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

polynomial selection for 155 digits

Expand Messages
  • Guirgis Sameh
    Hi I have two question about polynomial selection for 155 digits 1- is it possible to run polyselect in parallel. for example, on core 2 quad? how? 2-what are
    Message 1 of 25 , Feb 3, 2010
    • 0 Attachment
      Hi

      I have two question about polynomial selection for 155 digits
      1- is it possible to run  polyselect in parallel. for example, on core 2 quad? how?

      2-what are the expectation  (or near optimal) values of the parameters maxs1, maxskew, enum, e0, 1, j0, j1   to run polyselect?


      thanks alot.

      Guirgis


    • Jeff Gilchrist
      ... I m not familiar with the exact command line but yes you can run this in parallel by specifying a different range for each core to search. Do you have
      Message 2 of 25 , Feb 5, 2010
      • 0 Attachment
        On Wed, Feb 3, 2010 at 6:35 PM, Guirgis Sameh <guirgis.sameh@...> wrote:

        > I have two question about polynomial selection for 155 digits
        > 1- is it possible to run  polyselect in parallel. for example, on core 2 quad? how?

        I'm not familiar with the exact command line but yes you can run this
        in parallel by specifying a different range for each core to search.

        Do you have access to a machine with an Nvidia graphics card that
        supports running CUDA applications (one made in the past couple of
        years)? If yes, there is a new release of msieve
        (http://sourceforge.net/projects/msieve/) v1.44 that was just put out
        yesterday that supports doing polynomial selection (155 digits is
        supported) using GPUs that is magnitudes faster than using CPUs. If
        you can get access to any computer with such a video card, running a
        single instance on that will take significantly less time than using a
        quad core.

        Jeff.
      • ali maher
        i have an error when try to factor this no.
        Message 3 of 25 , Feb 6, 2010
        • 0 Attachment
          i have an error when try to factor this no.

          231375786962718406685740634517908336011157041513365446768177743893598925006969224030324622854605578754606929728233762180972304401742933872804034141944924829



          c:\ggnfs\gad>factmsieve.py h1.n
          -> ________________________________________________________________
          -> | Running factmsieve.py, a Python driver for MSIEVE with GGNFS |
          -> | sieving support. It is Copyright, 2010, Brian Gladman and is |
          -> | a conversion of factmsieve.pl that is Copyright, 2004, Chris |
          -> | Monico.          This is version 0.2, dated 13 January 2010. |
          -> |______________________________________________________________|
          -> This is client 1 of 1
          -> Using 2 threads
          -> Working with NAME = h1
          -> Error: Polynomial file h1.poly does not exist!
          Traceback (most recent call last):
            File "C:\ggnfs\gad\factmsieve.py", line 1885, in <module>
              m = re.search('n:\s*(\d+)', TMP[0])
          IndexError: list index out of range

          c:\ggnfs\gad>



        • br_gladman
          ... You are using an outdated version of factmsieve.py - the latest version is here: http://gladman.plushost.co.uk/oldsite/computing/factmsieve.py.zip Brian
          Message 4 of 25 , Feb 7, 2010
          • 0 Attachment
            --- In ggnfs@yahoogroups.com, ali maher <gad_12006@...> wrote:
            >
            > i have an error when try to factor this no.
            >
            > 231375786962718406685740634517908336011157041513365446768177743893598925006969224030324622854605578754606929728233762180972304401742933872804034141944924829
            >
            >
            >
            > c:\ggnfs\gad>factmsieve.py h1.n
            > -> ________________________________________________________________
            > -> | Running factmsieve.py, a Python driver for MSIEVE with GGNFS |
            > -> | sieving support. It is Copyright, 2010, Brian Gladman and is |
            > -> | a conversion of factmsieve.pl that is Copyright, 2004, Chris |
            > -> | Monico. This is version 0.2, dated 13 January 2010. |
            > -> |______________________________________________________________|
            > -> This is client 1 of 1
            > -> Using 2 threads
            > -> Working with NAME = h1
            > -> Error: Polynomial file h1.poly does not exist!
            > Traceback (most recent call last):
            > File "C:\ggnfs\gad\factmsieve.py", line 1885, in <module>
            > m = re.search('n:\s*(\d+)', TMP[0])
            > IndexError: list index out of range
            >
            You are using an outdated version of factmsieve.py - the latest version is here:

            http://gladman.plushost.co.uk/oldsite/computing/factmsieve.py.zip

            Brian Gladman
          • ali maher
            i have an error when try to use the new ver. of python     Microsoft Windows [Version 6.0.6000] Copyright (c) 2006 Microsoft Corporation.  All rights
            Message 5 of 25 , Feb 9, 2010
            • 0 Attachment

              i have an error when try to use the new ver. of python

               

               

              Microsoft Windows [Version 6.0.6000]
              Copyright (c) 2006 Microsoft Corporation.  All rights reserved.

              C:\Users\gad>cd c:\ggnfs\error\

              c:\ggnfs\error>factmsieve.py h.n
              -> ________________________________________________________________
              -> | Running factmsieve.py, a Python driver for MSIEVE with GGNFS |
              -> | sieving support. It is Copyright, 2010, Brian Gladman and is |
              -> | a conversion of factmsieve.pl that is Copyright, 2004, Chris |
              -> | Monico.       This is version 0.51, dated 8th February 2010. |
              -> |______________________________________________________________|
              -> This is client 1 of 1
              -> Using 2 threads
              -> Working with NAME = h
              -> Error: Polynomial file h.poly does not exist!
              -> Found n = 1534280224047781566463738853176008890515146232331206132188773536740
              377701919806749002484133391354053.
              -> Running polynomial selection ...
              -> msieve -s h.dat -l h.log -i h.ini -nf h.fb -v -np
              Traceback (most recent call last):
                File "C:\ggnfs\error\factmsieve.py", line 1980, in <module>
                  run_msieve_poly(fact_p)
                File "C:\ggnfs\error\factmsieve.py", line 883, in run_msieve_poly
                  ret = run_exe(MSIEVE, args)
                File "C:\ggnfs\error\factmsieve.py", line 297, in run_exe
                  p = subprocess.Popen([ex] + args.split(' '), **al)
                File "C:\Python26\lib\subprocess.py", line 621, in __init__
                  errread, errwrite)
                File "C:\Python26\lib\subprocess.py", line 830, in _execute_child
                  startupinfo)
              WindowsError: [Error 2] The system cannot find the file specified
              siever terminated

              c:\ggnfs\error>


            • Jeff Gilchrist
              ... You probably forgot to edit the new factmsieve.py file to set your paths for where msieve.exe and the ggnfs binaries are located. Jeff.
              Message 6 of 25 , Feb 9, 2010
              • 0 Attachment
                On Tue, Feb 9, 2010 at 11:43 AM, ali maher <gad_12006@...> wrote:

                > i have an error when try to use the new ver. of python
                >
                > c:\ggnfs\error>factmsieve.py h.n
                >
                > WindowsError: [Error 2] The system cannot find the file specified
                > siever terminated

                You probably forgot to edit the new factmsieve.py file to set your
                paths for where msieve.exe and the ggnfs binaries are located.

                Jeff.
              • ali maher
                i edit the new factmsieve.pyfile  with  msieve.exe and the ggnfs binaries are located. but the same error appear agian.      Microsoft Windows [Version
                Message 7 of 25 , Feb 9, 2010
                • 0 Attachment

                  i edit the new factmsieve.py file  with  msieve.exe and the ggnfs binaries are located.

                  but the same error appear agian.

                   

                   

                   

                  Microsoft Windows [Version 6.0.6000]
                  Copyright (c) 2006 Microsoft Corporation.  All rights reserved.

                  C:\Users\gad>cd c:\ggnfs\error\

                  c:\ggnfs\error>factmsieve.py h.n
                  -> ____________ _________ _________ _________ _________ _________ _______
                  -> | Running factmsieve.py, a Python driver for MSIEVE with GGNFS |
                  -> | sieving support. It is Copyright, 2010, Brian Gladman and is |
                  -> | a conversion of factmsieve.pl that is Copyright, 2004, Chris |
                  -> | Monico.       This is version 0.51, dated 8th February 2010. |
                  -> |___________ _________ _________ _________ _________ _________ ______|
                  -> This is client 1 of 1
                  -> Using 2 threads
                  -> Working with NAME = h
                  -> Error: Polynomial file h.poly does not exist!
                  -> Found n = 1534280224047781566 4637388531760088 9051514623233120 6132188773536740
                  3777019198067490024 84133391354053.
                  -> Running polynomial selection ...
                  -> msieve -s h.dat -l h.log -i h.ini -nf h.fb -v -np
                  Traceback (most recent call last):
                    File "C:\ggnfs\error\ factmsieve. py", line 1980, in <module>
                      run_msieve_poly( fact_p)
                    File "C:\ggnfs\error\ factmsieve. py", line 883, in run_msieve_poly
                      ret = run_exe(MSIEVE, args)
                    File "C:\ggnfs\error\ factmsieve. py", line 297, in run_exe
                      p = subprocess.Popen( [ex] + args.split(' '), **al)
                    File "C:\Python26\ lib\subprocess. py", line 621, in __init__
                      errread, errwrite)
                    File "C:\Python26\ lib\subprocess. py", line 830, in _execute_child
                      startupinfo)
                  WindowsError: [Error 2] The system cannot find the file specified
                  siever terminated


                • Jeff Gilchrist
                  ... Instead of just saying you edited the file, can you provide us with your actual settings so we can better help you. What do you have for these lines in
                  Message 8 of 25 , Feb 10, 2010
                  • 0 Attachment
                    On Tue, Feb 9, 2010 at 5:14 PM, ali maher <gad_12006@...> wrote:

                    > i edit the new factmsieve.py file  with  msieve.exe and the ggnfs binaries are located.
                    > but the same error appear agian.

                    Instead of just saying you edited the file, can you provide us with
                    your actual settings so we can better help you. What do you have for
                    these lines in factmsieve.py

                    # Set binary directory paths

                    GGNFS_PATH = '../'
                    MSIEVE_PATH = '../'

                    Jeff.
                  • br_gladman
                    ... I see that you are running in the working directory: C: ggnfs error Could you tell ue where your ggnfs and msieve binaries are? Could you also show us the
                    Message 9 of 25 , Feb 10, 2010
                    • 0 Attachment
                      --- In ggnfs@yahoogroups.com, ali maher <gad_12006@...> wrote:
                      >
                      > i edit the new factmsieve.pyfile  with  msieve.exe and the ggnfs binaries are located.
                      > but the same error appear agian.
                      >
                      >  
                      >  
                      >  Microsoft Windows [Version 6.0.6000]
                      > Copyright (c) 2006 Microsoft Corporation.  All rights reserved.
                      > C:\Users\gad>cd c:\ggnfs\error\
                      > c:\ggnfs\error>factmsieve.py h.n
                      > -> ____________ _________ _________ _________ _________ _________ _______
                      > -> | Running factmsieve.py, a Python driver for MSIEVE with GGNFS |
                      > -> | sieving support. It is Copyright, 2010, Brian Gladman and is |
                      > -> | a conversion of factmsieve.pl that is Copyright, 2004, Chris |
                      > -> | Monico.       This is version 0.51, dated 8th February 2010. |
                      > -> |___________ _________ _________ _________ _________ _________ ______|
                      > -> This is client 1 of 1
                      > -> Using 2 threads
                      > -> Working with NAME = h
                      > -> Error: Polynomial file h.poly does not exist!
                      > -> Found n = 1534280224047781566 4637388531760088 9051514623233120 6132188773536740
                      > 3777019198067490024 84133391354053.
                      > -> Running polynomial selection ...
                      > -> msieve -s h.dat -l h.log -i h.ini -nf h.fb -v -np
                      > Traceback (most recent call last):
                      >   File "C:\ggnfs\error\ factmsieve. py", line 1980, in <module>
                      >     run_msieve_poly( fact_p)
                      >   File "C:\ggnfs\error\ factmsieve. py", line 883, in run_msieve_poly
                      >     ret = run_exe(MSIEVE, args)
                      >   File "C:\ggnfs\error\ factmsieve. py", line 297, in run_exe
                      >     p = subprocess.Popen( [ex] + args.split(' '), **al)
                      >   File "C:\Python26\ lib\subprocess. py", line 621, in __init__
                      >     errread, errwrite)
                      >   File "C:\Python26\ lib\subprocess. py", line 830, in _execute_child
                      >     startupinfo)
                      > WindowsError: [Error 2] The system cannot find the file specified
                      > siever terminated

                      I see that you are running in the working directory:

                      C:\ggnfs\error\

                      Could you tell ue where your ggnfs and msieve binaries are?

