Loading ...
Sorry, an error occurred while loading the content.

Some Benchmark Primality Test

Expand Messages
  • calimero22
    I post some time test for a primality or pseudoprimality check. Tests on AMD Sempron 3000+ (1800 Mhz) WIndows XP n=31838235*2^29717+1 digits: 8954 Tests
    Message 1 of 5 , Aug 31 1:05 PM
    • 0 Attachment
      I post some time test for a primality or pseudoprimality check.

      Tests on AMD Sempron 3000+ (1800 Mhz) WIndows XP
      n=31838235*2^29717+1 digits: 8954
      Tests run by GIOVANNI DI MARIA - email: calimero22@...
      Tests give at least a Pseudoprimality PRP.

      ===================================================================
      GMP Library
      mpz_probab_prime_p(n,r) r=1 = 72 secs.
      mpz_probab_prime_p(n,r) r=0 = 34 secs.
      fermat (mpz_powm(2,n,n) = 2^n MOD n = 34 secs.
      My implementation Fermat algorytm = 33 secs.
      ===================================================================
      PFGW
      pfgw 3.2.2 (On AMD 1800 Mhz) = 3.1 secs.
      pfgw 3.2.2 (On INTEL P4 1800 Mhz) = 1.7 secs. <-----
      ===================================================================
      MATHEMATICA
      Mathematica PowerMod[2,n,n] (Fermat) = 38 secs.
      PrimeQ[n] = 157 secs.
      ===================================================================
      PARI/GP
      ispseudoprime(n) BPSW test = 130 secs.
      ispseudoprime(n,1) strong Rabin-Miller test for 1 base = 43 secs.
      powermod(x,k,m)=lift(Mod(x,m)^k) --> powermod(2,n,n) = 42 secs.
      isprime(n) = too long
      ===================================================================
      PROTH.EXE
      Normal Test = 5 secs.
      ===================================================================
      LLR.EXE
      Normal Test = 7 secs.
      ===================================================================
      PRP.EXE
      Normal Test = 7 secs.
      ===================================================================
      PrimeForm.EXE
      Normal Test = 24 secs.
      ===================================================================
    • cino hilliard
      Try (00:25:59) gp ispseudoprime(31838235*2^29717+1) %129 = 1 (00:33:39) gp ## *** last result computed in 6mn, 29,015 ms. (17:33:32) gp
      Message 2 of 5 , Sep 1, 2009
      • 0 Attachment
        Try
        (00:25:59) gp > ispseudoprime(31838235*2^29717+1)
        %129 = 1
        (00:33:39) gp > ##
        *** last result computed in 6mn, 29,015 ms.





        (17:33:32) gp > isprime(31838235*2^29717+1)
        %74 = 1
        (18:35:16) gp > ##
        *** last result computed in 51mn, 53,938 ms.



        c:\pfgw>pfgw -tc -q31838235*2^^29717+1
        PFGW Version 20090725.Win_Dev (Beta 'caveat utilitor') [GWNUM 25.12]

        Primality testing 31838235*2^29717+1 [N-1/N+1, Brillhart-Lehmer-Selfridge]
        Running N-1 test using base 17
        Running N-1 test using base 23
        Running N+1 test using discriminant 31, base 10+sqrt(31)
        Calling N-1 BLS with factored part 100.00% and helper 0.07% (300.08% proof)
        31838235*2^29717+1 is prime! (15.3991s+0.0311s)



        This 8954 digit number will be the top primo candidate if you have the time

        So before you do primo, do pfgw -tc



        pfgw ROCKS!



        Cino







        To: primenumbers@yahoogroups.com
        From: calimero22@...
        Date: Mon, 31 Aug 2009 20:05:01 +0000
        Subject: [PrimeNumbers] Some Benchmark Primality Test






        I post some time test for a primality or pseudoprimality check.

        Tests on AMD Sempron 3000+ (1800 Mhz) WIndows XP
        n=31838235*2^29717+1 digits: 8954
        Tests run by GIOVANNI DI MARIA - email: calimero22@...
        Tests give at least a Pseudoprimality PRP.

