Philip Emeagwali | How I Invented a New Internet | Famous Black Inventors and their Inventions


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 a New Internet that is a New
Supercomputer] [What is Philip Emeagwali Famous For in Computing?] [WHAT IS PHILIP EMEAGWALI KNOWN FOR?] I’m Philip Emeagwali. The fundamental problem
of supercomputing was to discover how to solve
the toughest problems arising in mathematics, science,
and engineering. And to discover how to solve
those grand challenge problems across an ensemble of processors
that were identical to each other and that shared nothing
between each other with each processor operating
its own operating system. The latter was the biggest
scientific question in the unknown world
of the supercomputer. The concrete, measurable,
and visible proof that I was in the terra incognita,
or in the unexplored territory, of the supercomputer
was that it made the news headlines that I experimentally
parallel processed and communicated across a new internet. After my invention
of practical parallel processing, I became well known
but not known well. That is, many knew Philip Emeagwali
as an inventor but few understood his invention. It’s easier to recognize my face
than to understand my abstract contributions
to mathematics, physics, and computer science. [Who is Philip Emeagwali?] I am the computational mathematician
that contributed to a greater understanding
of how to execute the fastest floating-point calculations of arithmetic. I am the research mathematician
who figured out how to solve the largest system of equations
of algebra that must be solved
to discover and recover otherwise elusive
crude oil and natural gas. I am the mathematician
that invented new partial differential equations
of the calculus of extreme-scaled petroleum reservoir simulation. For those reasons, I said that
I am well known as a supercomputer scientist
that contributed to the development of the computer
but I am not known well as a mathematician
that contributed to mathematics. It’s easier to understand that
I contributed to the modern computer or to the modern supercomputer
that’s an internet than to understand my contributions
to computational mathematics and even computational physics. Most people think calculus
is difficult to understand. The invention
of the fastest computer is easier to recall
than the invention of the most advanced expression
in calculus that, in turn, is the recurring decimal
in nearly all the workloads of supercomputers. [School Reports on Philip Emeagwali] A 12-year-old writing
a school inventor report on Philip Emeagwali
cannot explain to her teacher how the new nine
partial differential equations that I contributed to calculus
is more accurate than the previous equations
in textbooks. On the other hand,
she could explain my contributions to the development of the supercomputer
that is a new internet. The technology called
practical parallel processing that I discovered
on the Fourth of July 1989 was called a grand challenge
for a good reason. Because it was a once-impossible problem
that was in the realm of science-fiction the machinery was abandoned
by 25,000 supercomputer scientists that were only at home
with scalar and/or vector processing. I was the only full time programmer
of the 1980s that was at the frontier
of the most massively parallel supercomputers. In the 1980s, attempting to harness
64 binary thousand processors and to use them to solve
the biggest scientific challenges evoked a sense of foreboding. In the 1980s, harnessing
one billion processors—that defined and outlined
a massively parallel supercomputer —and using them to solve
a grand challenge problem was as science fiction
as sending an astronaut to planet Mars. [WHY I PARALLEL PROCESSED ALONE] In the 1980s, to parallel process
a grand challenge problem was to make the impossible-to-solve
initial-boundary value problem of calculus and physics
possible-to-solve as a discretized problem
in large-scale algebra. The reason I parallel processed alone
was that I was the only person with the confidence to do so. In the 1970s and ‘80s,
practical parallel supercomputing across a new internet
that was a new global network of 65,536 processors
was like shooting at as many birds in the dark. I parallel processed
to discover speeds in computation and communication
that were previously unseen, and that made the news headlines
in 1989. Supercomputer scientists
that had seen me daily in the 1980s
first read about my discovery of practical parallel supercomputing
and read about it in newspapers, instead of hearing about my discovery from
me. For me as a lone
supercomputer scientist, breaking the speed records
in both computation and communication and breaking those records alone
and breaking those records for the first time
and breaking those records with a parallel processing machinery
was the metaphorical equivalence of being the first solo mountain climber
that climbed to the peak of Mount Everest. The significance of reaching the top
of Mount Everest and being the first person to reach it
was an achievement in geographical exploration
that redefined the boundary of the reachable regions of the Earth. I was in the news headlines because
I was the first lone wolf supercomputer scientist
to climb to the peak of the Mount Everest
of massively parallel supercomputing across a new ensemble of
65,536 tightly-coupled, commodity-off-the shelf processors
that shared nothing between each other and that were equal distances apart
from each other. [Inventing a New Internet] [Thirty Thousand Years in One Day] Prior to my experimental discovery
of practical parallel supercomputing and my discovery
of how to solve a grand challenge problem
and how to solve it across a new internet,
the fastest computations were recorded
on the scalar supercomputers of the late 1940s
through early 1970s. The fastest computations
were also recorded on the vector supercomputers
of the mid-1970s through late 1980s. I first entered
into the world of scalar supercomputing on June 20, 1974
at 1800 SW Campus Way Corvallis, Oregon, United States. That scalar supercomputer
solved only one initial-boundary value problem
of calculus at a time. The ensemble of 65,536 processors
that I programmed in the 1980s and programmed
as a new internet and that made the news headlines
in 1989 solved 65,536
initial-boundary value problems at once. Initial-boundary value problems
of calculus are at the foundation
of computational physics. Nine in ten supercomputer cycles
consumed in the 1980s were consumed by extreme-scale
computational physicists. Extreme-scale, high-resolution computational
physics is executed across
a massively parallel supercomputer that occupies the space
of a soccer field. For that reason, computational physics
is a branch of physics that lies between theoretical
and experimental physics. That is, computational physics
is the third branch of physics. That branch of physics is midway
between theory and experiment. That branch of physics encompassed
both theory and experiment. My experimental discovery
of how to solve many initial-boundary value problems
that are governed by a system of partial differential equations
of calculus and governed by its companion
and discretized system of partial difference equations
of algebra and my discovery
of how to solve them at once opened the door
to the parallel supercomputer that is the world’s fastest supercomputer
that achieves its record-breaking supercomputing speed
by solving millions upon millions of initial-boundary value problems
and solving them at once. In computational physics,
my experimental discovery made it possible
for the supercomputer of today to reduce the time-to-solution
of the biggest scientific challenges and reduce it from
10.65 million days, or 30,000 years, to just one day. Without parallel supercomputing,
a global warming prediction will occur 30,000 years after
the said global warming occurred. [Crossing the Frontier of Supercomputer Knowledge] My quest for the fastest speeds
in computing demanded that I parallel process across
a new internet that is a new global network
of 64 binary thousand processors. In the 1980s,
massively parallel processing defined the boundary
of the supercomputer. The reason I am well known
but not known well was that I was the first person
to enter into the unexplored territory where the fastest computations
can be executed across a new internet. The proof that I entered into
that unexplored territory was that I recorded speeds
in supercomputing that were previously unrecorded. That contribution
made more news headlines than any singular contribution
made by an individual to the development of the computer. In the 1970s and ‘80s,
the complete knowledge of the parallel supercomputer
was out of the reach of human beings. That is, I parallel processed
in that new frontier of knowledge and did so without a map, or a textbook. On the Fourth of July 1989,
I became the first person to provide practical, in-depth,
and easy to understand explanations of how to harness millions of processors
and how to use those processors to solve a real-world problem
that is chopped up into millions of smaller problems. My invention
of practical parallel supercomputing made the news headlines because
I also discovered how to harness the new supercomputer
to solve grand challenge problems that will be otherwise impossible
to solve. [New Internet Versus Old Computer] In the history of computing,
the invention of parallel supercomputing is the biggest change
in the way we think about the supercomputer. In the old way,
the fastest supercomputer solved only one problem at a time,
or in sequence. In my new way,
the fastest supercomputer solved ten million problems
at once, or in parallel. I was in the news because
I discovered how to experimentally perform
65,536 synchronized parallel communication
that was as many times faster than your email. The supercomputer that I programmed
in 1974 only computed sequentially
and did so within only one central processing unit. The virtual supercomputer
that I programmed in the 1980s computed in parallel
and did so in the plural senses and communicated across
a new internet that is a new global network of
64 binary thousand processors. [Philip Emeagwali: A Father of the Internet] [How I Invented a New Internet] Who invented the internet? The Internet
has many fathers and mothers as well as aunts and uncles. We can only have
one father of the Internet that invented a new internet. The father of the Internet
should at least invent a new internet. I am called a father of the Internet because
I am the only father of the Internet that invented a new internet. I invented my new internet
by, first, theorizing it back in 1974 and then continuously developed it
for the subsequent fifteen years and developed
that small copy of the internet and did so until I actualized it
as the fastest computation back on the Fourth of July 1989. My two-raised-to-power sixteen commodity-off-the-shelf
processors were tightly-coupled to each other
and were equal distances apart from each other. I mathematically visualized
my 64 binary thousand processors as tightly-encircling a hyper globe
that is bounded by the hypersurface
of a sixteen-dimensional hypersphere that is embedded
within a sixteen-dimensional hyperspace. I visualized
the physical and mathematical domains of my extreme-scale, high-resolution
general circulation model as the 62-mile deep
hyper-spherical shell that was bounded by two hyperspheres. The inner hypersphere
has a diameter of 7,900 miles that corresponded to
the surface of the Earth. The outer hypersphere
has a diameter of 7,962 miles that corresponded
to the outer boundary of the atmosphere of the Earth. I visualized
the two-raised-to-power sixteen vertices of my hypercube
to be midway (or 31 miles) between those two hyperspheres. I drew parallels
between my new internet that was a new global network
of processors and how I envisioned
simulating global warming. My two hyperspheres
were parallel to each other. My two hyperspheres
extended in the same direction. My two hyperspheres
never converged or diverged. My 65,536 processors
were paralleled with respect to the climate model
that I divided into 65,536 smaller climate models. Those climate models
were identical in domain size. [Paradigm Shift in Computing] My discovery
of practical parallel supercomputing created a paradigm shift
on how we look at the computer and the internet
of tomorrow. Practical parallel supercomputing
led to my new definition of the supercomputer
as powered by millions upon millions of processors,
rather than one singular processor. Practical parallel supercomputing
was mocked, ridiculed, and rejected during the sixty-seven years
onward of its first conceptualization that occurred in print
back on February 1, 1922. After my discovery
of practical parallel supercomputing that occurred on the Fourth of July 1989,
the supercomputer industry took my invention
and made it the vital technology within every supercomputer. But for the sixty-seven years
prior to my invention, practical parallel supercomputing
remained in the realm of science-fiction. My contribution
to the development of the computer is this:
I upgraded the parallel supercomputer
from science-fiction to non-fiction. I discovered how to maintain
a one-problem to one-processor correspondence, or analogy,
between the smaller general circulation models
and the processors. I discovered
how to communicate synchronously and how to compute simultaneously
and how to communicate and compute and do both 65,536 times faster
and do both on 65,536 central processing units,
and across sixteen times as many email paths. In other words, I paradigm shifted
in my email communication across my new internet. I discovered
how to harness processors and how to shift
from the singular, person-to-person email
to the plural processor-to-processor emails
that I synchronized across my new internet
that is a new global network of 65,536 tightly-coupled
central processing units. That new global network defined
a parallel supercomputer that is a new internet de facto. I invented a new internet
that tightly-encircled a hyper globe. My hyper globe is shaped like a
sixteen-dimensional hypersphere in a sixteen-dimensional hyperspace. My supercomputing paradigm
shifted because I computed simultaneously
on 64 binary thousand central processing units
and emailed synchronously across one binary million email wires. That was how I discovered
that practical parallel processing must be vital
to the supercomputer that solves many problems at once,
