20101017, 08:29  #1 
Jul 2003
So Cal
2^{3}·5^{2}·13 Posts 
Another milestone
We've hit another milestone! log[(3^6071)/2] / log(2^32) > 30. Took me a few minutes to figure out why it wasn't working! A quick recompile and filtering is underway.

20101017, 08:55  #2 
"Serge"
Mar 2008
Phi(4,2^7658614+1)/2
23511_{8} Posts 
That's the usual one. Limbs, shlimbs.

20101017, 14:19  #3 
Tribal Bullet
Oct 2004
3^{2}×5×79 Posts 
Fixed in SVN. Soon everyone's codebase will look like Greg's local one :)
Note that I've received a report of a crash in the initial singleton removal pass for a dataset with 1.1G relations, and have not been able to diagnose it remotely. There is an additional fix for a potential problem in the hashtable code, but maybe I should be dusting off the JUDY library and adding it into places that are expected to build massive hashtables. I'm thinking of the initial singleton pass and the initial matrix build. It's a sign of the times when both of these have already had a lot of their working set pushed off to disk, and even with tens of GB of RAM I still have to worry about scaling badly. Last fiddled with by jasonp on 20101017 at 16:20 
20101017, 17:02  #4 
Jul 2003
So Cal
2^{3}·5^{2}·13 Posts 

20101017, 19:16  #5 
"Bob Silverman"
Nov 2003
North of Boston
2^{2}·1,877 Posts 
YES!!!! By curious coincidence, I have just starting sieving 2,1870L
using 31 bit LPS. Since my MP software uses 30bit limbs, I ran into a few problems.... One oversight that I had: If QS fails to split two LP's I try SQUFOF. But I had forgotten that my SQUFOF routine is limited to composites less than 2^60... My siever would run for a while, then hang... not core dump, just hang. It had gotten into an infinite loop in squfof. Tracing it down was a pain.... I have had a few similar problems.... For example, in one place I was calling a special prime test routine that is designed internally to work only on double precision (2^60) integers. It too was hanging. Replacing it with a general routine fixed the problem. etc. etc. I think that everything is OK now. BTW, the CWI post processing code doesn't handle ideals > 2^30 or (a,b) lattice values > 2^30 at all..... 
20101019, 21:12  #6 
Oct 2006
vomit_frame_pointer
2^{3}·3^{2}·5 Posts 
How many fingers am I holding up?

20120425, 22:06  #7 
Sep 2008
Kansas
2^{5}·7·17 Posts 
Faster Deterministic Integer Factorization

20120425, 22:12  #8  
Jun 2003
2·2,719 Posts 
Quote:


Thread Tools  
Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Another milestone!  tcharron  PrimeNet  3  20130829 06:44 
Big milestone coming up  schickel  Aliquot Sequences  8  20110729 10:54 
New milestone  tha  Data  526  20101123 00:09 
New Milestone  opyrt  Prime Sierpinski Project  65  20101006 13:18 
Milestone  davieddy  PrimeNet  2  20070908 12:38 