The Quantum Computer is the Biggest Obstacle in Supercomputing | Philip Emeagwali | Famous Engineers

Back in the 1970s and ‘80s, my unorthodox parallel processing approach to supercomputing met a lot of resistance. I was rejected and mocked whenever I proposed that parallel processing will work. In those two decades, my massively parallel processing supercomputing premise was that the logic of the grand challenge problem should determine how the problem should be solved, not vice-versa. That is, it’s only the laws of logic and physics that are sacrosanct, not the technology that, in the first place, must bend for the laws of logic and physics. In an article dated June 14, 1976, the Computer World magazine interviewed the foremost supercomputer experts that were attending the National Computer Conference in New York. Those supercomputer experts
unanimously told
the Computer World magazine
that the supercomputer theory
of parallel processing will be
[quote]
“a waste of time.”
[unquote]
In 1989, twenty-five thousand [25,000] research supercomputer scientists
logged on each day
onto conventional vector supercomputers.
Due to that skepticism and negative press,
it was widely believed that
parallel processing
is a huge waste of everybody’s time.
For that reason, I was the only person
that was logged on each day
onto the most powerful
and the most massively parallel processing supercomputer
in the world.
I visualized my modern massively
parallel processing supercomputer
as my new internet
powered by a global network of
64 binary thousand processors.
I visualized my new internet
as married together
as one seamless, cohesive
whole supercomputer.
I visualized my new supercomputer
as an ensemble of
64 binary thousand processors
that were married together
by one binary million email wires.
In 1989, I was researching alone
on how to use
sixty-five thousand
five hundred and thirty-six [65,536] commodity processors
and how to use them
to solve
one grand challenge problem.
In nineteen eighty-nine [1989],
it made the news headlines
that I—Philip Emeagwali,
an African Supercomputer Wizard
in the United States
has experimentally discovered
how to use a new internet
that’s a global network of
sixty-five thousand
five hundred and thirty-six [65,536] commodity processors
and how to use that new internet
to solve the toughest
initial-boundary value problems
in calculus and physics.
I experimentally discovered
how to use my internet
as a massively parallel processing supercomputer
and use that internet
to reduce the time-to-solution
of the most computation-intensive
grand challenge problems.
I experimentally discovered
how to speed up
from one hundred and eighty [180] years,
or sixty-five thousand
five hundred and thirty-six [65,536] days, within only one processor
to just one day
across one internet.
I visualized that new internet
as a global network of
sixty-five thousand
five hundred and thirty-six [65,536] commodity processors.
The two leading lights
of sequential and vector processing supercomputing paradigms
—namely, Gene Amdahl and Seymour Cray, respectively—
argued that it will be impossible
to experimentally record
the speed increase in supercomputing
that I recorded in 1989.
The 25,000 [quote unquote] “hot brains,”
or conventional supercomputer scientists
at National Science Foundation supercomputer centers stayed with conventional vector processing supercomputers.
The reason those 25,000
supercomputer scientists
stayed away
from the massively parallel processing
supercomputer
was that each believed that
it will be impossible
to use 64 binary thousand processors
to solve
one grand challenge problem.
In contrast, I believed that it will be possible
to parallel process
and to do so when it seemed impossible
to do so.
Beyond faster computation speeds,
using several cores
—in both computers and supercomputers—
has other rich consequences.
One such advantage
of multicore processing technology
is that it increased
the reliability of the modern computer
and improved
the fault-tolerance
of the modern massively parallel processing supercomputer.

A 12-year-old writing a school report
on the contributions of Philip Emeagwali
to the development of the computer
asked me:
“How do we increase
the speed
of quantum computers?”
I answered:
In classical parallel computing,
I experimentally discovered
how to solve
all sixty-five thousand
five hundred and thirty-six [65,536] challenging problems
and how to solve them
at the same time
and how to solve them across
a global network of one million
forty-eight thousand
five hundred and seventy-six [1,048,576] commodity email wires
that fed data and answers
from initial-boundary value problems
and fed them to and from
sixty-five thousand
five hundred and thirty-six [65,536] commodity processors.

TAGS
supercomputer how to build, supercomputer iq, supercomputer, parallel supercomputer versus quantum supercomputer, supercomputer vs mainframe

Philip Emeagwali 180125 1 5+6 of 8

Questions?

%d bloggers like this: