Why do we decipher the circumference? A company in the United States has already decoded the circumference to 105 trillion digits, and simply storing data requires 1 million gigabytes (G) of memory. But why are humans so obsessed with calculating more places on the circumference? What does this background mean?
Everyone knows the circumference is 3.14159, but the problem is, it’s not just elementary school knowledge. Someone calculated this number down to 105 trillion digits, and a million gigabytes of storage was filled just to record these numbers. It’s not about making movies or designing games, it’s about scientists, programmers, and hardware engineers doing something that “seems meaningless”: working out the circumference (π) to the end.
This is not a joke. In 2022, the U.S. supercomputing team ran the top-level server for five months and calculated π to 105 trillion digits. The formula didn’t pop out just with a click of a mouse, but it pushed an irrational number that had positive signs intersecting and never converging to zero to a depth that an ordinary person could not imagine.
105 trillion numbers, every single one is not duplicated and there are no rules. If you get one wrong, you have to do it all over again. Someone spent half a year on electricity, thousands of computing power, hundreds of terabytes (TB) of memory and tried to get just a few new numbers. Why is that?
This may seem crazy to outsiders, but this “game of madmen” has been going on for thousands of years. It started in Gobabylonia, Go China, and Go India and continued into the AI era. Cho’s value was already between 3.1415926 and 3.1415927 more than 1,500 years ago, almost a thousand years ahead of Europe. Instead of using a calculator, he used abacus, bamboo, paper and pen to split and break an infinite number of fire 循環. It wasn’t until the end of the 16th century that Europe slowly caught up.
In the Newtonian and Euler era, circumference calculation became the threshold of mathematical precision. Someone could prove that their algorithm is strong and their calculus system is strong by calculating more π. In the 20th century, circumference has completely fallen into the realm of robots. IBM 701, one of the first electronic calculators, did this, and with the advent of the Gray supercomputer in 1973, π jumped to hundreds of thousands of digits, more than a hundred.
That’s right. It is now a competition of hard cores to determine who has the most computing power. The last time Google broke the record, it used its cloud server platform to push π to 31.4 trillion in 121 days. Later, the University of Glaruben Application Technology in Switzerland challenged and used high-performance workstations to reach 62.8 trillion. They used fast Fourier transform (FFT), Chudnovsky formula, integer division, and multi-thread optimization to build a mathematical abyss through computational power.
Then, California threw another 105 trillion digits. This time, the math formula has reached its limit, not just memory distribution, cache scheduling, disk read/write speed. Note that the circumferential rate calculation is not just a violent one, but it should be checked to avoid errors. If even one digit is wrong, the whole thing collapses. That’s why scientists usually calculate each with two different algorithms and have to match to pass.
What really makes me feel like my scalp is that the meaning of this π war is already completely out of the “I want to know how much” category. In fact, up to 15 digits (3.14159265358979) of π used for rocket launch, aerospace, and missile location tracking is sufficient. It can cover up to thousands of kilometers of error. So why do you keep calculating?
Because this isn’t “to use” at all. If someone calculates more, he or she can prove “my machine is strong,” “my algorithm is stable,” “my server bandwidth is large,” and “my chip is stable.” It’s pure power competition. The winner takes the new “mathematics hegemony.”
Even if you think circumference is useless, the engineer says, “It’s for aerospace navigation.” Even if you think 105 trillion digits are too much, the cryptographer says, “The number sequence of π is almost unpredictable. Perfect for pseudorandom numbers.” Even if you think it’s you, the fact is that national laboratories, commercial giants, semiconductor industries, and AI developers are collectively betting. Even in some neural network models, the fractional precision of π is used for high-dimensional spatial data synthesis structures.
When training AI, a single numerical precision error can cause the result to deviate hundreds of kilometers. The precision of the π value itself is part of the underlying math. Moreover, labs specialize in using circumference as a system test item. If you calculate a million-digit π to see if memory scheduling doesn’t collapse, if floating-point units are stable, and if I/O throughput comes with it, you’ll see how valuable the system is. π has become a 試. It tests not only the formula, but the “muscles” of the entire digital world.
After all, it’s “hard leverage.” No one wants to lose on this table, even though they know they don’t need that many seats for real use. Like a marathon, I don’t expect anyone to really run all the way to Beijing to work, but the one who wins is at the pinnacle of global algorithms.
In addition, circumference calculations produce by-products. They indirectly develop faster high-precision algorithms, more stable data verification methods, stronger compression storage structures, and even liquid-cooled 散熱 and chip power optimization. So this battle isn’t over yet. It’s 105 trillion seats now, but teams are already looking for Giga-jo positions in the future.
You could say this game is useless, but you have to admit that one breakthrough involved generations of hard work, thousands of chips, and millions of degrees of electricity. It’s not just a simple calculation. It’s a deeper question: how far are the limits of humanity?