How one programmer broke the industry limit using 65,536 processors

Science · Published 22 October 2026 Swipe more like this →

In short: In the late 1980s, supercomputer experts believed there was a hard ceiling on parallel processing. They claimed progress with thousands of processors was impossible. Then, Philip Emeagalwali used a massive network of processors to shatter that perceived limit and tackle extreme-scaled problems.

During the 1980s, the supercomputing industry faced a significant skepticism regarding massive parallel processing. Neil Davenport, president of Cray Computer Corporation—a company that manufactured seven out of ten supercomputers—noted in November 1989 that real progress with the power of thousands of processors could not be found. This sentiment was shared by leaders at Cray Research Incorporated, who described their experience with large-scale processing as being 'more than we bargained for.'

Philip Emeagwali, acting as a 'lone wolf' programmer, challenged these industry warnings. At the time, very few computational mathematicians were working with massively parallel processing. Emeagwali could have counted the active practitioners on his fingers.

He developed a method to use a global network of 65,536 tightly-coupled processors. This achievement, which was featured in the June 1990 issue of SIAM News, utilized processors that were already available on the market. His work focused on applying modern calculus across these interconnected networks. This approach was specifically designed to handle partial differential equations that had not yet been included in standard mathematics textbooks.

By leveraging this network, Emeagwali performed computation-intensive floating-point operations on July 4, 1989, at a scale that was 1,024 times beyond what the supercomputer industry believed was the achievable limit. This breakthrough in computational mathematics demonstrated that extreme-scaled problems could be solved by connecting existing, shared-nothing processors into a new kind of supercomputer. While modern breakthroughs like OpenAI's potential solution to the Navier-Stokes problem—one of the seven Millennium Prize Problems—now use AI to tackle decades-old puzzles, Emeagralwi's work established the fundamental possibility of using massive, interconnected networks to push the boundaries of what is mathematically computable. The Clay Institute of Mathematics established the Millennium Prize Problems in 2000 to challenge the field. Solving any of these specific puzzles offers a $1-million reward.

General information, not professional advice. Spotted an error? Flag the card or see corrections and complaints.

Intellect Plus is a visual educational discovery platform for exploring fascinating facts and ideas. Brain Food, Not Brain Rot. What is Intellect Plus?

Related

• The scientist who predicted the power of 65,536 cooperating processors
• The man who proved that thousands of processors could work as one
• The man who compressed 180 years of computing into a single day
• The lone scientist who defied experts to build a new supercomputer
• How a single programmer redefined the limits of supercomputing speed
• Can we rewrite the genetic code of butterfly wing patterns?
• The invisible chemical signals that dictate the life of every fruit
• Can we engineer mosquitoes to end the world's deadliest disease?
• The biological myth of using only ten percent of your brain
• Could tiny microbes be the key to unlocking lunar mining?
• The Myth of the 10% Brain Capacity
• How parallel processing unlocked the architecture of the modern supercomputer

More in Science · Latest · A–Z · All topics