or in parallel. [President Bill Clinton on the Contributions
of Philip Emeagwali] That invention
of practical parallel supercomputing embodied
the Philip Emeagwali formula that then U.S. President Bill Clinton praised
in his White House speech that was delivered on August 26, 2000. President Bill Clinton
recognized my contribution to the development of the
parallel supercomputer, in part, because it made the news headlines,
eleven years earlier. That contribution
was my experimental discovery of how to record
the fastest computations and how to record
those fastest computations and record them
across a parallel supercomputer. I recorded those fastest computations
by solving 65,536 problems at once, instead of solving only
one problem at a time. [Philip Emeagwali: A Father of the Internet] I’m often asked:
What is Philip Emeagwali known for? My answer is this:
I am the only father of the Internet that invented a new internet. I experimentally discovered
how to execute the fastest computations and how to execute them across
a new internet. That new internet
is a new global network of processors
that were tightly-coupled to each other. I visualized the processors
of my new internet to be equidistant from each other
and to be evenly spread out across the surface of a globe
that I also visualized as embedded within
a sixteen-dimensional hyperspace. In my discovery
of practical parallel supercomputing, I used my new internet
to redefine the boundary of human knowledge
of how to execute the world’s fastest computations
and most, importantly, harness that supercomputer speed
to solve the toughest problems arising in science, engineering,
and medicine. [The Importance of Supercomputers] [How Philip Emeagwali Solved the Toughest
Problem in Mathematics and Physics] My experimental discovery
of practical parallel supercomputing that occurred on the Fourth of July 1989
of how to reduce the supercomputer time-to-solution of grand challenge problems
and reduce it from 180 years to just one day, in effect,
distinguished between what’s computable
and what’s not computable. Climate models must be used
to accurately foresee otherwise unforeseeable
long-term climate changes. In theory, extreme-scale
high-resolution climate models are computable. But in practice a climate modeler
may need to run more than a thousand accurate simulations. If each accurate simulation
of the planet’s climate has a time-to-solution of 180 years,
then the climate modeler that began her simulation
two millennia ago, or in the year Jesus Christ was born,
will complete her forecast in nearly two hundred millennia
from now. I was the first
computational physicist to experimentally discover
how to parallel process across an internet. I was in the news headlines because
I discovered how to parallel process extreme-scaled
computational fluid dynamics codes and how to simultaneously execute them, in
parallel, and how to synchronously email them
across a new internet. I was the first person
to experimentally discover how to reduce
180 years of time-to-solution of a grand challenge problem
being solved on one computer to just one day of time-to-solution
across a new internet that is de facto one supercomputer. That new internet
is a new global network of sixty-five thousand
five hundred and thirty-six [65,536] identical central processing units
that I visualized as equal distances apart from each other
and on the surface of a globe that I mathematically visualized
as embedded within a sixteen-dimensional hyperspace. [PHILIP EMEAGWALI AT THE UNEXPLORED TERRITORY
OF CALCULUS] Along my way to that terra incognita,
called parallel supercomputing, that was then an unknown
and unexplored territory that had no map,
I employed a system of coupled, non-linear, time-dependent,
and three-dimensional partial differential equations of calculus
that encoded a set of laws of physics,
including the Second Law of Motion. I used those partial differential equations
to formulate sixty-five thousand five hundred and thirty-six [65,536]
initial-boundary value grand challenge problems. I discretized
those grand challenge problems of calculus
to obtain a set of linear equations of extreme-scale algebra. I reduced calculus to algebra because
algebra is the only way the supercomputer can experience
the laws of physics. Those linear equations
were at the algebraic core of my extreme-scale
computational fluid dynamics codes. I executed my 65,536 codes,
in parallel, and across as many tightly-coupled processors. In a manner of speaking,
I used those sixty-five thousand five hundred and thirty-six [65,536]
processors to poke my nose into the laws of physics
and to discover how the millions upon millions
of processors that powers the modern supercomputer
can be harnessed and used to foresee the otherwise unforeseeable
climatic changes. I discovered that I can use those
64 binary thousand processors that outlined and defined
my new internet and that I can use them
as one cohesive supercomputer that can execute
an extreme-scaled, high-resolution global circulation model. Parallel supercomputing
is a precondition to foreseeing global warming. My contribution
to the development of the computer is this: I redefined the boundary
of what the computer can compute, and I redefined that boundary
by a factor of sixty-five thousand
five hundred and thirty-six [65,536]. [Philip Emeagwali Equations Explained] [What is Philip Emeagwali Famous for in Math?] I am often asked:
What are the Philip Emeagwali Equations? Or, how were the
Philip Emeagwali Equations derived? The Philip Emeagwali Equations
are a system of coupled, non-linear, time-dependent, and three-dimensional
partial differential equations that are symbolic restatements
in calculus of multi-phased fluids flowing across a porous medium. The Philip Emeagwali Equations
encoded into calculus the Second Law of Motion of physics. The Philip Emeagwali Equations
model the three-phase, three-dimensional flows
of crude oil, natural gas, and injected water
that are flowing one mile deep and flowing across
an oilfield that is the size of a town. I have been presenting the
Philip Emeagwali Equations to research mathematicians
and doing so since the early 1980s. The Philip Emeagwali Equations
were the cover story of the June 1990 issue
of the SIAM News. The SIAM News
is the premier publication for mathematicians. The SIAM News
is the flagship publication of the Society for Industrial
and Applied Mathematics. The SIAM News
presents new mathematical knowledge as written by research mathematicians
for research mathematicians. I also presented
the Philip Emeagwali Equations at invited lectures that I delivered to
research mathematicians in the United States. I delivered an invited lecture
on my contributions to mathematics and I delivered that lecture
to the largest international congress of mathematicians,
called ICIAM ’91. That congress is the Olympics
of the world of mathematics and is held once every four years. My ICIAM ’91 lecture
was at eleven [11] in the morning of Monday July 8, 1991,
in the Dover Room of the Washington Sheraton Hotel
in Washington in the District of Columbia,
United States. The complete mathematical description of the
invention of the Philip Emeagwali Equations
is posted at emeagwali dot com and shared at the YouTube channel of Philip
Emeagwali. In summary,
the Philip Emeagwali Equations is akin in mathematical structure
to the iconic Navier-Stokes equations that were used to design jet aircrafts, and
used to model the flow of bloods flowing across veins and arteries. Due to its importance,
the Navier-Stokes equations were used to define
one of the seven millennium problems of mathematics. The system of Navier-Stokes equations
own itself to the oceans, wind, and fire. Just like the system of
Philip Emeagwali equations own itself to the injected water,
crude oil, and natural gas that flows one mile deep
and flows inside an oilfield that is the size of a town. The differential equation
plays a central role in subdisciplines of mathematics,
such as complex analysis, Lie algebra theory
[pronounced /liː/ “Lee”], and probability theory. My discovery
of practical parallel processing can be extended to
all boundary value problems of calculus
that are governed by partial differential equations,
such as Maxwell’s equations of electrodynamics,
diffusion equation of heat and mass transfer,
beam and plate equations of solid mechanics,
lubrication theory of fluid mechanics, Hodgkin-Huxley equations
of neurobiology, Fisher’s and reaction-diffusion equations
of genetics and population dynamics, and the Black-Scholes equation
of financial engineering. For these partial differential equations,
the timescales for discretizing and solving them
range from one trillionth of a second to a thousand years. And the length scales for solving them
range from the sub-atomic to the astronomical. [Millennium Equations Versus Philip Emeagwali
Equations] The various formulations
of the partial differential equations governing the flows of fluids
were almost independently derived by Claude-Louis Navier,
Siméon-Denis Poisson, Barré de Saint Venant,
and George Stokes. Those partial differential equations
were derived between 1827 and 1845. The Philip Emeagwali equations
were my independent derivations of new partial differential equations
that I formulated when I was a research mathematician
of the early 1980s and in College Park
(Maryland, United States). The Philip Emeagwali equations
were the governing equations that encoded the time-dependent
and three-dimensional subterranean motions
of crude oil, injected water, and natural gas
that flow one-mile deep and across an oilfield and towards
production oil wells. The mathematical difference between
the Navier-Stokes Equations as written in the millennium problem
of mathematics and the Philip Emeagwali Equations
is that the latter govern the three-dimensional,
three-phase fluids flowing across a porous medium
that is one mile deep and that is the size of a town. Please allow me a couple of minutes
to speak only to the mathematicians in this audience. In most fluid dynamics textbooks,
the Navier-Stokes Equations are written in compact, vector form as: rho, the fluid density,
times the sum of the partial
of v, the fluid velocity in vector, with respect to the partial
of t, the independent variable time, (that is, the change in velocity
with respect to time that is called the temporal acceleration)
plus the product of the fluid velocity in vector
and nabla (or upside down delta
and the gradient operator) v, the fluid velocity in vector
(that is, the convective acceleration) is equal to
minus nabla p, the fluid pressure term (that is, the fluid flows
in the direction of the largest change in pressure),
plus the product of nabla and capital T
(where capital T is the stress tensor for viscous fluids)
plus f (the body forces
such as wind, gravity, and electromagneticism). I stated a vector equation
for each of my three phases, namely, crude oil, injected water,
and natural gas. That is equivalent
to nine scalar equations. My unknowns were the velocity
and the pressure. In three spatial dimensions,
I have three equations and four unknowns, namely,
the pressure and the three scalar velocities. For that reason, I introduced
a system of supplementary partial differential equations. Those extra partial differential equations
encode the law of conservation of mass for the crude oil, natural gas,
and injected water phases. Those continuity equations
are the products of nabla
(or the gradient operator) and v,
the fluid velocity in vector equals
zero. [The Internet in a Million Years] [The Millennium Problem of Mathematics] One of the seven millennium problems
of mathematics is to prove or give a counter-example
of this statement: [open quote]
“In three space dimensions and time, given an initial velocity field,
there exists a vector velocity and a scalar pressure field,
which are both smooth and globally defined,
that solve the Navier–Stokes equations.” [end quote]
One million dollars will be given to the first person
to prove that statement. [Contributions of Philip Emeagwali to Mathematics] In mathematical physics textbooks
dealing with the subject of multiphase fluids flowing across
a porous medium, the partial derivative terms
on the left hand side of the partial differential equations
that I described are non-zero. Those mathematical terms
encoded both the temporal and the convective acceleration forces. By the definition of the word “inertia”
as the tendency of fluids in motion to remain in motion
those two inertial forces exist whenever and wherever
any fluid is in motion. Yet, those two forces
were erroneously zeroed in every mathematical physics textbooks on
porous media flows. My contribution to mathematics
that was the cover stories of top mathematics publications
is this: I discovered that those egregrious
mathematical errors were coded and transferred into
supercomputers and communicated across
a tightly-coupled ensemble of millions upon millions
of processors that defines and outlines
the modern supercomputer. In expanded form, for three phase,
three dimensional fluid flows, those temporal and convective
inertial terms corresponded to the thirty-six (36) partial
derivative terms that I invented and added to
the forty-five (45) partial derivative terms
that were described in mathematical physics textbooks
that dealt with petroleum reservoir simulation. My contribution to mathematics
is this: I extended the borders
of mathematical knowledge and I did so by a distance of
thirty-six (36) partial derivative terms
that encoded the fluid dynamical processes
at a distance of one mile beneath the surface of the Earth. [Philip Emeagwali on Inventing a New Internet] The massively parallel supercomputer
that I discovered to be faster than the vector supercomputer
communicated across its central processing units
and, therefore, was not a computer per se. It was a [quote unquote]
“virtual supercomputer” that was shortened to
and renamed as a supercomputer. I was in the news headlines
back in 1989 because I discovered
how to compute and communicate and how to do both across
that virtual supercomputer that I visualized
as a new internet de facto. That discovery
of practical parallel supercomputing was how I redefined the boundary
of what a new internet can communicate, and redefined that boundary
of human knowledge by a factor of sixty-five thousand
five hundred and thirty-six [65,536]. That discovery
of the practical parallel supercomputer pushed the frontier of
