By David Shiga A Yahoo researcher has made a record-breaking calculation of the digits of pi using his company’s computers. The feat comes hot on the heels of a breakthrough Rubik’s cube result that used Google’s computers. Together, the results highlight the growing power of internet search giants to make mathematical breakthroughs. One way to show off computing power is to calculate pi to as many digits as possible, creating a string that starts with 3.14 and continues to the nth digit. The more digits one wants, the more computations it takes. But it is also possible to skip ahead to the nth digit without calculating the preceding ones – for example, determining that the 10th digit is 3, without having to find the first 9 digits: 3.14159265. This is another way of testing computing power, since more computations are required to find higher values of n. Now, Tsz-Wo Sze, a computer scientist at Yahoo in Sunnyvale, California, has used the company’s computers to calculate the most distant digits yet. His computer program represents pi in binary notation, and calculated the 2 quadrillionth (2 x 1015) binary digit, or bit, of pi. It is twice as distant as the previous record, which found a string of bits around the 1 quadrillionth bit. Calculations that do not skip any digits have come nowhere near this remote territory. The latest record for that kind of calculation is 2.7 trillion digits in decimal notation, which works out to around 9 trillion (9 x 1012) bits. Sze’s program was installed on 1000 Yahoo computers, each equipped with eight processors. They ran the calculations in July, when they were in low demand for regular work, doing in 23 days what would have taken half a millennium using just one processor. The computation was made possible by open-source software called Hadoop that allows thousands of networked computers to be used as if they constituted a single extremely powerful machine, a concept called cloud computing. Yahoo programmers have done much of the work to develop Hadoop, though it draws on ideas first published by Google. Yahoo is not the only internet giant delving into abstruse mathematics calculations. A recent result showing that any configuration of a Rubik’s cube can be solved in 20 moves or less relied on distributing calculations across many computers at Google, completing in a few weeks what would have taken a single computer 35 years. Sze says the computing power that companies like Yahoo and Google can bring to bear on these problems is a by-product of their need to speedily process vast amounts of web-related data. “That is why we are building large-scale computation systems,” he says. “As a technology company with big data at its core, we are excited by the possibilities of distributed software systems, most notably, Hadoop,” says Eric Baldeschwieler, vice president of Hadoop engineering at Yahoo. Calculations of pi are especially suited to distributed computing because they are easily broken into smaller parts, says David Bailey of the Lawrence Berkeley National Laboratory in California, which is setting up cloud-computing facilities to run such problems. “They can be divided into sections and assigned to separate computational processors, which can then operate almost completely independently of the others,” says Bailey, who in 1996 co-discovered the first formula allowing one to skip ahead to compute distant digits of pi. Distributed computing is also being used in a project called the Great Internet Mersenne Prime Search. It searches for large examples of a special class of prime numbers using computing power donated by individual volunteers via the internet. More on these topics: