Yes, that’s the idea, it’s just that the “spoiler” likely only revealed something that was already known (that specific digit), or at any rate, something that could be computed on a much smaller computer and in less time. Mostly though, that’s a bit of mathematically interesting info.
I don’t feel like watching a video but maybe there will be a more informative article sometime. I wonder if they used some existing software like Y-cruncher: https://en.wikipedia.org/wiki/Y-cruncher
Says they used a cluster with 2.2PB of flash drives (that was Kioxia’s contribution) and the calculation took 7.5 months. Article is otherwise sort of useless. I’d like to know more technical and mathematical details. It gives the “spoiler” that the 300 trillionth digit of pi is 5, but that was already relatively easy to compute using the Borwein-Bailey-Plouffe algorithm and was probably already known. The BBP algorithm lets you compute a specific digit like the 300 trillionth, using fairly little memory and much less compute time than computing all of the digits.
https://en.wikipedia.org/wiki/Bailey–Borwein–Plouffe_formula
Well ackshually…
Gluster? I don’t know much about it. As someone else asked, how much data are you talking about? Do you need a file store (mutable files), or are you ok with an object store (immutable)?
NFS is still a thing.
Nice. It appears to run on a phone under termux, rather than being a server app. Good to know about either way. Why flask? That’s a web server thing, I thought.
Cool, yeah, the digits themselves are at most 0.5 byte each ;). I don’t know enough about the higher level algorithm to say exactly how the rest of the storage is being used. There is a book called “Pi and the AGM” about pi computation and similar algorithms that is supposed to be really good, but it looks over my head mathematically.