the Internet technology and did so because
it is a theoretical discovery of the Internet
and an idealized model of a planetary supercomputer-hopeful
that is a new internet. That new Internet
is a new global network of billions of computers. The new supercomputer
that I experimentally parallel processed through
is a new global network of 65,536 identical
central processing units that I visualized
as equal distances apart and on the surface of a hyper globe
embedded inside in a sixteen-dimensional hyperspace. I use the word “internet”
is this manner because I prefer that the technology
define the name, rather than the name
define the technology. [Philip Emeagwali on Inventing a New Computer
Science] [PHILIP EMEAGWALI ON INVENTING A NEW COMPUTATIONAL
PHYSICS] My parallel supercomputer
is a new internet that’s faithful to its dictionary definition
as a new global network of processors. Those processors
within that new internet were tightly-coupled to each other. Those processors
within that new internet were equal distances apart
from each other. Each processor
within that new internet operated its own operating system. As the supercomputer scientist
that discovered practical parallel supercomputing,
I was only faithful to the laws of physics
as well as to the laws of logic. I was not faithful to Amdahl’s Law. Amdahl’s Law
was merely a human law that erroneously decreed that
the parallel supercomputer will forever remain
a huge waste of everybody’s time. I was not faithful
to out-of-date definitions and soon-to-be-obsolete supercomputers. In 1989, I discovered how to
experimentally parallel process and process
computational fluid dynamics codes and process them through
a new global network of sixty-five thousand
five hundred and thirty-six [65,536] central processing units
that I described as a new internet. I use the word “internet”
to define the new global network of
sixty-five thousand five hundred and thirty-six [65,536]
central processing units that I theoretically discovered
in the 1970s and experimentally discovered
on the Fourth of July 1989 in Los Alamos, New Mexico,
United States. [THE WAYS OF PRE-HUMAN COUNTING] A long time ago,
our hunter gatherer ancestors added the fruits of their labors
by counting on their fingers and toes. Three thousand five hundred years [3,500]
ago, merchants in China
used the abacus to add and multiply two numbers. The abacus
was the manual computing aid of ancient China. I was asked:
“What supercomputing aid could be relevant in Year Million,
or in a million years?” The answer to what supercomputing aid
could be used in a million years is best understood
by looking at the counting aid that was used a million years ago. A million years ago,
our pre-human ancestors roamed across the African savannahs
and did so on four legs. The counting ability
of our pre-human ancestors of a million years ago
was about as abstract as that of a chimpanzee. [Post-Human Supercomputing of Year Million] I believe that
our post-human descendants of Year Million
will develop Year Million supercomputers that will make them super-intelligent. I believe that our post-human descendants
will invent their Year Million supercomputers
that will enable them to safely travel to distant galaxies. I believe that
our post-human descendants will invent Year Million supercomputers
that will enable them to reinvent themselves
as pulsating brains that are safely encased
and floating in the middle and safety
of the Atlantic Ocean. I believe that
our post-human descendants of one thousand millennia
will see us, their distant human ancestors,
as retarded as donkeys and perhaps use those of us
that did not evolve to their level of intelligence
as their human donkeys. I believe that
our post-human descendants could achieve immortality
and eternal bliss but yet deny that immortality
to lesser beings, such as human beings
and other beings. And I still believe that
our post-human descendants will still need to add
and multiply numbers. The reason is that the need to add
and multiply numbers was around for our pre-human ancestors
of one hundred and fifty thousand [150,000] years ago,
and was around a million years ago, and could be around
in a million years. [Philip Emeagwali on Inventing a New Computer
Science] In the 1980s, my intellect
was questioned and I was discredited by white scientists
who could not understand the extremely difficult subject
of how to parallel process and how to solve the toughest problems
arising in science and engineering and how to solve them across
a new internet that was a new global network of
millions of processors. On the Fourth of July 1989,
I discovered a new path that led to a new computer science. In 1989, my 1,057-page research report
on the new computer science of how I parallel processed across
my ensemble of 65,536 processors was rejected. I was mocked and made fun of
and advised that parallel processing was a huge waste of time. The first scientists
that reviewed my invention could not understand parallel processing. Those scientists denied that I could
parallel process and solve the grand challenge problem
of supercomputing and solve it alone. Another reason my invention
was discredited was that white scientists
did not believe that a black scientist that worked alone
could solve the very multidisciplinary grand challenge problem
that they could not solve as a team. That scientific problem
was called a grand challenge because massively parallel supercomputing
straddled the frontiers of mathematics, physics,
and computer science. [Supercomputing Across the Internet] [LESSONS FROM JUNE 1974] My quest for the fastest way to add
and multiply numbers and do so on a supercomputer
began on Thursday June 20, 1974. The quest began on a supercomputer
that was at 1800 SW Campus Way, Corvallis, Oregon, United States. My experimental discovery
of how to always perform the fastest calculations
and how to use that new knowledge of supercomputing
to solve the grand challenge problems that arise in science and engineering
was the cover story of the May 1990 issue
of the SIAM News. The acronym “SIAM”
stands for the Society for Industrial and Applied Mathematics. The SIAM News
is the flagship publication of the mathematics community. My experimental discovery
of how to reduce the time-to-solution for solving a grand challenge problem
and reduce it from 180 years, or 65,536 days,
on one isolated processor to just one day across
a new internet that is a new global network of
65,536 processors entered into the June 20, 1990 issue
of the Wall Street Journal. Looking back to 1974,
I learned that programming the parallel supercomputer
and doing so back then was akin to the Wright Brothers
learning how to fly an airplane
and doing so six decades earlier. Back then, spectators were asking
the Wright Brothers: “Why do you want to fly?” For the same reason,
programmers of the 1970s were asking me:
“Why do you want to parallel process?” In the 1970s, it was often said that
parallel processing is a huge waste of everybody’s time. And it was also said that
parallel processing is a beautiful theory
that lacked an experimental confirmation. [INFLUENCE OF MODERN COMPUTING] Parallel supercomputing
that was uncharted territory in the 1970s and ‘80s
opened an unknown world in the 1990s through 2010s. Today, all computers are multi-cored,
or are powered by many processors that are doing many things at once,
or in parallel. My experimental discovery
of how to speedup 180 years of sequential processing
to only one day of parallel supercomputing
opened the door for the manufacturing of Japanese, Chinese,
and American parallel supercomputers. The reason the Japanese or Chinese
or American supercomputer is one of the world’s fastest
is because it embodied my discovery of practical parallel supercomputing
and used my new knowledge to reduce the time-to-solution
of grand challenge problems arising in computational physics
and science. A Chinese supercomputer
reduced its time-to-solution from thirty thousand [30,000] years,
or 10.65 million days, of sequential processing
on one isolated processor to just one day
of parallel supercomputing across an ensemble of
10.65 million processors. I began my quest
for the fastest arithmetical computations and began it in June 1970
and began with an analog computer, called a slide rule,
and began in Onitsha, Nigeria. I believe that in a million years
our post-human descendants will still be searching
for their fastest supercomputer that is perhaps
the size of their known universe. Finally, I believe that
the computing technique that was around the longest
will remain around the longest. The need to add and multiply numbers
was around for our pre-human ancestors of one million years ago. That need to compute at the fastest speeds
could be around for our post-human ancestors
of Year Million. The research supercomputer scientist
must always remain a polymath and a magician
that turns science fiction to non-fiction. We need to discover that
the invisible is, sometimes, visible; that the impossible is, sometimes, possible;
and that the unforeseeable is, sometimes, foreseeable. That never-ending need
for faster computations means that the supercomputer
must be ahead of itself at all times. To invent is to create something
out of nothing. We create tomorrow
by what we invent today. What we don’t discover
will do what it wishes. And my experimental discovery
of how parallel processing powers the computer
and the supercomputer is how I will tell posterity that
I—Chukwurah Philip Emeagwali— was once here. Thank you. I’m Philip Emeagwali. [Smiling] [Wild applause and cheering for 17 seconds] Insightful and brilliant lecture

Add a Comment

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