        ===================================================================
        GMP Library
        mpz_probab_prime_p(n,r) r=1 = 72 secs.
        mpz_probab_prime_p(n,r) r=0 = 34 secs.
        fermat (mpz_powm(2,n,n) = 2^n MOD n = 34 secs.
        My implementation Fermat algorytm = 33 secs.
        ===================================================================
        PFGW
        pfgw 3.2.2 (On AMD 1800 Mhz) = 3.1 secs.
        pfgw 3.2.2 (On INTEL P4 1800 Mhz) = 1.7 secs. <-----
        ===================================================================
        MATHEMATICA
        Mathematica PowerMod[2,n,n] (Fermat) = 38 secs.
        PrimeQ[n] = 157 secs.
        ===================================================================
        PARI/GP
        ispseudoprime(n) BPSW test = 130 secs.
        ispseudoprime(n,1) strong Rabin-Miller test for 1 base = 43 secs.
        powermod(x,k,m)=lift(Mod(x,m)^k) --> powermod(2,n,n) = 42 secs.
        isprime(n) = too long
        ===================================================================
        PROTH.EXE
        Normal Test = 5 secs.
        ===================================================================
        LLR.EXE
        Normal Test = 7 secs.
        ===================================================================
        PRP.EXE
        Normal Test = 7 secs.
        ===================================================================
        PrimeForm.EXE
        Normal Test = 24 secs.
        ===================================================================










        [Non-text portions of this message have been removed]
      • djbroadhurst
        ... OK ... Using ECPP to re-prove a prime already proven by BLS would be an exercise in fatuity. David
        Message 3 of 5 , Sep 1, 2009
        • 0 Attachment
          --- In primenumbers@yahoogroups.com,
          cino hilliard <hillcino368@...> wrote:

          > 31838235*2^29717+1 is prime! (15.3991s+0.0311s)

          OK

          > This 8954 digit number will be the top primo candidate
          > if you have the time

          Using ECPP to re-prove a prime already proven by BLS
          would be an exercise in fatuity.

          David
        • cino hilliard
          Hi David, ... anayway, I thought we were benchmarking in this last post? I am still curious how long it would take primo to prove 31838235*2^29717+1 is prime.
          Message 4 of 5 , Sep 1, 2009
          • 0 Attachment
            Hi David,



            >So before you do primo, do pfgw -tc


            anayway, I thought we were benchmarking in this last post?



            I am still curious how long it would take primo to prove 31838235*2^29717+1 is prime.

            At fiirst, I thought time was proportional to the number of digits but now I realize
            that ain't so. "When in doubt, specialize"



            Not all lost though. I learned a new word.



            I tried pfgw on the primo top entry of the top 20.



            http://www.ellipsa.eu/public/primo/top20.html



            c:\pfgw>pfgw -tc -q2^^29727+20273
            PFGW Version 20090725.Win_Dev (Beta 'caveat utilitor') [GWNUM 25.12]

            Primality testing 2^29727+20273 [N-1/N+1, Brillhart-Lehmer-Selfridge]
            Running N-1 test using base 11
            Running N+1 test using discriminant 19, base 9+sqrt(19)
            Calling N+1 BLS with factored part 0.08% and helper 0.07% (0.30% proof)
            2^29727+20273 is Fermat and Lucas PRP! (12.4107s+0.0008s)



            Interesting.

            What dominion does 31838235*2^29717+1 have over 1*2^29727+20273?

            Is 80 days the best that can be done for the top primo prime?


            http://primes.utm.edu/primes/page.php?id=89447



            pfgw still ROCKS!


            Cino




            To: primenumbers@yahoogroups.com
            From: d.broadhurst@...
            Date: Tue, 1 Sep 2009 22:24:24 +0000
            Subject: [PrimeNumbers] Re: Some Benchmark Primality Test





            --- In primenumbers@yahoogroups.com,
            cino hilliard <hillcino368@...> wrote:

            > 31838235*2^29717+1 is prime! (15.3991s+0.0311s)

            OK

            > This 8954 digit number will be the top primo candidate
            > if you have the time

            Using ECPP to re-prove a prime already proven by BLS
            would be an exercise in fatuity.

            David










            [Non-text portions of this message have been removed]
          • djbroadhurst
            ... Very roughly, I reckon that the Primo time is proportional to digits^(9/2). David
            Message 5 of 5 , Sep 1, 2009
            • 0 Attachment
              --- In primenumbers@yahoogroups.com,
              cino hilliard <hillcino368@...> wrote:

              > At first, I thought time was proportional
              > to the number of digits but now I realize
              > that ain't so.

              Very roughly, I reckon that the Primo time is proportional to digits^(9/2).

              David
            Your message has been successfully submitted and would be delivered to recipients shortly.