                      Could you also show us the lines with:

                      GGNFS_PATH =
                      MSIEVE_PATH =

                      in your C:\ggnfs\error\factmsieve.py script?

                      Brian
                    • ali maher
                      thank you the problem is solved.
                      Message 10 of 25 , Feb 10, 2010
                      • 0 Attachment

                        thank you the problem is solved.

                         


                      • Orkblutt
                        Hi guys, very longtime I didn t follow the optimisations made on ggnfs. I m totaly outdated... It looks there are many improvment since the last time I
                        Message 11 of 25 , Feb 19, 2010
                        • 0 Attachment
                          Hi guys,
                           
                          very longtime I didn't follow the optimisations made on ggnfs. I'm totaly outdated... It looks there are many improvment since the last time I checked.
                           
                          I seen a great news some days ago: the msieve v1.44 using CUDA!
                          I downloaded it but I didn't take the time to test it.
                           
                          Also downloaded the B.Gladman's python driver.
                           
                           
                          Weekend starting so time to test... but my first one didn't go very far:
                           
                          D:\ggnfs\exemple>..\factmsieve.py example.n
                          -> ________________________________________________________________
                          -> | Running factmsieve.py, a Python driver for MSIEVE with GGNFS |
                          -> | sieving support. It is Copyright, 2010, Brian Gladman and is |
                          -> | a conversion of factmsieve.pl that is Copyright, 2004, Chris |
                          -> | Monico.       This is version 0.52, dated 9th February 2010. |
                          -> |______________________________________________________________|
                          -> This is client 1 of 1
                          -> Using 2 threads
                          -> Working with NAME = example
                          -> Error: Polynomial file example.poly does not exist!
                          -> Found n = 2881039827457895971881627053137530734638790825166127496066674320241571446494762386620442953820735453.
                          -> Running polynomial selection ...
                          -> msieve -s example.dat -l example.log -i example.ini -nf example.fb -v -np
                           

                          Msieve v. 1.44
                          Sat Feb 20 00:36:43 2010
                          random seeds: 3f66fd4c ba647312
                          factoring 2881039827457895971881627053137530734638790825166127496066674320241571446494762386620442953820735453 (100 digits)
                          searching for 15-digit factors
                          commencing number field sieve (100-digit input)
                          commencing number field sieve polynomial selection
                          time limit set to 0.35 hours
                          searching leading coefficients from 1 to 6474027
                          using GPU 0 (GeForce GTX 295)
                          error (line 184): CUDA_ERROR_FILE_NOT_FOUND
                          Msieve Error: return value -1. Terminating...
                          siever terminated
                           
                           
                          Is this error because factmsieve.py doesn't support yet msieve using GPU?
                           
                          Thanks for your help.
                           
                          Orkblutt

                          Sent: Friday, February 05, 2010 5:26 PM
                          Subject: Re: [ggnfs] polynomial selection for 155 digits

                           

                          On Wed, Feb 3, 2010 at 6:35 PM, Guirgis Sameh <guirgis.sameh@ yahoo.com> wrote:

                          > I have two question about polynomial selection for 155
                          digits
                          > 1- is it possible to run  polyselect in parallel. for
                          example, on core 2 quad? how?

                          I'm not familiar with the exact command line but yes you can run this
                          in parallel by specifying a different range for each core to search.

                          Do you have access to a machine with an Nvidia graphics card that
                          supports running CUDA applications (one made in the past couple of
                          years)? If yes, there is a new release of msieve
                          (http://sourceforge. net/projects/ msieve/) v1.44 that was just put out
                          yesterday that supports doing polynomial selection (155 digits is
                          supported) using GPUs that is magnitudes faster than using CPUs. If
                          you can get access to any computer with such a video card, running a
                          single instance on that will take significantly less time than using a
                          quad core.

                          Jeff.

                        • Jason Papadopoulos
                          ... The msieve GPU binary comes with a number of .ptx files; those files have to be in the same directory as the msieve binary for the GPU code to work. jasonp
                          Message 12 of 25 , Feb 19, 2010
                          • 0 Attachment
                            Quoting Orkblutt <orkblutt@...>:

                            > Is this error because factmsieve.py doesn't support yet msieve using GPU?
                            >

                            The msieve GPU binary comes with a number of .ptx files; those files
                            have to be in the same directory as the msieve binary for the GPU code
                            to work.

                            jasonp

                            ------------------------------------------------------
                            This message was sent using BOO.net's Webmail.
                            http://www.boo.net/
                          • Orkblutt
                            Yes I have the 6 .ptx file at the same level than msieve.exe... But same result. All the best, Orkblutt From: Jason Papadopoulos Sent: Saturday, February 20,
                            Message 13 of 25 , Feb 20, 2010
                            • 0 Attachment
                              Yes I have the 6 .ptx file at the same level than msieve.exe... But same result.
                               
                              All the best,
                               
                              Orkblutt

                              Sent: Saturday, February 20, 2010 1:33 AM
                              Subject: Re: [ggnfs] factmsieve.py & CUDA's msieve

                               

                              Quoting Orkblutt <orkblutt@msn. com>:

                              > Is
                              this error because factmsieve.py doesn't support yet msieve using GPU?
                              >

                              The msieve GPU binary comes with a number of .ptx files; those files
                              have to be in the same directory as the msieve binary for the GPU code
                              to work.

                              jasonp

                              ------------ --------- --------- --------- --------- ------
                              This message was sent using BOO.net's Webmail.
                              http://www.boo. net/

                            • br_gladman
                              ... I think the *.ptx files have to be in the current working directory, which is not necessarily the same directory as msieve is in. Since numbers are
                              Message 14 of 25 , Feb 20, 2010
                              • 0 Attachment
                                --- In ggnfs@yahoogroups.com, "Orkblutt" <orkblutt@...> wrote:
                                >
                                > Yes I have the 6 .ptx file at the same level than msieve.exe... But same result.
                                >
                                > All the best,
                                >
                                > Orkblutt
                                >
                                >
                                > From: Jason Papadopoulos
                                > Sent: Saturday, February 20, 2010 1:33 AM
                                > To: ggnfs@yahoogroups.com
                                > Subject: Re: [ggnfs] factmsieve.py & CUDA's msieve
                                >
                                >
                                >
                                > Quoting Orkblutt <orkblutt@...>:
                                >
                                > > Is this error because factmsieve.py doesn't support yet msieve using GPU?
                                > >
                                >
                                > The msieve GPU binary comes with a number of .ptx files; those files
                                > have to be in the same directory as the msieve binary for the GPU code
                                > to work.
                                >
                                > jasonp
                                >
                                > ------------------------------------------------------
                                > This message was sent using BOO.net's Webmail.
                                > http://www.boo.net/
                                >
                                I think the *.ptx files have to be in the current working directory, which is not necessarily the same directory as msieve is in. Since numbers are normally run from a directory that is used to hold the files generated by the script, the *.ptx files may need to be here.

                                I hope to find a way of running msieve in its own directory while it writes files to the 'normal' working directory by manipulating the various paths used in factmsieve.py.

                                Brian
                              • Orkblutt
                                Thanks Brian ! It works like a charm... Kindest regards, Orkblutt From: br_gladman Sent: Saturday, February 20, 2010 10:24 AM To: ggnfs@yahoogroups.com
                                Message 15 of 25 , Feb 20, 2010
                                • 0 Attachment
                                  Thanks Brian !
                                  It works like a charm...
                                   
                                  Kindest regards,
                                   
                                  Orkblutt

                                  Sent: Saturday, February 20, 2010 10:24 AM
                                  Subject: [ggnfs] Re: factmsieve.py & CUDA's msieve

                                   



                                  --- In ggnfs@yahoogroups. com, "Orkblutt" <orkblutt@.. .> wrote:

                                  >
                                  > Yes I have the 6 .ptx file at
                                  the same level than msieve.exe.. . But same result.
                                  >
                                  > All the
                                  best,
                                  >
                                  > Orkblutt
                                  >
                                  >
                                  > From: Jason
                                  Papadopoulos
                                  > Sent: Saturday, February 20, 2010 1:33 AM
                                  > To:
                                  href="mailto:ggnfs@yahoogroups.com">ggnfs@yahoogroups. com
                                  >
                                  Subject: Re: [ggnfs] factmsieve.py & CUDA's msieve
                                  >
                                  >
                                  >
                                  > Quoting Orkblutt <orkblutt@.. .>:
                                  >
                                  > > Is
                                  this error because factmsieve.py doesn't support yet msieve using GPU?
                                  > >
                                  >
                                  > The msieve GPU binary comes with a number of .ptx files;
                                  those files
                                  > have to be in the same directory as the msieve binary for
                                  the GPU code
                                  > to work.
                                  >
                                  > jasonp
                                  >
                                  >
                                  ------------ --------- --------- --------- --------- ------
                                  >
                                  This message was sent using BOO.net's Webmail.
                                  >
                                  href="http://www.boo.net/">http://www.boo. net/
                                  >
                                  I think the *.ptx files have to be in the current working directory, which is not necessarily the same directory as msieve is in. Since numbers are normally run from a directory that is used to hold the files generated by the script, the *.ptx files may need to be here.

                                  I hope to find a way of running msieve in its own directory while it writes files to the 'normal' working directory by manipulating the various paths used in factmsieve.py.

                                  Brian

                                • Orkblutt
                                  I noticed msieve for GPU can choose the working GPU (-g switch). The python (and perl) script should be updated to take care about this parameter. The
                                  Message 16 of 25 , Feb 20, 2010
                                  • 0 Attachment
                                    I noticed msieve for GPU can choose the working GPU (-g switch). The python (and perl) script should be updated to take care about this parameter.
                                     
                                    The polynomial selection step took more time I was thinking:
                                     
                                    Sat Feb 20 11:38:39 2010  Msieve v. 1.44
                                    Sat Feb 20 11:38:39 2010  random seeds: 1f2f4d0c 5dda8154
                                    Sat Feb 20 11:38:39 2010  factoring 2881039827457895971881627053137530734638790825166127496066674320241571446494762386620442953820735453 (100 digits)
                                    Sat Feb 20 11:38:40 2010  searching for 15-digit factors
                                    Sat Feb 20 11:38:40 2010  commencing number field sieve (100-digit input)
                                    Sat Feb 20 11:38:40 2010  commencing number field sieve polynomial selection
                                    Sat Feb 20 11:38:40 2010  time limit set to 0.35 hours
                                    Sat Feb 20 11:38:40 2010  searching leading coefficients from 1 to 6474027
                                    Sat Feb 20 11:38:41 2010  using GPU 1 (GeForce GTX 295)
                                    Sat Feb 20 12:22:53 2010  polynomial selection complete
                                    Sat Feb 20 12:22:53 2010  R0: -1277996108135008554742004
                                    Sat Feb 20 12:22:53 2010  R1:  90597371296321
                                    Sat Feb 20 12:22:53 2010  A0:  27241995657721635205536043245
                                    Sat Feb 20 12:22:53 2010  A1: -1221351916941001527056
                                    Sat Feb 20 12:22:53 2010  A2: -33768914437945541
                                    Sat Feb 20 12:22:53 2010  A3:  261332752
                                    Sat Feb 20 12:22:53 2010  A4:  1080
                                    Sat Feb 20 12:22:53 2010  skew 3406178.83, size 1.331082e-013, alpha -5.711799, combined = 1.375442e-008
                                    Sat Feb 20 12:22:53 2010  elapsed time 00:44:14
                                     
                                     
                                    I have to test with CPU but I think there is a problem here... what do you think?
                                     
                                    Regards,
                                    Orkblutt

                                    Sent: Saturday, February 20, 2010 10:24 AM
                                    Subject: [ggnfs] Re: factmsieve.py & CUDA's msieve

                                     



                                    --- In ggnfs@yahoogroups. com, "Orkblutt" <orkblutt@.. .> wrote:

                                    >
                                    > Yes I have the 6 .ptx file at
                                    the same level than msieve.exe.. . But same result.
                                    >
                                    > All the
                                    best,
                                    >
                                    > Orkblutt
                                    >
                                    >
                                    > From: Jason
                                    Papadopoulos
                                    > Sent: Saturday, February 20, 2010 1:33 AM
                                    > To:
                                    href="mailto:ggnfs@yahoogroups.com">ggnfs@yahoogroups. com
                                    >
                                    Subject: Re: [ggnfs] factmsieve.py & CUDA's msieve
                                    >
                                    >
                                    >
                                    > Quoting Orkblutt <orkblutt@.. .>:
                                    >
                                    > > Is
                                    this error because factmsieve.py doesn't support yet msieve using GPU?
                                    > >
                                    >
                                    > The msieve GPU binary comes with a number of .ptx files;
                                    those files
                                    > have to be in the same directory as the msieve binary for
                                    the GPU code
                                    > to work.
                                    >
                                    > jasonp
                                    >
                                    >
                                    ------------ --------- --------- --------- --------- ------
                                    >
                                    This message was sent using BOO.net's Webmail.
                                    >
                                    href="http://www.boo.net/">http://www.boo. net/
                                    >
                                    I think the *.ptx files have to be in the current working directory, which is not necessarily the same directory as msieve is in. Since numbers are normally run from a directory that is used to hold the files generated by the script, the *.ptx files may need to be here.

                                    I hope to find a way of running msieve in its own directory while it writes files to the 'normal' working directory by manipulating the various paths used in factmsieve.py.

                                    Brian

                                  • br_gladman
                                    ... I have just updated thye Python version here: http://gladman.plushost.co.uk/oldsite/computing/factmsieve.py.zip It shoulkd now be able to run msieve
                                    Message 17 of 25 , Feb 20, 2010
                                    • 0 Attachment
                                      --- In ggnfs@yahoogroups.com, "Orkblutt" <orkblutt@...> wrote:
                                      >
                                      > I noticed msieve for GPU can choose the working GPU (-g switch). The python (and perl) script should be updated to take care about this parameter.
                                      >
                                      > The polynomial selection step took more time I was thinking:
                                      >
                                      > Sat Feb 20 11:38:39 2010 Msieve v. 1.44
                                      > Sat Feb 20 11:38:39 2010 random seeds: 1f2f4d0c 5dda8154
                                      > Sat Feb 20 11:38:39 2010 factoring 2881039827457895971881627053137530734638790825166127496066674320241571446494762386620442953820735453 (100 digits)
                                      > Sat Feb 20 11:38:40 2010 searching for 15-digit factors
                                      > Sat Feb 20 11:38:40 2010 commencing number field sieve (100-digit input)
                                      > Sat Feb 20 11:38:40 2010 commencing number field sieve polynomial selection
                                      > Sat Feb 20 11:38:40 2010 time limit set to 0.35 hours
                                      > Sat Feb 20 11:38:40 2010 searching leading coefficients from 1 to 6474027
                                      > Sat Feb 20 11:38:41 2010 using GPU 1 (GeForce GTX 295)
                                      > Sat Feb 20 12:22:53 2010 polynomial selection complete
                                      > Sat Feb 20 12:22:53 2010 R0: -1277996108135008554742004
                                      > Sat Feb 20 12:22:53 2010 R1: 90597371296321
                                      > Sat Feb 20 12:22:53 2010 A0: 27241995657721635205536043245
                                      > Sat Feb 20 12:22:53 2010 A1: -1221351916941001527056
                                      > Sat Feb 20 12:22:53 2010 A2: -33768914437945541
                                      > Sat Feb 20 12:22:53 2010 A3: 261332752
                                      > Sat Feb 20 12:22:53 2010 A4: 1080
                                      > Sat Feb 20 12:22:53 2010 skew 3406178.83, size 1.331082e-013, alpha -5.711799, combined = 1.375442e-008
                                      > Sat Feb 20 12:22:53 2010 elapsed time 00:44:14
                                      >
                                      >
                                      > I have to test with CPU but I think there is a problem here... what do you think?
                                      >
                                      > Regards,
                                      > Orkblutt
                                      >
                                      >
                                      > From: br_gladman
                                      > Sent: Saturday, February 20, 2010 10:24 AM
                                      > To: ggnfs@yahoogroups.com
                                      > Subject: [ggnfs] Re: factmsieve.py & CUDA's msieve
                                      >
                                      >
                                      >
                                      >
                                      >
                                      > --- In ggnfs@yahoogroups.com, "Orkblutt" <orkblutt@> wrote:
                                      > >
                                      > > Yes I have the 6 .ptx file at the same level than msieve.exe... But same result.
                                      > >
                                      > > All the best,
                                      > >
                                      > > Orkblutt
                                      > >
                                      > >
                                      > > From: Jason Papadopoulos
                                      > > Sent: Saturday, February 20, 2010 1:33 AM
                                      > > To: ggnfs@yahoogroups.com
                                      > > Subject: Re: [ggnfs] factmsieve.py & CUDA's msieve
                                      > >
                                      > >
                                      > >
                                      > > Quoting Orkblutt <orkblutt@>:
                                      > >
                                      > > > Is this error because factmsieve.py doesn't support yet msieve using GPU?
                                      > > >
                                      > >
                                      > > The msieve GPU binary comes with a number of .ptx files; those files
                                      > > have to be in the same directory as the msieve binary for the GPU code
                                      > > to work.
                                      > >
                                      > > jasonp
                                      > >
                                      > > ------------------------------------------------------
                                      > > This message was sent using BOO.net's Webmail.
                                      > > http://www.boo.net/
                                      > >
                                      > I think the *.ptx files have to be in the current working directory, which is not necessarily the same directory as msieve is in. Since numbers are normally run from a directory that is used to hold the files generated by the script, the *.ptx files may need to be here.
                                      >
                                      > I hope to find a way of running msieve in its own directory while it writes files to the 'normal' working directory by manipulating the various paths used in factmsieve.py.
                                      >
                                      > Brian

                                      I have just updated thye Python version here:

                                      http://gladman.plushost.co.uk/oldsite/computing/factmsieve.py.zip

                                      It shoulkd now be able to run msieve without moving the *.ptx files (you need to put them back in the msieve directory of course).

                                      I have also added a USE_CUDA and a GPU_NUM define.

                                      Do remember to set these, other settings and the paths to your executables in this script before using it.

                                      I would be interested to know if it works.

