News: 1629176408

  ARM Give a man a fire and he's warm for a day, but set fire to him and he's warm for the rest of his life (Terry Pratchett, Jingo)

Pi calculated to '62.8 trillion digits' with a pair of 32-core AMD Epyc chips, 1TB RAM, 510TB disk space

(2021/08/17)


Switzerland's University of Applied Sciences Graubünden has challenged the world record for calculating Pi, claiming it has computed the mathematical constant to 62.8 trillion digits.

The university yesterday [1]claimed it had broken the record, asserting it beat the previous record of 50 trillion digits, set by Timothy Mullican last year, by 12.8 trillion digits, and completed the task in just over 108 days versus Mullican's 303.

Helpfully, the uni has also [2]published details of the hardware used for its feat.

[3]

A pair of 32-core AMD Epyc 7542 processors powered the uni's rig. AMD [4]states the CPU cores spend most of their time at 2.9GHz, can burst to 3.4GHz, have 128MB L3 cache and happily run 64 threads apiece. A server with 1TB of RAM was also employed, with Ubuntu Linux 20.04 installed on a pair of solid-state disks of unspecified size.

[5]

[6]

A JBOD housed 38 7200RPM hard disks, each with 16TB capacity.

Thirty-four of those disks were used to store values swapped from RAM – a design chosen because memory is very expensive. Hard disks were chosen over SSDs because SSD performance degrades over time and the university's designers feared their intensive calculations could cause problems. In all, the uni said 510TB of disk space was used.

[7]That's Numberwang! Google Cloud staffer breaks record for most accurate Pi calculation

[8]Pi(e) Day of the Century is upon us! Time to celebrate 3/14/15 in style, surely?

[9]Linux gods at last turn their gaze to Pi 400: Computer-in-a-keyboard receives mainline kernel support with v5.14

The uni's rig was modest compared to the cloudy 96-vCPU effort that Googler Emma Haruka Iwao employed over 121 days to calculate Pi to 31.4 trillion decimal places in 2019, we note.

An app called [10]y-Cruncher did the work, and was configured to move data in parallel from the server to the 34 disks at around 8.5GB/sec.

[11]

Sharp-eyed readers will have noted that the JBOD housed 38 disks – the other four were used to store the value of Pi itself.

The last ten digits stored on those disks are 7817924264 and they are now the last known digits of Pi, if the uni's calculations are correct. If the world record is confirmed by Guinness, the full number will be published, the uni said. ®

Get our [12]Tech Resources



[1] https://www.fhgr.ch/news/newsdetail/die-fh-graubuenden-kennt-pi-am-genauesten-weltrekord/

[2] https://www.fhgr.ch/fachgebiete/angewandte-zukunftstechnologien/davis-zentrum/pi-challenge/#c15509

[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/servers&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2YRuI3Pr5uDxoUdGZJPY8fAAAAEE&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0

[4] https://www.amd.com/en/products/cpu/amd-epyc-7542

[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/servers&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44YRuI3Pr5uDxoUdGZJPY8fAAAAEE&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0

[6] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/servers&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33YRuI3Pr5uDxoUdGZJPY8fAAAAEE&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0

[7] https://www.theregister.com/2019/03/14/thats_numberwang_google_cloud_staffer_breaks_world_record_for_pi_calculations/

[8] https://www.theregister.com/2015/03/14/pi_day/

[9] https://www.theregister.com/2021/06/15/pi_400_linux_support/

[10] http://www.numberworld.org/y-cruncher/

[11] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/servers&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44YRuI3Pr5uDxoUdGZJPY8fAAAAEE&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0

[12] https://whitepapers.theregister.com/



Bah. I've done more.

Anonymous Coward

I did it on my fingers & toes. (I have a lot of them.) I stopped counting when I hit one Googolplex of digits. I figure the next time I want to speed it up a bit, I'll use my nipples as well.

Re: Bah. I've done more.

Anonymous Coward

Quite a feet.

they are now the last known digits of Pi

Neil Barnes

I suppose 355/113 isn't good enough in this modern, hurry-up, high-precision world!

Re: they are now the last known digits of Pi

Little Mouse

Back in the low-precision 80's, my old maths teacher would often boast that his friend knew PI to *20* (gasp!) decimal places, expecting us all to be amazed.

I never understood why, if he was genuinely actually that impressed about it, he didn't just sit down one night and commit them to memory himself. Or hell, even go one better and learn it to 25, or 30, or maybe something really huge.

