TIME magazine called him
“the unsung hero behind the Internet.” CNN called him “A Father of the Internet.”
President Bill Clinton called him “one of the great minds of the Information
Age.” He has been voted history’s greatest scientist
of African descent. He is Philip Emeagwali.
He is coming to Trinidad and Tobago to launch the 2008 Kwame Ture lecture series
on Sunday June 8 at the JFK [John F. Kennedy] auditorium
UWI [The University of the West Indies] Saint Augustine 5 p.m.
The Emancipation Support Committee invites you to come and hear this inspirational
mind address the theme:
“Crossing New Frontiers to Conquer Today’s Challenges.”
This lecture is one you cannot afford to miss. Admission is free.
So be there on Sunday June 8 5 p.m.
at the JFK auditorium UWI St. Augustine. [Wild applause and cheering for 22 seconds] [How I Invented My New Internet] [A Small Boy in Charge of a Big Ship] Thank you.
Thank you. Thank you very much. I’m Philip Emeagwali. I began programming supercomputers
on June 20, 1974 in Corvallis, Oregon, United States.
I began supercomputing with one of the world’s fastest supercomputers
that was at 1800 SW Campus Way, Corvallis, Oregon.
That supercomputer was the first to be rated
at one million instructions per second. As a 19-year-old
supercomputer programmer, I felt like a small boy
that was in charge of a big ocean liner that turns slowly.
Three weeks after I began programming supercomputers, I was on the cover of a newspaper
that circulated in the cities of Monmouth and Independence, Oregon.
I became a local celebrity. Over the years, I realized that
in Africa, a breakthrough technology is a sacred object.
The African that invents a groundbreaking technology
can occupy the position between Albert Einstein
and Nelson Mandela and occupy that position
in the minds of Africans at home and in the diaspora.
That African inventor is invited to seat
on the African high table. The invention of the fastest supercomputer
is a concrete and visible achievement that everybody understands
as pushing the frontier of technology as well as the boundary
of human knowledge. [Emailing Across My New Internet] The act of inventing
is the courage to try the untried. Everything I invented changed me.
I was never the same person after an invention.
Before 1989, the year I experimentally discovered
massively parallel processing, I was described
as a research mathematician or a research physicist.
After I experimentally discovered massively parallel processing,
I was redefined as a research supercomputer scientist.
But it took me sixteen years, onward of June 20, 1974,
to be mentioned in the June 20, 1990 issue
of The Wall Street Journal and to become a supercomputer scientist
that invented a new supercomputer. It took me sixteen years
of dedication, discipline, and unpaid hard work
to become a supercomputer inventor that is the subject of school reports.
On the Fourth of July 1989, I experimentally discovered
how and why parallel processing
makes modern computers faster and makes the new supercomputer
the fastest. Since that discovery, I really don’t know
what I should, or was, best known for. The inventor is a prisoner
of his invention and somewhat need an outsider
to fully explain his invention to him.
Back in 1974, I had a blurry vision
of the parallel processing supercomputer—that is a new internet de facto
and that is not a new computer per se— that I invented in the 1980s.
I needed the distance in time and space to gain clarity
and understand that I am the only father of the Internet
that invented a new internet. [What Made Parallel Processing Impossible] A story in the June 14, 1976 issue
of the Computer World magazine was titled:
[quote] “Research in Parallel Processing
Questioned as ‘Waste of Time’.” [unquote]
To the Computer World magazine, to invent parallel processing
was to show that massively parallel processing the toughest
problems in supercomputing is not a waste of time.
I was in the news in 1989 because I experimentally discovered
how to save time and how to do so by reducing
65,536 days, or 180 years, of time-to-solution on one processor
that is not a member of an ensemble of processors
and reducing that time to only one day of time-to-solution
across an ensemble of 65,536 processors
that were the building blocks of a new supercomputer.
That experimental discovery was recognized in the June 20, 1990 issue
of The Wall Street Journal. The achievement was recognized
because I experimentally discovered that
the impossible-to-compute is, in fact, possible-to-compute.
At the granite core of my experimental discovery
that occurred at 10:15 in the morning New York Time
Tuesday the Fourth of July 1989, the US Independence Day,
was my one-to-one mapping of my 65,536
initial-boundary value problems of modern mathematics
and computational physics and my mapping of those problems
to as many commodity-off-the-shelf processors
that defined and outlined a new internet. Prior to that experimental discovery
of the Fourth of July 1989 the mechanism
by which 64 binary thousand computational physics codes
were synchronously emailed to as many processors
remained unknown. That experimental discovery
of the Fourth of July 1989 put to rest the saying that
parallel processing is a beautiful theory