                                      Brian
                                    • Orkblutt
                                      Are you sure the .py is updated on your website? I still have the version dated of 09/02. regards, Orkblutt From: br_gladman Sent: Saturday, February 20, 2010
                                      Message 18 of 25 , Feb 20, 2010
                                      • 0 Attachment
                                        Are you sure the .py is updated on your website?
                                        I still have the version dated of 09/02.
                                         
                                        regards,
                                         
                                        Orkblutt

                                        Sent: Saturday, February 20, 2010 2:17 PM
                                        Subject: [ggnfs] Re: factmsieve.py & CUDA's msieve

                                         



                                        --- In ggnfs@yahoogroups. com, "Orkblutt" <orkblutt@.. .> wrote:

                                        >
                                        > I noticed msieve for GPU can
                                        choose the working GPU (-g switch). The python (and perl) script should be updated to take care about this parameter.
                                        >
                                        > The polynomial
                                        selection step took more time I was thinking:
                                        >
                                        > Sat Feb 20
                                        11:38:39 2010 Msieve v. 1.44
                                        > Sat Feb 20 11:38:39 2010 random seeds:
                                        1f2f4d0c 5dda8154
                                        > Sat Feb 20 11:38:39 2010 factoring
                                        2881039827457895971 8816270531375307 3463879082516612 7496066674320241 5714464947623866 2044295382073545 3 (100 digits)
                                        > Sat Feb 20 11:38:40 2010 searching for 15-digit
                                        factors
                                        > Sat Feb 20 11:38:40 2010 commencing number field sieve
                                        (100-digit input)
                                        > Sat Feb 20 11:38:40 2010 commencing number field sieve
                                        polynomial selection
                                        > Sat Feb 20 11:38:40 2010 time limit set to 0.35
                                        hours
                                        > Sat Feb 20 11:38:40 2010 searching leading coefficients from 1 to
                                        6474027
                                        > Sat Feb 20 11:38:41 2010 using GPU 1 (GeForce GTX 295)
                                        >
                                        Sat Feb 20 12:22:53 2010 polynomial selection complete
                                        > Sat Feb 20
                                        12:22:53 2010 R0: -127799610813500855 4742004
                                        > Sat Feb 20 12:22:53
                                        2010 R1: 90597371296321
                                        > Sat Feb 20 12:22:53 2010 A0:
                                        2724199565772163520 5536043245
                                        > Sat Feb 20 12:22:53 2010 A1:
                                        -122135191694100152 7056
                                        > Sat Feb 20 12:22:53 2010 A2:
                                        -33768914437945541
                                        > Sat Feb 20 12:22:53 2010 A3: 261332752
                                        > Sat
                                        Feb 20 12:22:53 2010 A4: 1080
                                        > Sat Feb 20 12:22:53 2010 skew 3406178.83,
                                        size 1.331082e-013, alpha -5.711799, combined = 1.375442e-008
                                        > Sat Feb 20
                                        12:22:53 2010 elapsed time 00:44:14
                                        >
                                        >
                                        > I have to test
                                        with CPU but I think there is a problem here... what do you think?
                                        >
                                        > Regards,
                                        > Orkblutt
                                        >
                                        >
                                        > From: br_gladman
                                        > Sent: Saturday, February 20, 2010 10:24 AM
                                        > To:
                                        href="mailto:ggnfs%40yahoogroups.com">ggnfs@yahoogroups. com
                                        >
                                        Subject: [ggnfs] Re: factmsieve.py & CUDA's msieve
                                        >
                                        >
                                        >
                                        >
                                        >
                                        > --- In
                                        href="mailto:ggnfs%40yahoogroups.com">ggnfs@yahoogroups. com, "Orkblutt" <orkblutt@> wrote:
                                        > >
                                        > > Yes I have the 6 .ptx file at
                                        the same level than msieve.exe.. . But same result.
                                        > >
                                        > > All the best,
                                        > >
                                        > > Orkblutt
                                        > >
                                        > >
                                        > > From: Jason Papadopoulos
                                        > > Sent: Saturday,
                                        February 20, 2010 1:33 AM
                                        > > To:
                                        href="mailto:ggnfs%40yahoogroups.com">ggnfs@yahoogroups. com
                                        > > Subject: Re: [ggnfs] factmsieve.py & CUDA's msieve
                                        > >
                                        > >
                                        > >
                                        > > Quoting Orkblutt
                                        <orkblutt@>:
                                        > >
                                        > > > Is this error because
                                        factmsieve.py doesn't support yet msieve using GPU?
                                        > > >
                                        > >
                                        > > The msieve GPU binary comes with a number of .ptx files;
                                        those files
                                        > > have to be in the same directory as the msieve binary
                                        for the GPU code
                                        > > to work.
                                        > >
                                        > > jasonp
                                        > >
                                        > >
                                        ------------ --------- --------- --------- --------- ------
                                        > > This message was sent using BOO.net's Webmail.
                                        > >
                                        href="http://www.boo.net/">http://www.boo. net/
                                        > >
                                        > I
                                        think the *.ptx files have to be in the current working directory, which is not necessarily the same directory as msieve is in. Since numbers are normally run from a directory that is used to hold the files generated by the script, the *.ptx files may need to be here.
                                        >
                                        > I hope to find a way of
                                        running msieve in its own directory while it writes files to the 'normal' working directory by manipulating the various paths used in factmsieve.py.
                                        >
                                        > Brian

                                        I have just updated thye Python version here:

                                        http://gladman. plushost. co.uk/oldsite/ computing/ factmsieve. py.zip

                                        It shoulkd now be able to run msieve without moving the *.ptx files (you need to put them back in the msieve directory of course).

                                        I have also added a USE_CUDA and a GPU_NUM define.

                                        Do remember to set these, other settings and the paths to your executables in this script before using it.

                                        I would be interested to know if it works.

                                        Brian

                                      • br_gladman
                                        ... I havew just uploaded from my site and the new version is definitely there. But I have uploaded it to the files area as well. Brian
                                        Message 19 of 25 , Feb 20, 2010
                                        • 0 Attachment
                                          --- In ggnfs@yahoogroups.com, "Orkblutt" <orkblutt@...> wrote:
                                          >
                                          > Are you sure the .py is updated on your website?
                                          > I still have the version dated of 09/02.

                                          I havew just uploaded from my site and the new version is definitely there. But I have uploaded it to the files area as well.