Re: they are now the last known digits of Pi

Anonymous Coward

"go one better and learn it to 25, or 30"

If you'd paid more attention to the subject, you might realise that "one better" than 20 is actually 21.

Re: they are now the last known digits of Pi

Ciaran McHale

Innumeracy affects 8 out of every 5 people.

Re: they are now the last known digits of Pi

spireite

My optician told me my eyes set is better than perfect this year.... it's 20/21.

Re: they are now the last known digits of Pi

Anonymous Coward

In my day it was 22/7 and we were lucky to get that...

Engineering approximation

trevorde

I was a mechanical engineer and once told a PhD maths colleague that you could approximate both pi and e as 3. He was appalled.

Re: Engineering approximation

Negative Charlie

A real engineer can approximate anything to 3.

Re: Engineering approximation

Doctor Syntax

Did you ever wonder why your tyres didn't fit properly?

Re: Engineering approximation

Anonymous Coward

I don't often wear my tyres*. I just thought I'd put on weight.

* Although I am getting a little bald...

Two questions

Chris G

Did they show their working out and who checked the result.

Actually, as someone who gets a different answer every time I count my fingers and toes, is is knowing pi to a bazillion places useful for anything?

Re: is knowing pi to a bazillion places useful for anything?

sabroni

It generates a lot of "news" articles.

Dilligent research provides: [1]https://www.theguardian.com/science/2021/aug/17/new-mathematical-record-whats-the-point-of-calculating-pi

extract: Mathematicians have estimated that an approximation of pi to 39 digits is sufficient for most cosmological calculations – accurate enough to calculate the circumference of the observable universe to within the diameter of a single hydrogen atom.

[1] https://www.theguardian.com/science/2021/aug/17/new-mathematical-record-whats-the-point-of-calculating-pi

Re: is knowing pi to a bazillion places useful for anything?

Howard Sway

The diameter of a single hydrogen atom is still fairly massive on the quantum scale : you don't have to go too much further though to reach the Planck length, after which all measurements become meaningless due to quantum fluctuations. So, no, it's not practically useful for anything in the real universe..

Re: Two questions

ssharwood

Yes All explained in the stuff we linked to.

Re: Two questions

jmch

"is knowing pi to a bazillion places useful for anything"

according to the article I read about this including an interview with the team lead, knowing pi to a bazillion places is not that useful, but knowing *how* to calculate pi to a bazillion places in a highly efficient way is useful because the techniques can be applied in other fields of numerical analysis.

It's very Swiss..

Anonymous Coward

.. to demand more precision.

:)

7817924264

Pascal Monett

Damnit, that was my phone number !

Re: 7817924264

adam 40

Damn! Now I've got to change my PIN number number!!!!

So, now what ?

Dwarf

Although its a "cool" * project and they have a big pile of data to play with now, but what are they actually going to do with it ?

Or was it just a cool thing to put on a bunch of CV's / an enabler for doing some other HPC tasks that do real work?

* Or whatever the 2021 version of "cool" is.

This all seems.....

sreynolds

Just so irrational.

Re: This all seems.....

adam 40

not very complex.

Anonymous Coward

I can help them to carry it on if needed...

cat /dev/urandom

Random numbers

Anonymous IV

Could they print out these 62.8 trillion digits and use them as a source of truly random numbers?

Re: Random numbers

Ellipsis

Proving the sequence is random would seem like a more worthwhile endeavour than simply burning electricity running a program for a little bit longer than the last lot, just to willy-wave a meaningless record until the next lot come along and waste even more electricity running it even longer…

Secret messages?

Just A Quick Comment

Have they found any secret messages tucked away in the depts of Pi, maybe from the Creator to us Subjects?

"This is a Beta version. Please report any problems..."

"Hello to mum and dad. Told you I'd get to 32 trillion..."

"Congrations! You have just discovered the secret message."

"I think I made an error back a few digits. Better check..."

Guiness World Records

Anonymous Coward

How are Guiness going to verify the numbers? short of calculating Pi themselves...

karlkarl

Would all those threads provided by the AMD chip actually be beneficial for this specific task?

It isn't an easy problem to break up into parallel equations is it?

I mean, yeah its convenient that they can also surf the web and answer emails whilst their machine is crunching away but I am fairly certain they have other machines for that ;)

adam 40

.... actually, they _did_ have a big email to send, to Greggs.co.uk, after adding _e_ [geddit!!???]

vektorweg

Time to replace the infinite monkey theorem with the pi encoded theorem. If you encode numbers as characters, pi is containing every work ever written.

Harrison's Postulate:
For every action, there is an equal and opposite criticism.