches, new governing equations (or descriptions) for important problems, and new numerical algorithms.
Emeagwali designed a suitable algorithm for petroleum reservoir simulation by modifying a set of governing equations developed in 1938 by the Russian mathematician B. K. Risenkampf. Although these equations were abandoned for various historical and computational reasons and have never been used in any serious applications, Emeagwali argues that they are suitable for the newer supercomputer architectures such as that of the Connection Machine. More importantly, his approach is also applicable to a wide range of important scientific and engineering problems, including the problem of calculating the movement of buried nuclear wastes.
Another challenge associated with the use of the Connection Machine is the time spent in inter-processor communication, which has made it extremely difficult to obtain very high performance. Emeagwali was able to reduce inter-processor communication time drastically by creating and using more than 8 million virtual processors instead of the original 65,000 processors.
ResultsUsing this new approach in combination with the Connection Machine, Emeagwali's model ran at the exceptionally speed of 3.1 billion calculations per second --- twice the speed of the 1988 Gordon Bell Prize winning entry and 24 times faster than the 1987 winner. The speed of Emeagwali's model even exceeds the theoretical peak calculation speed of much more expensive conventional supercomputers, including the widely used $30 million Cray Y-MP.
Running at such high speeds, petroleum reservoir simulation problems that formerly took several hours to solve on conventional supercomputers can now be solved in only a few seconds. In fact, the implementation of Emeagwali's new approach on the Connection Machine supercomputer took the petroleum reservoir simulation problem from the "I wish I could ..." stage to the "I can see how to do it" stage.
Research BackgroundEmeagwali will soon receive his Ph.D. degree in Civil Engineering and Scientific Computation from U-M. He was a civil engineer for the U.S. government before he accepted a doctoral fellowship from the University of Michigan in 1987. Emeagwali received two M.S. degrees in Engineering from George Washington University, an M.A. in Mathematics from the University of Maryland at College Park, A B.S. in mathematics from Oregon State University, and a General Certificate of Education from the University of London.Please like subscribe comment and share this post. Also, check out our facebook group: If you would like to start a discussion join the group and IM me.
|
Comments
Post a Comment