                                          Brian
                                        • Orkblutt
                                          Ok I have finally downloaded the correct file. line 901: if run_msieve( -g {0:d} -v - np .format(GPU_NUM) if USE_CUDA else -v -np ): should be if
                                          Message 20 of 25 , Feb 20, 2010
                                          • 0 Attachment
                                            Ok I have finally downloaded the correct file.
                                             
                                            line 901:
                                            if run_msieve('-g {0:d} -v - np'.format(GPU_NUM) if USE_CUDA else '-v -np'):
                                            should be
                                            if run_msieve('-g {0:d} -v -np'.format(GPU_NUM) if USE_CUDA else '-v -np'):
                                             
                                            All working fine, no need the .ptx in the working directory anymore.
                                             
                                            Thanks Brian.
                                             
                                            Orkblutt

                                            Sent: Saturday, February 20, 2010 4:56 PM
                                            Subject: [ggnfs] Re: factmsieve.py & CUDA's msieve

                                             



                                            --- In ggnfs@yahoogroups. com, "Orkblutt" <orkblutt@.. .> wrote:

                                            >
                                            > Are you sure the .py is updated
                                            on your website?
                                            > I still have the version dated of 09/02.

                                            I havew just uploaded from my site and the new version is definitely there. But I have uploaded it to the files area as well.

                                            Brian

                                          • br_gladman
                                            ... Thanks for the bug report - I ve updated it on my site but not in the files area here. Brian
                                            Message 21 of 25 , Feb 20, 2010
                                            • 0 Attachment
                                              --- In ggnfs@yahoogroups.com, "Orkblutt" <orkblutt@...> wrote:
                                              >
                                              > Ok I have finally downloaded the correct file.
                                              >
                                              > line 901:
                                              > if run_msieve('-g {0:d} -v - np'.format(GPU_NUM) if USE_CUDA else '-v -np'):
                                              > should be
                                              > if run_msieve('-g {0:d} -v -np'.format(GPU_NUM) if USE_CUDA else '-v -np'):

                                              Thanks for the bug report - I've updated it on my site but not in the files area here.

                                              Brian
                                            • Orkblutt
                                              ... Total yield: 611653 0/0 mpqs failures, 38913/17554 vain mpqs milliseconds total: Sieve 484159 Sched 0 medsched 177727 TD 421744 (Init 35584, MPQS 97821)
                                              Message 22 of 25 , Feb 20, 2010
                                              • 0 Attachment
                                                ...
                                                Total yield: 611653
                                                0/0 mpqs failures, 38913/17554 vain mpqs
                                                milliseconds total: Sieve 484159 Sched 0 medsched 177727
                                                TD 421744 (Init 35584, MPQS 97821) Sieve-Change 203378
                                                TD side 0: init/small/medium/large/search: 4680 35239 5575 9530 102681
                                                sieve: init/small/medium/large/search: 16775 68880 8429 57225 67753
                                                TD side 1: init/small/medium/large/search: 1344 30523 5514 9312 181682
                                                sieve: init/small/medium/large/search: 9412 43575 8531 66432 137147
                                                appending spairs.out.T0 to spairs.out
                                                appending spairs.out.T1 to spairs.out
                                                appending spairs.out to example.dat
                                                Found 6046136 relations, 101.0% of the estimated minimum (5985000).
                                                Traceback (most recent call last):
                                                  File "D:\ggnfs\factmsieve.py", line 1997, in <module>
                                                    run_siever(CLIENT_ID, NUM_THREADS, fact_p, lats_p)
                                                  File "D:\ggnfs\factmsieve.py", line 1709, in run_siever
                                                    if run_msieve(MSIEVE, '-t {0:d} - nc1'.format(NUM_CPUS)):
                                                TypeError: run_msieve() takes exactly 1 argument (2 given)
                                                 
                                                 
                                                Looks like there is an extra white space before nc1 parameter.
                                                 
                                                regards,
                                                 
                                                orkblutt

                                                Sent: Saturday, February 20, 2010 7:46 PM
                                                Subject: [ggnfs] Re: factmsieve.py & CUDA's msieve

                                                 



                                                --- In ggnfs@yahoogroups. com, "Orkblutt" <orkblutt@.. .> wrote:

                                                >
                                                > Ok I have finally downloaded
                                                the correct file.
                                                >
                                                > line 901:
                                                > if run_msieve(' -g
                                                {0:d} -v - np'.format(GPU_ NUM) if USE_CUDA else '-v -np'):
                                                > should
                                                be
                                                > if run_msieve(' -g {0:d} -v -np'.format( GPU_NUM) if USE_CUDA
                                                else '-v -np'):

                                                Thanks for the bug report - I've updated it on my site but not in the files area here.

                                                Brian

                                              • br_gladman
                                                ... This: if run_msieve(MSIEVE, -t {0:d} - nc1 .format(NUM_CPUS)): should be: if run_msieve( -t {0:d} - nc1 .format(NUM_CPUS)): I ve now updated the version
                                                Message 23 of 25 , Feb 20, 2010
                                                • 0 Attachment
                                                  --- In ggnfs@yahoogroups.com, "Orkblutt" <orkblutt@...> wrote:
                                                  >
                                                  > ...
                                                  > Total yield: 611653
                                                  > 0/0 mpqs failures, 38913/17554 vain mpqs
                                                  > milliseconds total: Sieve 484159 Sched 0 medsched 177727
                                                  > TD 421744 (Init 35584, MPQS 97821) Sieve-Change 203378
                                                  > TD side 0: init/small/medium/large/search: 4680 35239 5575 9530 102681
                                                  > sieve: init/small/medium/large/search: 16775 68880 8429 57225 67753
                                                  > TD side 1: init/small/medium/large/search: 1344 30523 5514 9312 181682
                                                  > sieve: init/small/medium/large/search: 9412 43575 8531 66432 137147
                                                  > appending spairs.out.T0 to spairs.out
                                                  > appending spairs.out.T1 to spairs.out
                                                  > appending spairs.out to example.dat
                                                  > Found 6046136 relations, 101.0% of the estimated minimum (5985000).
                                                  > Traceback (most recent call last):
                                                  > File "D:\ggnfs\factmsieve.py", line 1997, in <module>
                                                  > run_siever(CLIENT_ID, NUM_THREADS, fact_p, lats_p)
                                                  > File "D:\ggnfs\factmsieve.py", line 1709, in run_siever
                                                  > if run_msieve(MSIEVE, '-t {0:d} - nc1'.format(NUM_CPUS)):
                                                  > TypeError: run_msieve() takes exactly 1 argument (2 given)
                                                  >
                                                  >
                                                  > Looks like there is an extra white space before nc1 parameter.

                                                  This:

                                                  if run_msieve(MSIEVE, '-t {0:d} - nc1'.format(NUM_CPUS)):

                                                  should be:

                                                  if run_msieve('-t {0:d} - nc1'.format(NUM_CPUS)):

                                                  I've now updated the version on my site - thanks for debugging it.

                                                  Brian
                                                • Orkblutt
                                                  if fast there is 2 errors in the line 1709... line 1709 should be if run_msieve( -t {0:d} -nc1 .format(NUM_CPUS)): all the best, Orkblutt From: Orkblutt Sent:
                                                  Message 24 of 25 , Feb 20, 2010
                                                  • 0 Attachment
                                                    if fast there is 2 errors in the line 1709...
                                                    line 1709 should be
                                                    if run_msieve('-t {0:d} -nc1'.format(NUM_CPUS)):
                                                     
                                                    all the best,
                                                     
                                                    Orkblutt
                                                     

                                                    From: Orkblutt
                                                    Sent: Saturday, February 20, 2010 10:38 PM
                                                    Subject: Re: [ggnfs] Re: factmsieve.py & CUDA's msieve

                                                     

                                                    ...
                                                    Total yield: 611653
                                                    0/0 mpqs failures, 38913/17554 vain mpqs
                                                    milliseconds total: Sieve 484159 Sched 0 medsched 177727
                                                    TD 421744 (Init 35584, MPQS 97821) Sieve-Change 203378
                                                    TD side 0: init/small/medium/ large/search: 4680 35239 5575 9530 102681
                                                    sieve: init/small/medium/ large/search: 16775 68880 8429 57225 67753
                                                    TD side 1: init/small/medium/ large/search: 1344 30523 5514 9312 181682
                                                    sieve: init/small/medium/ large/search: 9412 43575 8531 66432 137147
                                                    appending spairs.out.T0 to spairs.out
                                                    appending spairs.out.T1 to spairs.out
                                                    appending spairs.out to example.dat
                                                    Found 6046136 relations, 101.0% of the estimated minimum (5985000).
                                                    Traceback (most recent call last):
                                                      File "D:\ggnfs\factmsiev e.py", line 1997, in <module>
                                                        run_siever(CLIENT_ ID, NUM_THREADS, fact_p, lats_p)
                                                      File "D:\ggnfs\factmsiev e.py", line 1709, in run_siever
                                                        if run_msieve(MSIEVE, '-t {0:d} - nc1'.format( NUM_CPUS) ):
                                                    TypeError: run_msieve() takes exactly 1 argument (2 given)
                                                     
                                                     
                                                    Looks like there is an extra white space before nc1 parameter.
                                                     
                                                    regards,
                                                     
                                                    orkblutt

                                                    Sent: Saturday, February 20, 2010 7:46 PM
                                                    Subject: [ggnfs] Re: factmsieve.py & CUDA's msieve

                                                     



                                                    --- In ggnfs@yahoogroups. com, "Orkblutt" <orkblutt@.. .> wrote:

                                                    >
                                                    > Ok I have finally downloaded
                                                    the correct file.
                                                    >
                                                    > line 901:
                                                    > if run_msieve(' -g
                                                    {0:d} -v - np'.format(GPU_ NUM) if USE_CUDA else '-v -np'):
                                                    > should
                                                    be
                                                    > if run_msieve(' -g {0:d} -v -np'.format( GPU_NUM) if USE_CUDA
                                                    else '-v -np'):

                                                    Thanks for the bug report - I've updated it on my site but not in the files area here.

                                                    Brian

                                                  • Orkblutt
                                                    no problem with that whitespace - nc1 ? From: br_gladman Sent: Saturday, February 20, 2010 10:56 PM To: ggnfs@yahoogroups.com Subject: [ggnfs] Re:
                                                    Message 25 of 25 , Feb 20, 2010
                                                    • 0 Attachment
                                                      no problem with that whitespace "- nc1"?
                                                       

                                                      Sent: Saturday, February 20, 2010 10:56 PM
                                                      Subject: [ggnfs] Re: factmsieve.py & CUDA's msieve

                                                       



                                                      --- In ggnfs@yahoogroups. com, "Orkblutt" <orkblutt@.. .> wrote:

                                                      >
                                                      > ...
                                                      > Total yield:
                                                      611653
                                                      > 0/0 mpqs failures, 38913/17554 vain mpqs
                                                      > milliseconds
                                                      total: Sieve 484159 Sched 0 medsched 177727
                                                      > TD 421744 (Init 35584, MPQS
                                                      97821) Sieve-Change 203378
                                                      > TD side 0:
                                                      init/small/medium/ large/search: 4680 35239 5575 9530 102681
                                                      > sieve:
                                                      init/small/medium/ large/search: 16775 68880 8429 57225 67753
                                                      > TD
                                                      side 1: init/small/medium/ large/search: 1344 30523 5514 9312 181682
                                                      >
                                                      sieve: init/small/medium/ large/search: 9412 43575 8531 66432 137147
                                                      >
                                                      appending spairs.out.T0 to spairs.out
                                                      > appending spairs.out.T1 to
                                                      spairs.out
                                                      > appending spairs.out to example.dat
                                                      > Found 6046136
                                                      relations, 101.0% of the estimated minimum (5985000).
                                                      > Traceback (most
                                                      recent call last):
                                                      > File "D:\ggnfs\factmsiev e.py", line 1997, in
                                                      <module>
                                                      > run_siever(CLIENT_ ID, NUM_THREADS, fact_p,
                                                      lats_p)
                                                      > File "D:\ggnfs\factmsiev e.py", line 1709, in
                                                      run_siever
                                                      > if run_msieve(MSIEVE, '-t {0:d} -
                                                      nc1'.format( NUM_CPUS) ):
                                                      > TypeError: run_msieve() takes exactly
                                                      1 argument (2 given)
                                                      >
                                                      >
                                                      > Looks like there is an extra
                                                      white space before nc1 parameter.

                                                      This:

                                                      if run_msieve(MSIEVE, '-t {0:d} - nc1'.format( NUM_CPUS) ):

                                                      should be:

                                                      if run_msieve(' -t {0:d} - nc1'.format( NUM_CPUS) ):

                                                      I've now updated the version on my site - thanks for debugging it.

                                                      Brian

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