Google Smashes the World Record For Calculating Digits of Pi (wired.co.uk) 132
Pi just got bigger. Google's Compute Engine has calculated the most digits of pi ever, setting a new world record. From a report: Emma Haruka Iwao, who works in high performance computing and programming language communities at Google, used infrastructure powered by Google Cloud to calculate 31.4 trillion digits of pi. The previous world record was set by Peter Trueb in 2016, who calculated the digits of pi to 22.4 trillion digits. This is the first time that a publicly available cloud software has been used for a pi calculation of this magnitude.
Iwao became fascinated by pi when she learned about it in math class at school. At university, one of her professors, Daisuke Takahashi, was the record holder for the most-calculated digits of pi using a supercomputer. Now, y-cruncher is the software of choice for pi enthusiasts. Created in 2009, y-cruncher is designed to compute mathematical constants like pi to trillions of digits. "You need a pretty big computer to break the world record," says Iwao. "But you can't just do this with a computer from a hardware store, so people have previously built custom machines." In September of 2018, Iwao started to consider how the process of calculating even more digits of pi would work technically. Something which came up quickly was the amount of data that would be necessary to carry out the calculations, and store them -- 170 terabytes of data, which wouldn't be easily hosted by a piece of hardware. Rather than building a whole new machine Iwao used Google Cloud.
Iwao used 25 virtual machines to carry out those calculations. "But instead of clicking that virtual machine button 25 times, I automated it," she explains. "You can do it in a couple of minutes, but if you needed that many computers, it could take days just to get the next ones set up." Iwao ran y-cruncher on those 25 virtual machines, continuously, for 121 days.
Iwao became fascinated by pi when she learned about it in math class at school. At university, one of her professors, Daisuke Takahashi, was the record holder for the most-calculated digits of pi using a supercomputer. Now, y-cruncher is the software of choice for pi enthusiasts. Created in 2009, y-cruncher is designed to compute mathematical constants like pi to trillions of digits. "You need a pretty big computer to break the world record," says Iwao. "But you can't just do this with a computer from a hardware store, so people have previously built custom machines." In September of 2018, Iwao started to consider how the process of calculating even more digits of pi would work technically. Something which came up quickly was the amount of data that would be necessary to carry out the calculations, and store them -- 170 terabytes of data, which wouldn't be easily hosted by a piece of hardware. Rather than building a whole new machine Iwao used Google Cloud.
Iwao used 25 virtual machines to carry out those calculations. "But instead of clicking that virtual machine button 25 times, I automated it," she explains. "You can do it in a couple of minutes, but if you needed that many computers, it could take days just to get the next ones set up." Iwao ran y-cruncher on those 25 virtual machines, continuously, for 121 days.
Where... (Score:2)
... can I download it from?
Re: (Score:1)
Re: (Score:1)
Re: Where... (Score:4, Funny)
Re: (Score:1)
Hardware Store (Score:5, Funny)
"But you can't just do this with a computer from a hardware store...
I always get my computers from Home Depot.
Re: (Score:3)
Don't you know "Ace is the Place?"
Re:Where... (Score:5, Funny)
Re: (Score:3)
You mean 1.0 (in base pi), right?
Re: (Score:3)
Re: (Score:3)
*sigh* I knew that.
[slowly turns in nerd-card]
Re: (Score:2)
Re: (Score:2)
1 in any base is 1.
A number in its own base is 10.
What about base 1 / unary?
1 is 1 (or |), not 10 (or |0 or | ). There is no 0.
Re: (Score:2)
Probably not, 1 in any base is 1.
I assume you meant that 1 base anything is 1 base 10. If so, then you are wrong. 1 base zero is infinity base ten.
Re: (Score:2)
-1 - -1
0 - 0
1 - 1
2 - 2
3 - 3
4 - 10.22012202112111030100001011...
5 - 11.22012202112111030100001011...
My ID would be 11213000130002.2221120030010203010210012201...
Re:Where... (Score:5, Funny)
Back in college, I jokingly sent my friend 1 million digits of PI. This was on a VAX terminal college e-mail system and he didn't know how to delete the message without scrolling through the entire thing. So he sat there hitting page down over and over until he reached the bottom. For some reason, he didn't seem to appreciate the practical joke.
Re: (Score:2)
Remind me of the time I drop rock on friend Grog. He not like rock smash him. He beat me with club. Then saber tooth tiger eat whole family. Me laugh last!
Re: (Score:2)
Well, thank God you didn't have 30 Trillion digits available. He'd still be there.
Re: (Score:3)
It's waiting for the copyright lawyers from Hollywood to make a claim.
After all, the binary data from every move that was ever made and ever will be made is already in the number pi.
Actually, if you would describe the whole universe in binary format -or decimal if you wish-, it's already in the number pi. Somewhere.
Re: (Score:2)
Or Stanislaw Lem's "His Master's Voice", which probably comes first.
Re:Where... (Score:4, Interesting)
Actually, if you would describe the whole universe in binary format -or decimal if you wish-, it's already in the number pi. Somewhere.
Can't be, because some universal constants are irrational, and therefore cannot be in another number.
Re: (Score:1)
Your mom is pretty irrational and I was inside her last night.
Re: (Score:1)
yes but if you can describe the constant with a generating function that could be encoded as a finite length algorithm and therefore a finite number
doesn't follow. (Score:1)
Can't be, because some universal constants are irrational, and therefore cannot be in another number.
imagine the following number: 0.131415926...
which is
0.1 followed by the digits of pi in base 10.
this is clearly irrational, (since pi is irrational, it will be neverending), yet it's a irrational number that contains another number in them.
Why isn't your favourite irrational number like this? Can you prove that pi is not contained within even e?
Re: (Score:2)
For that to work one infinitely long number would have to be longer than another infinitely long number.
Re: (Score:1)
Sure it contains another number. It contains it by the following procedure:
number B is 'in' number A if for all digits, in order, B is present in A consecutively.
The Nth digit of number B is the N+1th digit of number A. The 1st digit of number A is 1.
Since for all N digits of number B it is present in A therefor B is in A.
* The digits are in order (for all N, it is the case that position N+2 is higher than N+1)
* The digits are all there (for all N)
* The numbers are all consecutive (for all N there
Re: (Score:2)
Pi has not yet been proven to be a normal number. It probably is, but we might be surprised to find it's not after the 32nd trillion digit :)
And our rules for computing it haven't been proven to true. For all we know we're making some arbitrary constructable number while the ratio of the circumference of a circle to its diameter is something else once you get to the grain of spacetime.
Re: (Score:2)
It's been proven trascendental, so no chance it's constructable.
Re: (Score:2)
Re: (Score:3)
Whats the file size in terms of Firefox Send?
It it's too big, just compress it.
Re: (Score:1)
3. there i did it for you. it is a lossy compression algorithm
Re:Where... (Score:4, Funny)
Whats the file size in terms of Firefox Send?
It it's too big, just compress it.
And if that doesn't work, compress it again!
Re: (Score:2)
Here it is exceedingly compressed to 20 or so bytes: "pi to 33 trillion digits".
Re: (Score:2)
But why do you think pi is interesting? (Score:2)
My initial reaction to this story was to wonder how irrelevant this is from a real world perspective. The actual universe is not flat. Made me wonder how many decimal places actually apply to reality. I'm guessing that it's a larger number somewhere out between galaxies... Here on earth, probably less than 10 digits of pi are significant, and fewer than that if you were on Mercury.
Seems to make more sense to calculate an irrational number that has some rational relationship to the real world. How about e? I
Re: (Score:1)
Re: (Score:2)
Re: (Score:1)
I hate to say it, but I agree. What's the point of this? I believe going beyond 40 or so digits already gets you down in the range of accurate to within a planck length or so. And around 20 digits gets you down toward the atomic limit of accuracy for building stuff. So really, who cares? We can't even validate that it's actually correct. We just have an algorithm that say it is.
Re: (Score:2)
" I believe going beyond 40 or so digits already gets you down in the range of accurate to within a planck length or so."
But what if you draw a bigger circle?
Re: (Score:2)
As I seem to recall, pi to 12 digits allows you to pinpoint any point in the UNIVERSE to an atom's margin.
What bigger circle are you drawing that you need 31.2 trillion digits?
Re: (Score:1)
But what if you draw a bigger circle?
Bigger than the visible universe? [nasa.gov]
Well ok, I guess since the universe is larger than just the portion we can see, that would be possible.
Re: (Score:1)
As opposed to . . . athletes showing off.
Or Reality TV.
People with brains might actually appreciate nerds showing off.
I would also point out . . .
All that time and energy wasted on . . . Facebook.
Re: (Score:2)
People with brains might actually appreciate nerds showing off.
Yeah. I'll just leave this [gocomics.com] here.
Re: (Score:2)
So they used idle process in an already existing data center to actually calculate math, rather than just idle.
What the fuck is your problem again? This is exactly the kind of thing that is more efficient on cloud computing.
Re: (Score:1)
Clearly you don’t know much about idle vs under load power consumption. A blade that has an OS booted, but has nothing running, might use something like 150W. Under load, that same blade could use around 450W.
Re: (Score:2)
Re: (Score:2)
Which raises the question: how many valid BitCoin hashes are there in the 3.14trillion digits of pi?
Iwao smashed the record (Score:2)
Sounds like Iwao smashed the record, but she happened to use Google Cloud computers to do it. With just 25 machines, Google staff probably was unaware of her existence.
Re: (Score:2)
It's also Albert Einstein's birthday, relatively speaking.
Re: (Score:2)
You are one of Al's relatives?
Did you inherit his mustache? (I bet Al wouldn't have been so famous if he trimmed his 'stache and combed his hair)
Re: (Score:2)
Re: (Score:1)
How to prove it? (Score:1)
Re: (Score:1)
There are Spigot algorithms that amount to pi pez dispensers.
https://en.wikipedia.org/wiki/Spigot_algorithm
They are ludicrously slow, but you input a digit of position and they will output the digit at that position without calculating all the intermediate values.
https://en.wikipedia.org/wiki/Bailey%E2%80%93Borwein%E2%80%93Plouffe_formula
Re: (Score:3, Interesting)
Since PI has been independently calculated by so many different implementations over time, the initial digits of them can be cross checked for correctness. One early effort, was it in the 1950s maybe(?) calculated PI to 2000 places, and that was called a 'stunt' to show off a
Re: (Score:2)
For your record calculation to be recognized you need to demonstrate that you made some effort to determine if it was correct or not, which usually means using two different algorithms and comparing the results. So actually she calculated Pi to 31.4 trillion digits twice, the second time with different code for verification of the first calculation.
Re: (Score:2)
It's from Google so I imagine it's as reliable as Google's search engine.
Tolerance for error (Score:3)
So if her last digit was wrong, how far off would she be on a calculation of the diameter of the observable universe?
Does pi have any meaning when you get details beyond the Planck length?
Re:Tolerance for error (Score:4, Informative)
Re: (Score:2)
Re:Tolerance for error (Score:5, Informative)
Here's some data [davidpratten.com] on the size of the observable universe in Planck-lengths. 200 digits of pi should be sufficient to precisely compute the number of Planck-length cubed units in our observable universe. From a strictly physical perspective, more than this level of precision in meaningless.
Re: (Score:2)
Pi digits (Score:1)
So the same piece of software, y-cruncher, was used to break the record 6 times.
This latest record has 31,415,926,535,897 digits. Har har, get it?
Re: (Score:2)
Better tagline (Score:2)
Pi just got bigger.
You missed a real opportunity there to say:
There's now a lot more PI to go around.
It's even more delicious than you think at first glance... "around", get it?!?
Re: (Score:2)
When you make pi bigger everybody gets a bigger piece. I'll take mine in apple, if you please. How much more of a byte am I now entitled to?
Why does the headline say Google? (Score:2, Insightful)
Google did nothing other than sell the environment. Maybe the headline should say Intel because it may have built the actual hardware, or Cisco because its switches were used somewhere?
Re: (Score:2)
31.4 trillion digits is just a rounding error (Score:1)
Just sayin'
OK, that's just sad. Really. (Score:5, Funny)
"But instead of clicking that virtual machine button 25 times, I automated it,"
She needs 170 terabytes of space across 25 computers for 121 days to produce 31.4 (Ha!) trillion digits. And she's worried about clicking a button a few times?? Hell, even I'm not that anal unless it was a trivial solution. (for a in `seq 1 25` ; do ./push ; done)
First world problems, I guess.
So in all seriousness, how do you check that? Run it again and see if it produces the same number? If there's a timing bug, it'll differ. If there's (say) a BAD timing bug, it won't; but might differ on a different machine. Or numeric coprocessor problems: One [slashdot.org] Two [wikipedia.org] Three [wordpress.com]. Or cosmic rays actually flipping a bit somewhere. (ECC CPUs?) I realize this is all fun and games, but how do you know that it's actually correct? See if you can use it to successfully square the circle, in which case it's not?
Re: (Score:2)
For verification the tool in question has two methods of calculating digits of Pi, and compares the results.
For automating 25 clicks, I would assume it was more than a single click but even so, what geek hasn't wasted more time automating something that it would have taken to do manually? Manual repetitive tasks are boring, automation is an interesting little task.
Re: (Score:1)
(for a in `seq 1 25` ; do ./push ; done)
You just saved yourself 25 keystrokes...in a mere 38 keystrokes.
Pi Day. A test for factoring primes? (Score:2, Interesting)
Public-key/private-key encryption systems are based on factoring primes and the premise is no one can identify all the primes in a truly huge list of whole numbers starting at zero.
So now that we know what Google can do in corporate spare time with its processors, maybe someone out there with more knowledge that I have can answer the question "Can two-factor encryption be undermined by the computing power Google used today to generate a Pi Day (March 14th) news release?"
Re: (Score:2)
Everything about that statement is wrong.
We don't factor primes, we factor prime products. Furthermore, it's relatively easy to identify primes, or we couldn't come up with the two large primes to multiply together in the first place.
We can also test that the product isn't prime with good efficiency.
What we can't do is efficientl
I wonder how much it cost (Score:5, Interesting)
In terms of electricity.
Or how much would it have cost someone who doesn't work at Google.
25 servers, 121 days, 170 terabytes of data.
And then the real question, was it really WORTH it?
Re: (Score:1)
And then the real question, was it really WORTH it?
It placed a puff piece about Google Cloud in Wired, so their marketing department would say "Yes".
Re: (Score:2)
I found the Google estimate site and used the highest RAM servers available. 25 servers at 24/7 comes to over $277,000 if paid by a customer. Estimate variables below.
That's over $1.1 million USD for the 4 month period.
I'd call this exercise a massive waste of electricity. How much coal was used in this exercise? How many tons of CO2 did this add to the environment?
Site:
https://cloud.google.com/produ... [google.com]
Estimate Details:
25 x Calculating Pi.
18,250 total hours per month
VM class: regular
Instance type: n1-ul
Re: (Score:2)
I found the Google estimate site and used the highest RAM servers available. 25 servers at 24/7 comes to over $277,000 if paid by a customer. Estimate variables below.
That's over $1.1 million USD for the 4 month period.
I'd call this exercise a massive waste of electricity. How much coal was used in this exercise? How many tons of CO2 did this add to the environment?
Site: https://cloud.google.com/produ... [google.com]
Estimate Details: 25 x Calculating Pi. 18,250 total hours per month VM class: regular Instance type: n1-ultramem-160 (160 CPUs, 3844GB RAM) Region: Iowa Total available local SSD space 8x375 GB (3,000GB per server) Commitment term: 1 Year Estimated Component Cost: USD 277,868.11 per 1 month
That's far beyond what the record page [numberworld.org] shows. It appears the record was broken with a single dual-socket Xeon machine. Which is in line with how the previous record (a single 4-socket Xeon machine using older processors) was broken.
Re: (Score:2)
I figured she had access to the highest level cloud servers they offer, with her working there. Reading the article is says that every digit takes additional time, memory, and storage.
She works for Google in their "high performance computing and programming language communities" per the summary (I did read the article as well).
Re: (Score:2)
If you care about having a 'world record' and/or are obsessed with Pi, like this person happens to be, I suppose it is.
What I'm surprised by is the limit of only 25 VMs. In a previous job, we would spin up 64 VM clusters hundreds of times a day to chunk through 100TB of data in a few seconds and shut them down immediately after the data were processed.
While I realize you cannot always make a baby in 1 month with 9 women, I still wonder if this could not have been done more economically both financially and
Re: (Score:2)
I really like your use of "cannot always" there...
Re: (Score:2)
In terms of electricity.
Or how much would it have cost someone who doesn't work at Google.
25 servers, 121 days, 170 terabytes of data.
And then the real question, was it really WORTH it?
25 virtual machines (they say). The record page [numberworld.org] shows that the hardware was a single dual-socket Xeon machine [numberworld.org]. That screengrab doesn't show anything about virtual machines, and it is unclear to me that you would need such an arrangement since the software is fully multi-threaded.
It isn't really anything earthshattering, Peter Trueb used a 4-socket Xeon system with E7-8890 v3's to get the previous record. Emma took 16 more days (15% more) than Peter to do the calculation, and I assume the processors wer
burp (Score:2)
I don't think I can handle any more pie.
Re: (Score:2)
Why would you calculate Pi to so many digits? (Score:3, Funny)
Seems irrational to me.
Bah! Too much work (Score:2)
Way too much work. Just pass a law that sets pi to be 3. https://en.wikipedia.org/wiki/... [wikipedia.org]
Missed opportunity (Score:1)
Was there a long stretch of 0s in it? (Score:2)
I want a job like that... (Score:2)
Great (Score:2)
Re: (Score:2)
So... (Score:2)
Shepherd's Pi 2.0 (Score:2)
Now the Shepherd's Pi song can be extended from one million hours (derived from one billion Pi digits) to 30 billion hours, or 3.5 million years!
Elevator music for geological ages!
In other words (Score:2)
25 virtual machines spent 121 days not analysing protein folding or cures for cancer.
31.415926535897 trillion digits? (Score:2)
I'd like to know the first pi x 10 trillion digits!