Supercomputing Across a New Internet | Philip Emeagwali | A Father of the Internet | Famous Inventor

Supercomputing Across a New Internet | Philip Emeagwali | A Father of the Internet | Famous Inventor


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] [Supercomputing Across a New Internet] Back in April 1990 and in a night club,
I ran into a Senegalese research computational physicist.
He had left Senegal for Paris (France) and came to US
after several years in Paris. We were drinking beer
at the same table when I discovered his background
as a computational physicist and he discovered my background
as the only extreme-scale computational physicist
that was computing across a new internet
that was a global network of 65,536 processors.
The Senegalese computational physicist was shocked to learn that
a fellow west African research mathematical physicist
had experimentally discovered the new supercomputer knowledge
of how to build new supercomputers that will be powered by
an ensemble of commodity-off-the-shelf processors.
He said in amazement: “I’ve heard of parallel processing
but thought it was science fiction.” It was surreal to him
that he left Dakar (Senegal, West Africa) and went to Paris (France)
to learn computational physics. Here he is in a Night Club
in the United States and learning extreme-scale
computational physics and learning the massively
parallel processing supercomputer and learning the technology directly
from the primary source and from a person
that also grew up in West Africa and learning the technology
from the Nigerian that experimentally invented
the massively parallel processing supercomputer.
The Senegalese research computational physicist described himself
as the first student of the first person
that experimentally discovered how to make modern computers faster
and how to make the new supercomputer the fastest
and how to use that new knowledge
to build new supercomputers. We noted that almost three weeks earlier
and on February 11, 1990, that Nelson Mandela
had walked out of prison. It was 8 p.m. Friday April 20, 1990.
My voice was hoarse from giving my first marathon
daily media interviews on my experimental discovery
of massively parallel processing. Those daily media interviews
continued non-stop and for the past eight weeks
and since I won the top prize in supercomputing.
I began giving media interviews —on my experimental discovery—
and from the then Cathedral Hill Hotel in San Francisco, California
and on February 28, 1990. My first media interviews
took place at the International Computer Conference
of February 26, 1990 through March 2, 1990.
That computer conference was the largest computer conference
and was organized by The Computer Society of the IEEE,
the acronym for the Institute of Electrical
and Electronics Engineers. The experimental discovery
of massively parallel processing that I was being interviewed about
occurred on the Fourth of July 1989. But to my interviewers,
that discovery did not occur a year earlier
but occurred yesterday. Back in 1989,
that discovery sounded like science fiction, instead of the scientific discovery
that it was. Here I was in a night club
at 8 p.m. Friday April 20, 1990. I was straining my voice
to be heard over the loud extended mix and remix
of the 1987 dance hall hit song “Yeke Yeke” by Mory Kante,
the Guinean vocalist and player of the kora harp.
I continued my geek conversation with the Senegalese computational physicist
that was sharing my table. Using the language of a
supercomputer scientist, I said to the Senegalese research physicist:
“I achieved nearest-neighbor emailing by using a naming scheme
that’s binary reflected. It will take a thousand hours
to recite the names of all my sixty-five thousand
five hundred and thirty-six [65,536] commodity processors.”
“Just name eight processors,” the Senegalese physicist pleaded.
“Consider eight processors that are equal distances apart
and that are on the surface of globe and that are married together
as a global network of eight processors and that defined and outlined
both a new internet and a parallel processing supercomputer
computing with eight processing nodes,” I began.
“I visualized my email wires as the 24 bi-directional edges of the cube
and I visualized those eight processors as placed in the positions
of the eight vertices of the cube that I visualized
as tightly circumscribed by a globe.
For my 65,536 processors, I communicated across
higher spatial dimensions and I sent and received emails
by visualizing the hypercube in a sixteen-dimensional universe,
or hyperspace. I visualized my hypercube
as tightly circumscribed by the hypersphere, or the hyperglobe,
and in the same sixteen-dimensional hyperspace. I rephrased my sixteen-dimensional
email communication problem as follows:
“How do I walk along the bi-directional email wires
and pausing only once at each computer?”
Doing so is equivalent to my discovering the nearest-neighbor mapping
of my one-dimensional mesh onto all sixty-five thousand
five hundred and thirty-six [65,536] commodity processors
that are equal distances apart.
That was how I emailed eight petroleum reservoir codes
—each code, an initial-boundary value problem—
that represented eight adjacent production oilfields.
It’s counter-intuitive but my directly connected
petroleum reservoirs were mapped onto
directly connected commodity processors that simultaneously
had different three-bit long identification numbers,
that uniquely identified eight directly-connected processors
and eight contiguous blocks of oilfields that were also directly connected.
I experimentally discovered how to parallel process
across a new internet that is a global network of processors.
I experimentally discovered parallel processing
in the following eight steps for my eight processors.
And I extended that algorithm to my 65,536 steps
for my as many processors. First, I synchronously emailed
my petroleum reservoir code for oilfield number zero
to the processor I named zero. Second, I synchronously emailed
my petroleum reservoir code for oilfield number one
to the processor I named one; Third, I synchronously emailed
my petroleum reservoir code for oilfield number two
to the processor I named three, and not email it
to the processor I named two. For this processor, my email addressing
was counter-intuitive and unconventional. My parallel processing experiment
was counter-intuitive because I emailed my petroleum reservoir code
for oilfield number two to processor three
within my ensemble of processors. It was intuitive and conventional
to email my petroleum reservoir code for oilfield number two
to processor two and to email my code as taught
from the supercomputer textbooks of 1989 and earlier.
Fourth, I synchronously emailed my petroleum reservoir code
for oilfield number three to the processor I named two.
Here, my email addressing is counter-intuitive and unconventional
because I emailed my petroleum reservoir code
for oilfield number three to processor two,
instead of the intuitive and conventional emailing
of computer code three to processor three.
Fifth, I synchronously emailed my petroleum reservoir code
for oilfield number four to the processor I named six.
Here, my email addressing is counter-intuitive and unconventional
because I emailed from computer code four to processor six,
instead of the intuitive and conventional emailing
of computer code four to processor four. Sixth, I synchronously emailed
my petroleum reservoir code for oilfield number five
to the processor I named seven. Here, my email addressing
is counter-intuitive and unconventional because I emailed computer code five
to processor seven instead of the intuitive and conventional
emailing from computer code five to processor five.
Seventh, I synchronously emailed the petroleum reservoir code
for oilfield number six to the processor I named five.
Here, my email addressing is counter-intuitive and unconventional
because I emailed my petroleum reservoir simulation code
that I named six to the processor that I named five,
instead of the intuitive and conventional emailing of code six
to processor six. Eighth, I synchronously emailed
my petroleum reservoir code for oilfield number seven
to the processor that I named four. Here, my email addressing is
counter-intuitive and unconventional because I emailed
the computer code that I named seven
to the processor that I named four instead of the intuitive and conventional
emailing of code seven to processor seven.
I extended this one-to-one mapping algorithm
from the third dimension through the sixteenth dimension.
In the sixteenth dimension, I synchronously emailed
my two-raised-to-power sixteen, or 65,536,
petroleum reservoir simulation codes for 64 binary thousand oilfields
to 65,536 processors that I uniquely named
by 65,536 unique strings of sixteen zeroes and ones.
In the sixteenth dimension, my email addressing was
counter-intuitive and unconventional because I emailed my computer codes
to the processors that I named differently, instead of the intuitive and conventional
emailing of computer codes
to processors that were correspondingly named. My code-to-processor mappings
preserved nearest-neighbor connectivity and did so with
a one-to-one correspondence between my 64 binary thousand codes
and my as many processors. That one-to-one, nearest-neighbor mapping
was at the granite core of how I theoretically
and experimentally discovered how and why
parallel processing makes modern computers faster
and makes the new supercomputer the fastest
and for me to experimentally invent how and why to use
that new supercomputer knowledge to build a new supercomputer
that encircled the globe in the way the internet does.
I experimentally discovered the fastest
parallel processing supercomputer and I discovered it
as a new internet and I discovered it
at the crossroad of processor-to-processor email messaging
and I discovered it at the crossroads between
nearest-neighboring petroleum reservoir simulation codes
that I executed within nearest-neighboring processors
of an ensemble of 65,536 processors that were the building blocks
of a new supercomputer. [Wild applause and cheering for 17 seconds] Insightful and brilliant lecture

Danny Hutson

Leave a Reply

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