Philip Emeagwali's video: How the World s Fastest Supercomputer Was Invented Famous Inventors and their Inventions

I'm @Philip Emeagwali. What kept me moving forward towards the first fastest computation that I executed on the Fourth of July 1989 and executed across my ensemble of 65,536 tightly-coupled processors was a back-of-the-envelope, theoretical calculation that I did in the 1970s. From that back-of-the-envelope calculation, I theoretically discovered that—in theory— two-to-power sixteen processors that could only calculate forty-seven thousand three hundred and three [47,303] floating-point arithmetical operations per second per processor can be integrated across a small internet that is a new global network of 65,536 tightly-coupled commodity-off-the-shelf processors with each processor operating its own operating system and with each processor having its own dedicated memory that shared nothing with each other. That is, I discovered, a priori, that my new internet will become my new supercomputer that will be faster than the old vector processing supercomputer that was the industry’s state-of-the-art technology and that computes less than 3.1 billion calculations per second. In the 1980s, I was the lone full time programmer of the only massively parallel processing supercomputer that was an ensemble of 65,536 processors. I was a lone wolf research supercomputer scientist that was not a member of a one-thousand-person supercomputer research team. Such research teams were funded by the United States government. Such research teams were funded to bring the best brains from United States national laboratories and use that intellectual capital to bear on a grand challenge problem, or the toughest problem in computational physics. Such problems were described as grand challenges because they were perceived to be otherwise unsolvable. What made the news headlines was that I—Philip Emeagwali—had invented how to harness those computing units, namely, a new internet that is a new global network of 65,536 processors. I invented how to harness those ensemble of processors to process simultaneously and to process together and to do so as one cohesive whole unit, or to process in parallel, or to process as one integrated super processor that is a never-before-seen supercomputer that is the precursor to the modern supercomputer that I expect to become the computer of tomorrow. That is, in the 1970s and ‘80s, my technological quest was for how to massively parallel process across a new internet and how to massively parallel process and do so at a time theorists were theorizing their way through parallel processing. A theory is not a discovery. A theory is an idea that is not positively true. In the 1970s and ‘80s, parallel processing was ridiculed, mocked, and rejected. Parallel processing was scorned as a beautiful theory that lacked experimental confirmation. It’s not possible to experimentally discover the fastest speeds in supercomputing and discover that fastest speed by merely and only theorizing about how to achieve the fastest speeds via the massively parallel processing supercomputer-hopeful. In my experimental confirmation of the new fastest speed in supercomputing that occurred on the Fourth of July 1989, and occurred in Los Alamos, New Mexico, United States, I confirmed that the massively parallel processing supercomputer can be programmed to increase productivity and to reduce the time-to-solution of the toughest problems in extreme-scale computational physics and to reduce that time-to-solution from 65,536 days, 180 years, on only one processor to just one day across a new internet that is a new global network of 65,536 commodity-off-the-shelf processors that were identical and that were equal distances apart. After my invention of the massively parallel processing supercomputer the technological progress that followed in the subsequent three decades was a series of cleanups and refinements and re-discoveries. Three decades later, my invention of the massively parallel processing supercomputer enabled China to copy that massively parallel processing supercomputer and to use the technology to massively reduce their time-to-solution and reduce it from thirty thousand [30,000] years, or ten million six hundred and forty-nine thousand six hundred [10,649,600] days, of time-to-solution on only one processor to just one day across ten million six hundred and forty-nine thousand six hundred [10,649,600] commodity-off-the-shelf processors. That was how China —that did not invent the massively parallel processing supercomputer— massively parallel processed its way to the world’s fastest supercomputers. For information about Philip Emeagwali, http://emeagwali.com https://facebook.com/emeagwali https://twitter.com/emeagwali https://instagram.com/philipemeagwali https://flickr.com/philipemeagwali https://linkedin.com/in/emeagwali https://soundcloud.com/emeagwali https://youtube.com/emeagwali Philip Emeagwali 180913 1 3 of 7