Page 1 of 2

BOINC and e

Posted: Tue Apr 22, 2008 11:00 am
by CompKronos
well as a a project me and my math teacher and two other students are doing we will attempt to break a world record. What record you ask? well no less awesome then computing 200 billion digits of e (including the two). Yes in reality it seems to be a math spitting contest, but there are various things that can be learned along the way

but also with that is the method we are using to do it: clustered computing! {w00ts}. Mainly we will use a program called BOINC that is perfect for this sort of thing. This really is the main point of my post to hopefully get some people here to download and when i finally set up the project to download and maybe not just my schools computers could help calculate 1/1,000,000!
(factorial not exclamation mark.)

link for BOINC if you want to see anyway it is very neat and may helps some others as well ;) http://boinc.berkeley.edu/

its open-source, and the website has all the documentation needed

Posted: Tue Apr 22, 2008 11:16 am
by WaeV
That sounds cool. I have a book with phi to 5 million digits or so.

Re: BOINC and e

Posted: Tue Apr 22, 2008 11:21 am
by bcnipod
CompKronos wrote:well as a a project me and my math teacher and two other students are doing we will attempt to break a world record. What record you ask? well no less awesome then computing 200 billion digits of e (including the two). Yes in reality it seems to be a math spitting contest, but there are various things that can be learned along the way

but also with that is the method we are using to do it: clustered computing! {w00ts}. Mainly we will use a program called BOINC that is perfect for this sort of thing. This really is the main point of my post to hopefully get some people here to download and when i finally set up the project to download and maybe not just my schools computers could help calculate 1/1,000,000!
(factorial not exclamation mark.)

link for BOINC if you want to see anyway it is very neat and may helps some others as well ;) http://boinc.berkeley.edu/

its open-source, and the website has all the documentation needed
I will help if you want. I use Boinc for SETI.

Posted: Wed Apr 23, 2008 3:22 pm
by RaVNzCRoFT
200 billion digits of e, eh? Good luck. :wink:

Posted: Wed Apr 23, 2008 4:14 pm
by JacksonCougAr
Thats just odd: any of you math geeks figured out about how long this will take?

Posted: Wed Apr 23, 2008 4:27 pm
by Cryticfarm
Mmmm, last record it took a 3.0 dual core and 30 gigs of RAM, 6 days, 4 hours, 41 mins, 6.67 secs to compute 100 billion.

Posted: Wed Apr 23, 2008 6:10 pm
by ScottyGEE
JacksonCougAr wrote:Thats just odd: any of you math geeks figured out about how long this will take?
because e to the 200 biollionth makes someone a geek, if they did it on their own you'd have a point. What does reverse engineering a game do to you then Jackson ;0

That'd be long...
Print a book on it. That, or copy paste it here and I'll be able to read it.


Also, is there absolutely no limit to its decimal places? While I've used e a fair bit, I'm just curious if you could actually run into say, only 101billion before it stops

Posted: Wed Apr 23, 2008 6:13 pm
by GametagAeonFlux
ScottyGEE wrote:Also, is there absolutely no limit to its decimal places? While I've used e a fair bit, I'm just curious if you could actually run into say, only 101billion before it stops
e is usually defined by the following equation:

e = limn->infinity (1 + 1/n)^n.

Found that through the wondrous interwebs...basically, it's infinite.

Posted: Wed Apr 23, 2008 6:19 pm
by ScottyGEE
e to the 1 million digits in a text document is 1013 KB. so multiplying that by 2000 gives 1 978.51562 megabytes

Big text file

I might be wrong though.

Posted: Wed Apr 23, 2008 6:21 pm
by GametagAeonFlux
ScottyGEE wrote:e to the 1 million digits in a text document is 1013 KB. so multiplying that by 2000 gives 1 978.51562 megabytes

Big text file

I might be wrong though.
There's a 70,000,000 digit pi text file that's ~546mb...I think you may just be off a bit.

Posted: Thu Apr 24, 2008 4:11 am
by Cryticfarm
Oh yeah, the text file for 100 billion was 43 gigs.

Posted: Thu Apr 24, 2008 6:22 am
by JacksonCougAr
ScottyGEE wrote:
JacksonCougAr wrote:Thats just odd: any of you math geeks figured out about how long this will take?
because e to the 200 biollionth makes someone a geek, if they did it on their own you'd have a point. What does reverse engineering a game do to you then Jackson ;0

That'd be long...
Print a book on it. That, or copy paste it here and I'll be able to read it.


Also, is there absolutely no limit to its decimal places? While I've used e a fair bit, I'm just curious if you could actually run into say, only 101billion before it stops
Technically we're all geeks :o

Posted: Thu Apr 24, 2008 9:36 am
by CompKronos
to see how long it would take in theory if everything went as planned, my math teacher did 4 billion digits in 12 hours, but his processor was overclocked so there is probably a good bit of errors in the file, so if we kept a processed stock-clocked i am going to say it would take 14 hours for 4 billion accurate digits. So that means about 285714285.8 digits per hour (rounded) so times that by 200 billion thats takes about 5000000000000000000 hours or so (on one comp)

Posted: Thu Apr 24, 2008 10:14 am
by GametagAeonFlux
CompKronos wrote:14 hours for 4 billion accurate digits[...]so times that by 200 billion thats takes about 5000000000000000000 hours or so (on one comp)
What? Wouldn't you just multiply that by 50 and get 700 hours?

Otherwise the timeframe you're talking is ~570,776,255,707,762.6 years (5000000000000000000/24 = 208333333333333333.33 days / 365 [granted that doesn't account for the extra day in a leap year] = ^).

Posted: Thu Apr 24, 2008 12:28 pm
by CompKronos
yea i i knew i had something really wrong but i didn't have time to fix it because i was late for class

Posted: Thu Apr 24, 2008 12:53 pm
by bcnipod
If you want any help with the project, I would be more than willing to help you out.

Posted: Thu Apr 24, 2008 1:59 pm
by RaVNzCRoFT
GametagAeonFlux wrote:
ScottyGEE wrote:Also, is there absolutely no limit to its decimal places? While I've used e a fair bit, I'm just curious if you could actually run into say, only 101billion before it stops
e = limn->infinity (1 + 1/n)^n.
Yeah, it's essentially the right end behavior of the graph f(x)=(1+1/n)^n. The number of digits is infinite, although the value of the number lies at around 2.718.

Posted: Thu Apr 24, 2008 5:31 pm
by WaeV
CompKronos wrote:to see how long it would take in theory if everything went as planned, my math teacher did 4 billion digits in 12 hours, but his processor was overclocked so there is probably a good bit of errors in the file, so if we kept a processed stock-clocked i am going to say it would take 14 hours for 4 billion accurate digits. So that means about 285714285.8 digits per hour (rounded) so times that by 200 billion thats takes about 5000000000000000000 hours or so (on one comp)
Overclocking causes errors? :?

Posted: Thu Apr 24, 2008 6:02 pm
by CompKronos
yes in fact underclocked produces the most accurate results at slower speeds
_
btw somewhere around 699.9 hours so a little longer then the month of February with a leap year i.e. 29 days and about 2 hours

Posted: Thu Apr 24, 2008 6:07 pm
by GametagAeonFlux
WaeV wrote:Overclocking causes errors? :?
If the processor was meant to run at that speed, they would have made it run at that speed. You are sacrificing stability for speed. Think of it like drawing a line on paper...most people have to slow down their drawing speed to get a (close to) perfect line.