20171126, 16:14  #1 
"Mark"
Apr 2003
Between here and the
6,163 Posts 
fbncsieve  a new fixed n sieve
I am pleased to announce new sieving software designed to replace the fixedk sieving functionality of newpgen. fbncsieve is a piece of software that I have been working on for a few weeks. It has a number of distinct advantages over newpgen including:
There are two Windows executables included. One yes the FPU (the one ending in 87) for modular arithmetic, the other does not (the one ending in 86). In my testing they are close enough in speed that I can't determine if one is faster than the other. For some machines, I would expect that one will be faster than the other. The usage is fairly simple. Here are some examples:
The default maxp is 2^62, but fbncsieve will terminate at sqrt(kmax*b^n+/c). fbncsieve will remove remove numbers that are prime from the output, so if kmin*b^n+/c is prime and p > kmin*b^n+/c, then kmin will still be in the output file. One thing I discovered in testing is that newpgen removes those k from the output file when b = 2, but does not if b != 2. For those examples, in the first case, it will stop sieving at sqrt(kmax*b^n+/c) and output the file k_b3_n10+1.pfbw in ABCD format. For the second, it will sieve to 1e15, verify factors, and output the file k_b57_n100001.pfgw in ABCD format. For the third, it will stop at sqrt(kmax*b^n+/c) and output the file in newpgen format. For the fourth, it would convert from ABCD to newpgen format. If the input is not sieved deeply enough, it will continue sieving. For the last, it will continue sieving the input file to 1e16 (if not already sieved that deeply) and output a file in ABC format. I have done a lot of testing and comparing outputs between newpgen and fbncsieve and outside of newpgen's base 2 bug, the only difference is when the software chooses the max p to sieve to. The max p is very close, but not exactly the same. It is likely due to small differences between newpgen's and fbncsieve's algorithms in the calculation. Since fbncsieve uses all integer arithmetic in the calculation and newpgen uses floating point arithmetic, I assume that fbncsieve is more accurate. Of course, even though I done a lot of testing, it is always possible that something was missed. Please give it a spin and let me know if you run into any issues. Although this will help Conjectures 'R Us more than any other project, I intend to use this code as the base to write a Mac/Linux version of fermfact. 
20171126, 18:05  #2  
Banned
"Luigi"
Aug 2002
Team Italia
2·5·479 Posts 
Quote:


20171127, 03:05  #3  
Jun 2003
2·3^{2}·269 Posts 
The title says "fixed k", but all the examples appear to be the exact opposite, ie. "variable k" (aka fixed n).
Quote:


20171127, 13:46  #4  
"Mark"
Apr 2003
Between here and the
1813_{16} Posts 
Quote:
If k*b^n+c is prime and k*b^n+c < max prime sieved, then k will be in the output file. 

20171127, 14:16  #5 
Dec 2011
After milion nines:)
3·463 Posts 
Doesnot work for me on Win 7 x64
First you need to download two dll files ( can you compile it inside exe file) then crah with error 0x00000007 error code 
20171127, 14:33  #6  
"Mark"
Apr 2003
Between here and the
6,163 Posts 
Quote:
Just in case this is something specific to my code, what command line arguments are you using? 

20171127, 16:08  #7 
Dec 2011
After milion nines:)
10101101101_{2} Posts 

20171127, 17:21  #8 
"Mark"
Apr 2003
Between here and the
6163_{10} Posts 
Works fine for me:
Code:
fbncsieve v1.0.0, a CPU program to find factors of k*b^n+c for fixed b, n, and c Changing p_max to 42088836. All remaining terms will be prime. Sieve started: 1 < p < 42088836 with 1410065409 terms p=15485863, 4.896K p/sec, 9525905914 factors found, 46.64M f/sec, 36.8% percentDone. ETA 20171127 11:22 Sieve completed at 42088836. Writing terms file... ...429528886 terms written to k_b3_n10+1.pfgw Clock time: 215.37 seconds at 11852 p/sec. Primes tested 2552668. Factors found: 9570471115. Processor time: 210.14 sec. (0.02 sieving) (0.98 cores) 
20171127, 17:59  #9  
Dec 2011
After milion nines:)
3×463 Posts 
Quote:
I install the mingw64 runtime, restart windows, add GCC to the path... but still same error There is tutorial how to add those DLL to exe code, so it can be independent Last fiddled with by pepi37 on 20171127 at 18:00 Reason: Add more info 

20171127, 18:14  #10  
"Mark"
Apr 2003
Between here and the
6,163 Posts 
Quote:
Code:
Using builtin specs. COLLECT_GCC=gcc COLLECT_LTO_WRAPPER=C:/Program\ Files/mingww64/7.1.0/mingw64/bin/../libexec/gcc/x86_64w64mingw32/7.1.0/ltowrapper.exe Target: x86_64w64mingw32 Configured with: ... excluding list of libraries ... Thread model: posix gcc version 7.1.0 (x86_64posixsehrev2, Built by MinGWW64 project) Do you try some of the other examples? Is anyone else experiencing this problem? 

20171127, 20:26  #11  
Dec 2011
After milion nines:)
3×463 Posts 
Quote:
Doesnot work for me on Win 7 x64 

Thread Tools  
Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Sieving k * 2^n + c with Nvidia GPU's for fixed k  diep  GPU Computing  5  20160923 19:19 
A siever for K (b, n, c fixed)?  pepi37  Software  7  20150710 04:42 
Sieving k*2^n1 With Fixed n  c10ck3r  Riesel Prime Search  14  20130203 00:19 
User interface bug fixed on LLR V3.8.4  Jean PennĂ©  Software  0  20110122 16:47 
KEP is reporting computer fixed  KEP  Twin Prime Search  3  20070213 18:29 