20121017, 18:47  #122 
∂^{2}ω=0
Sep 2002
República de California
2^{3}×3×487 Posts 
Updated source (just the mainprogram .c file  headers same as before) attached to this message.
By way of clarification: For those doing deep sieving, since the code uses both stdout and stderr for output, you'll want to do redirecttofile by e.g. appending " &> err.log" to the command line. (The & causes all ostreams to get redirected to the target of the > ). (You can further append " &" to put into bg mode immediately, or do ctrl+z / bg after invocation to accomplish the same thing). Only changes to the source are slightly improved helpdiagnostics (try invoking the executable with no args, for example) and a fix for the bug I mentioned running into leading up to the first release, which allows me to get rid of the slow workaround and gain ~5% more speed. As always, let me know of any issues and suggestions for future versions. Cheers, Ernst 
20121017, 19:19  #123 
Banned
"Luigi"
Aug 2002
Team Italia
11346_{8} Posts 
On post #74 Ernst gave us the tables of the candidates for this deep sieving project.
On post #118 I asked for information about some ks that should have been sieved out, since the limits on Will Edgington's page were higher. I would like to avoid duplication of work, but can't find the factors that eliminate those ks. Is there anyone that has them? Maybe Will (I will ask him by email)? Or Phil? Or have they all just been ruled out by PRP? On post #120 LaurV told us that "2*92*M1257787+1 does not divide MM1257787": was he referring to a PRP test he did? I'm sorry to ask so many questions, but as I have take the task of centralize information on double Mersenne numbers on one site (maybe too) seriously, now I am worried of losing bits of data... Thank you for your patience... Luigi 
20121017, 19:22  #124 
Banned
"Luigi"
Aug 2002
Team Italia
2×41×59 Posts 

20121017, 21:15  #125  
∂^{2}ω=0
Sep 2002
República de California
2^{3}×3×487 Posts 
Quote:
Code:
Timing tests in 3 different ranges vs #sieve_prime, all with p = 43112609 (all run with fulltime LL test on other CPU): Commandline arguments #prime: 1e3 1e5 1e5 1e6   imax 1000000000 k 5 11.4s 11.6s 12.4s 21.3s imin 7000000000 imax 8000000000 k 665 9.3s 8.0s 9.2s 18.2s imin 10007000000000 imax 10008000000000 k 665 20.7s 19.2s 18.1s 28.7s ***** <<< optimize for this *** 

20121018, 02:32  #126  
Romulan Interpreter
"name field"
Jun 2011
Thailand
23215_{8} Posts 
Quote:
For things related to your post #118, I don't think that somebody has factors for those. Revisiting the post now, I see those k's are the starting values on each row, those which were grayed in my initial table (because they were under the watermark, i.e. the historical "k" was higher). I believe those k's were eliminated by trial factoring, same as I did in the paragraph above. You should ask Toni, or the persons involved. Last fiddled with by LaurV on 20121018 at 02:34 

20121018, 09:55  #127 
Banned
"Luigi"
Aug 2002
Team Italia
2×41×59 Posts 
@Ernst: I will do some tests during the weekend and post my results.
@LaurV: Thank you for the answer, I will contact Tony Forbes hoping to have some more data to integrate. Luigi 
20121018, 18:50  #128 
"Åke Tilander"
Apr 2011
Sandviken, Sweden
2×283 Posts 
Collecting residues
There will never ever be anyone who would be able to do a LL on these large MMs, so I think it would be very, very important thing to collect the residues since sooner or later somebody would like to do a doublecheck.

20121019, 12:20  #129 
Banned
"Luigi"
Aug 2002
Team Italia
12E6_{16} Posts 
Tony Forbes answered.
Now I have all data relative to MM34 and MM35.
Tony did his sieving work for MM34 and MM35 up to 20G for k<500,000, so we just doublechecked his work for k < 1,000. The "Deep Sieving" pages have been reworked, while I am still in process of inserting Tony's new k. It won't be an highpriority task. Meanwhile, I'd like to know the process of selecting ks, should I decide to raise them. Must they only be 0 or 1 mod 4, and 2kp+1 mod 120 = {16 known values over 60 possible}? Luigi Last fiddled with by ET_ on 20121019 at 14:06 
20121020, 11:29  #130  
Banned
"Luigi"
Aug 2002
Team Italia
1001011100110_{2} Posts 
@Ernst.
On post 74 you said: Quote:
Meanwhile, I started some benchmarking on factor_qmmp to optimize NUM_SIEVING_PRIME. Luigi 

20121021, 16:02  #131 
Banned
"Luigi"
Aug 2002
Team Italia
2·41·59 Posts 
factor_qnnp optimzation tables
I ran some benchmarks on factor_qmmp, varying the following parameters:
 The NUM:SIEVING_PRIME parameter  The exponent of the double Mersenne number  The range computed  The sieve position where the range was taken The results have been uploaded here. Considerations: 1  It would not be hard designing a regression line to grant each exponent/range the correct NUM_SIEVE_PRIMES 2  Changing k from 8 to 1000 won't affect performance 3  We have a limitation for k higher than (say) 100,000, and must wait a 64bit version of the siever 4  Note to Ernst: sieving ranges of 1G starting from k=500T produces a strange shrink of more than 10% in the calculation time, Did the program loose a loop or get a wrong exit flag? Apart from this, the best NUM_SIEVING_PRIMES value for each range is evident from the first and last table (those with range = 10G). Enjoy! Luigi 
20121027, 09:58  #132 
Banned
"Luigi"
Aug 2002
Team Italia
1001011100110_{2} Posts 
First one kocked down!
MM(32582657): k=20 has a factor: 665316631883
Next k=60 Luigi BTW, is anybody else testing/sieving? I had MM(1257787), k=93 reserved by LaurV... Last fiddled with by ET_ on 20121027 at 10:00 Reason: I just noticed that I can't edit the title, evidently wrong... 
Thread Tools  
Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Is CEMPLLA 1.5 "the only software in the world capable of discovering" something? Not really.  CRGreathouse  Number Theory Discussion Group  51  20181216 21:55 
Official "World cup 2014/2018" teat  LaurV  Hobbies  74  20180711 19:33 
Problem E7 of Richard Guy's "Unsolved problems in number theory"  Batalov  Computer Science & Computational Number Theory  40  20130316 09:19 
Is the USA the "new" peacekeeper of the world??  outlnder  Soap Box  20  20050203 09:30 
Would Minimizing "iterations between results file" may reveal "is not prime" earlier?  nitai1999  Software  7  20040826 18:12 