20061026, 21:41  #1 
"Jason Goatcher"
Mar 2005
5·701 Posts 
Alternative Sieving for 10M digit prime search
Because of interest in my idea for an alternative sieving project for the 10Mdigit prime hunt, I'm starting this thread. I'm putting it in LMH because at the moment there are only three of us, and this seemed the best place to come without cluttering the boards with a new forum. Basically, we'll just be doing combined trialfactoring, instead of the Prime95 version.
So far there are 3 different factoring efforts going on: k=5 plus 1, k=5 minus 1, and a k=9 minus 1, starting from n=33,000,000 for k=9(I really hope this is a typo or shorthand, because this wouldn't produce a a number with more than 10M digits) If anyone has any additional information, please post. At the moment, I'm headed over to Riesel Sieve to see if it's possible to use JJSieve with this project. 
20061026, 23:16  #2 
"Jason Goatcher"
Mar 2005
3505_{10} Posts 
correction: it's k=7 plus 1, k=7 minus 1, k=9 minus 1

20061028, 00:33  #3  
"Jason Goatcher"
Mar 2005
5×701 Posts 
Additional information having to do with FFT lengths, gleaned from another thread:
Quote:


20061028, 07:36  #4 
Mar 2004
Belgium
1505_{8} Posts 
Yes I am testing from 33M?

20061029, 01:06  #5 
"Jason Goatcher"
Mar 2005
5·701 Posts 
Unless your k is more than about 66,010 digits, the k/n pair won't meet the 10 million digit mark.
Edit: Try n=33,219,281 Last fiddled with by jasong on 20061029 at 01:07 
20061029, 01:15  #6 
"Jason Goatcher"
Mar 2005
5×701 Posts 
I've been informed that if we sieve 3^16*2^n+1 the sieve will be faster, because only numbers of the form 32x+1 will have to be checked. If JJSieve automatically makes us of this fact, then I will do a little sieving to start a file and state a goal to be reached for sieving. Probably the same goal as the bit depth the numbers are sieved to with Prime95(could someone please post the bit depth that Prime95 normally uses for these numbers?), although people can check out numbers early. Checked out numbers will continue to be sieved, just in case we(they?) get lucky.
I'm asking some questions over at Riesel Sieve in the "General Sieve" forum, if anybody wants to drop in and look. 
20061030, 10:10  #7 
Mar 2004
Belgium
3^{3}×31 Posts 

20061031, 00:29  #8  
Jun 2003
1575_{10} Posts 
Quote:


20061031, 23:52  #9  
"Jason Goatcher"
Mar 2005
5·701 Posts 
Quote:
I'm going to receive some AMD parts in about a week. Sometime in the week after that they'll be running Linux. Then I'll finally be able to help with this. 

20061103, 17:12  #10 
"Erling B."
Dec 2005
2^{2}×19 Posts 
Status:
I am sieving for k=7 For k=7, K*b^n+1, n=33219281  33259281 (40.000 candidates) Sieve with NPGen up to 1.1 trillion. there is 1996 n's remaining (candidates) For k=7, K*b^n1, n=33219281  33259281 (40.000 candidates) Sieve with NPGen up to 1.2 trillion. there is 840 n's remaining (candidates) I was estimating that sieving up to 1.500 trillion will be the goal here (bit >50 if I am correct) and see what will be left? ...and I will continue sieving …. 
20061103, 18:00  #11 
Mar 2004
Belgium
3^{3}·31 Posts 
FWIW, I am doing the search from 33.2M to 100M, 3M candidates remaining.

Thread Tools  
Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Search of all even15digit Aliquot cycles  Drdmitry  Aliquot Sequences  25  20161216 15:26 
Polynomial search for 204digit cofactor of M1009  fivemack  Factoring  45  20120214 08:50 
Deep Sieving 10m Digit Candidates  lavalamp  Open Projects  53  20081201 03:59 
Help Sieving 10 Million Digit Candidates  lavalamp  Riesel Prime Search  26  20080525 08:24 
idea about 10 million digit search(possibly dumb)  jasong  Math  5  20060607 10:39 