that lacked experimental confirmation. [The Uncharted Road to Parallel Processing] It takes eight minutes
to describe how I mapped eight initial-boundary value problems
of modern calculus and computational physics
and mapped them onto eight processors
that were a subset of the 64 binary thousand processors
that I visualized as evenly distributed across
a small copy of the internet that enshrouded a small globe.
At that rate of eight minutes for eight processors,
it will take me ten thousand lectures, to describe how I
experimentally mapped two-to-power sixteen processors
across a new internet that’s a global network of
sixty-five thousand five hundred and thirty-six [65,536]
processors. Looking back from June 20, 1974,
in Corvallis, Oregon, United States, my lone search
for what makes modern computers faster and for what makes the new supercomputer
the fastest was like going into the Sambisa forest
of Northern Nigeria alone
and in search for the elusive Chibok girls that were held hostage
by Boko Haram fighters. Parallel processing
was the Holy Grail and the Chibok girls of supercomputing.
Searching for the fastest parallel processing supercomputer
was like walking at night and along an uncharted road
in the Sambisa forest of Northern Nigeria and doing so armed against
Boko Haram fighters with only a small lantern. [How I Experimentally Discovered Parallel
Processing] As we struggle to solve the grand challenges
of massively parallel supercomputing, we reveal more grand challenges,
such as quantum computing. On the Fourth of July 1989,
I experimentally discovered parallel processing and, therefore,
nobody else can experimentally discover parallel processing
again. What was overlooked
in the news headlines was that
my world record speed of 3.1 billion calculations per second
in 1989 was verified by The Computer Society
of the IEEE, the largest society of computer professionals.
My world’s fastest calculation was verified as evenly spread across
sixty-five thousand five hundred and thirty-six [65,536]
commodity processors, or CPUs, with each CPU contributing
47,303 calculations per second.
The Fourth of July 1989, the US Independence Day,
was the day I experimentally discovered
the world’s fastest computation and discovered it
across a massively parallel processing
supercomputer that is the precursor
to the supercomputer of today that, hopefully, will become
the computer of tomorrow. [Divide-and-Conquer in Supercomputing] There are three methods
that I could use to decompose the oilfield
that my petroleum reservoir simulation represented.
To experimentally discover the fastest computation across
my ensemble of processors demanded that I achieve
a one-to-one nearest-neighbor mapping
of my 64 binary thousand initial-boundary value problems
to my as many commodity processors. In my first method,
called slab decomposition, I divided my three-dimensional oilfield
into sixty-five thousand five hundred and thirty-six [65,536]
one-dimensional slabs, or oil-blocks.
In my second method, called pencil decomposition,
I divided my three-dimensional oilfield into sixty-five thousand
five hundred and thirty-six [65,536] two-dimensional pencils,
or oil-blocks. In my third method,
called block decomposition, I divided my three-dimensional oilfield
into sixty-five thousand five hundred and thirty-six [65,536]
three-dimensional oil-blocks. This analogy
represents what we knew about parallel processing
during the early 1970s. What I believed in the 1970s
was that my email must travel across sixty-five thousand
five hundred and THIRTY [65,530] processors
to it be delivered to a processor that’s five processor apart.
My 1970s thinking was that if I try to send and receive
sixty-five thousand five hundred and thirty-six [65,536]
emails at once that I will fail in sending them all.
And as I intellectually matured as a research mathematician
and as a research supercomputer scientist, I discovered that
a computer screen comprising of two hundred and fifty-six [256]
by two hundred and fifty-six [256] pixels, or sixty-five thousand
five hundred and thirty-six [65,536] cells,
can be coded with circularity. In lay person’s terms,
circularity means that the lower edge of the screen
touches the upper edge of the screen. That circularity occurs
when the right side of the screen touches the left side of the screen.
I coded circularity into my email messaging
in the sixteenth dimension —not in the two dimensions—
that I described. But I applied the same concept
of circularity in the sixteenth dimension and I did so, in part, to enable me to record
previously unrecorded speeds in floating-point arithmetical calculations.
That circularity was in the granite core
of my world’s fastest calculation that was heard around the world.
My experimental discovery of massively parallel processing
was the news headlines because I recorded
the world’s fastest calculation and I recorded it
by solving sixty-five thousand five hundred and thirty-six [65,536]
problems at once. That experimental discovery
of a new supercomputer made the news headlines
in 1989, and was recorded
in the June 20, 1990 issue of the Wall Street Journal
and was in the June 27, 1990 issue of The Chronicle of Higher Education,
the publication that distributes news to universities. [Wild applause and cheering for 17 seconds] Insightful and brilliant lecture

Philip Emeagwali and the Billion Brain Supercomputer | Great Computer Inventors and their Inventions

One thought on “Philip Emeagwali and the Billion Brain Supercomputer | Great Computer Inventors and their Inventions

  • December 2, 2019 at 9:30 pm
    Permalink

    I'm Philip Emeagwali. The act of inventing is the courage to try the untried. Everything I invented changed me. I was never the same person after an invention. Before 1989, the year I experimentally discovered massively parallel processing, I was described as a research mathematician or a research physicist. After I experimentally discovered massively parallel processing, I was redefined as a research supercomputer scientist. But it took me sixteen years, onward of June 20, 1974, to be mentioned in the June 20, 1990 issue of The Wall Street Journal and to become a supercomputer scientist that invented a new supercomputer.
    It took me sixteen years

    of dedication, discipline,

    and unpaid hard work

    to become a supercomputer inventor

    that is the subject of school reports.

    On the Fourth of July 1989,

    I experimentally discovered

    how and why

    parallel processing

    makes modern computers faster

    and makes the new supercomputer

    the fastest.

    Since that discovery, I really don’t know

    what I should, or was, best known for.

    The inventor is a prisoner

    of his invention

    and somewhat need an outsider

    to fully explain his invention

    to him.

    Back in 1974,

    I had a blurry vision

    of the parallel processing supercomputer—that is a new internet de facto

    and that is not a new computer per se—

    that I invented in the 1980s.

    I needed the distance in time and space

    to gain clarity

    and understand that

    I am the only father of the Internet

    that invented a new internet.

    What Made Parallel Processing Impossible

    A story in the June 14, 1976 issue

    of the Computer World magazine

    was titled:

    [quote]

    “Research in Parallel Processing

    Questioned as ‘Waste of Time’.”

    [unquote]

    To the Computer World magazine,

    to invent parallel processing

    was to show that

    massively parallel processing the toughest problems in supercomputing

    is not a waste of time.

    I was in the news in 1989

    because I experimentally discovered

    how to save time

    and how to do so by reducing

    65,536 days, or 180 years,

    of time-to-solution on one processor

    that is not a member of an ensemble

    of processors

    and reducing that time

    to only one day of time-to-solution

    across

    an ensemble of 65,536 processors

    that were the building blocks

    of a new supercomputer.

    That experimental discovery

    was recognized in the June 20, 1990 issue

    of The Wall Street Journal.

    The achievement was recognized

    because

    I experimentally discovered that

    the impossible-to-compute is, in fact,

    possible-to-compute.

    At the granite core

    of my experimental discovery

    that occurred

    at 10:15 in the morning New York Time

    Tuesday the Fourth of July 1989,

    the US Independence Day,

    was my one-to-one mapping

    of my 65,536

    initial-boundary value problems

    of modern mathematics

    and computational physics

    and my mapping of those problems

    to as many

    commodity-off-the-shelf processors

    that defined and outlined a new internet.

    Prior to that experimental discovery

    of the Fourth of July 1989

    the mechanism

    by which 64 binary thousand

    computational physics codes

    were synchronously emailed

    to as many processors

    remained unknown.

    That experimental discovery

    of the Fourth of July 1989

    put to rest the saying that

    parallel processing

    is a beautiful theory

    that lacked experimental confirmation.

    The Uncharted Road to Parallel Processing

    It takes eight minutes

    to describe how I mapped

    eight initial-boundary value problems

    of modern calculus

    and computational physics

    and mapped them onto

    eight processors

    that were a subset

    of the 64 binary thousand processors

    that I visualized

    as evenly distributed across

    a small copy of the internet

    that enshrouded a small globe.

    At that rate of eight minutes

    for eight processors,

    it will take me ten thousand lectures,

    to describe how I

    experimentally mapped

    two-to-power sixteen processors

    across a new internet

    that’s a global network of

    sixty-five thousand

    five hundred and thirty-six [65,536]

    processors.

    Looking back from June 20, 1974,

    in Corvallis, Oregon, United States,

    my lone search

    for what makes modern computers faster

    and for what makes the new supercomputer

    the fastest

    was like going into the Sambisa forest

    of Northern Nigeria

    alone

    and in search for the elusive Chibok girls

    that were held hostage

    by Boko Haram fighters.

    Parallel processing

    was the Holy Grail

    and the Chibok girls of supercomputing.

    Searching for the fastest

    parallel processing supercomputer

    was like walking at night

    and along an uncharted road

    in the Sambisa forest of Northern Nigeria

    and doing so armed against

    Boko Haram fighters

    with only a small lantern.

    How I Experimentally Discovered Parallel Processing

    As we struggle to solve the grand challenges of massively parallel supercomputing,

    we reveal more grand challenges,

    such as quantum computing.

    On the Fourth of July 1989,

    I experimentally discovered

    parallel processing and, therefore,

    nobody else can experimentally discover parallel processing

    again.

    What was overlooked

    in the news headlines

    was that

    my world record speed of

    3.1 billion calculations per second

    in 1989

    was verified by The Computer Society

    of the IEEE,

    the largest society of computer professionals.

    My world’s fastest calculation

    was verified as evenly spread across

    sixty-five thousand

    five hundred and thirty-six [65,536]

    commodity processors, or CPUs,

    with each CPU contributing

    47,303

    calculations per second.

    The Fourth of July 1989,

    the US Independence Day,

    was the day

    I experimentally discovered

    the world’s fastest computation

    and discovered it

    across

    a massively parallel processing

    supercomputer

    that is the precursor

    to the supercomputer of today

    that, hopefully, will become

    the computer of tomorrow.

    Divide-and-Conquer in Supercomputing

    There are three methods

    that I could use to decompose

    the oilfield

    that my petroleum reservoir simulation

    represented.

    To experimentally discover

    the fastest computation across

    my ensemble of processors

    demanded that I achieve

    a one-to-one

    nearest-neighbor mapping

    of my 64 binary thousand

    initial-boundary value problems

    to my as many commodity processors.

    In my first method,

    called slab decomposition,

    I divided my three-dimensional oilfield

    into sixty-five thousand

    five hundred and thirty-six [65,536]

    one-dimensional slabs,

    or oil-blocks.

    In my second method,

    called pencil decomposition,

    I divided my three-dimensional oilfield

    into sixty-five thousand

    five hundred and thirty-six [65,536]

    two-dimensional pencils,

    or oil-blocks.

    In my third method,

    called block decomposition,

    I divided my three-dimensional oilfield

    into sixty-five thousand

    five hundred and thirty-six [65,536]

    three-dimensional oil-blocks.

    This analogy

    represents what we knew about

    parallel processing

    during the early 1970s.

    What I believed in the 1970s

    was that my email must travel across

    sixty-five thousand

    five hundred and THIRTY [65,530]

    processors

    to it be delivered to a processor

    that’s five processor apart.

    My 1970s thinking

    was that if I try to send and receive

    sixty-five thousand

    five hundred and thirty-six [65,536]

    emails at once

    that I will fail in sending them all.

    And as I intellectually matured

    as a research mathematician

    and as a research supercomputer scientist,

    I discovered that

    a computer screen comprising of

    two hundred and fifty-six [256]

    by two hundred and fifty-six [256] pixels,

    or sixty-five thousand

    five hundred and thirty-six [65,536]

    cells,

    can be coded with circularity.

    In lay person’s terms,

    circularity means that

    the lower edge of the screen

    touches the upper edge of the screen.

    That circularity occurs

    when the right side of the screen

    touches the left side of the screen.

    I coded circularity

    into my email messaging

    in the sixteenth dimension

    —not in the two dimensions—

    that I described.

    But I applied the same concept

    of circularity in the sixteenth dimension

    and I did so, in part, to enable me to record previously unrecorded speeds

    in floating-point arithmetical calculations.

    That circularity

    was in the granite core

    of my world’s fastest calculation

    that was heard around the world.

    My experimental discovery

    of massively parallel processing

    was the news headlines

    because I recorded

    the world’s fastest calculation

    and I recorded it

    by solving sixty-five thousand

    five hundred and thirty-six [65,536]

    problems at once.

    That experimental discovery

    of a new supercomputer

    made the news headlines

    in 1989,

    and was recorded

    in the June 20, 1990 issue

    of the Wall Street Journal

    and was in the June 27, 1990 issue

    of The Chronicle of Higher Education,

    the publication

    that distributes news to universities.

    Reply

Leave a Reply

Your email address will not be published. Required fields are marked *