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

GMP ECM 5.0 (revision c) released

Expand Messages
  • Jim Fougeron.
    I have uploaded the latest GMP ECM release. I have placed most of the changes I had previously done to the ECM 4c-r7a into this new release. Things added
    Message 1 of 4 , Mar 2 2:30 PM
    • 0 Attachment
      I have uploaded the latest GMP ECM release. I have placed most of the changes
      I had previously done to the ECM 4c-r7a into this new release. Things added
      "beyond" the stock 5.0 release are:

      1. expression parser (similar to the expression parser in ecm4c-r7a, but to
      this new one has been added: Factorials, multi-factorials, primorials.
      Also the expression parser will accept * or . for multiplication (as is
      in the OpenPFGW parser).
      2. -a automatic B1 incrementing switch
      3. -deep to continue factoring a number even after a factor is found (that is
      iff the cofactor is composite). The default in ecm4c-r7a was to be in
      -deep mode, and there was a -1 switch to stop processing a number after a
      factor was found, and now with ECM 5, the opposite is the default (the
      default is r7a's -1 mode).
      4. -n and -nn "nice" mode (below normal and idle priority)
      5. -t n (trial division). The -rho switch has not been added to ecm 5.0
      (yet).
      6. -i file will use file as the input (instead of requiring stdin
      redirection).
      7. -ib file (and a -b for stdin redirection mode) process the file in
      "breadth-first mode. This mode (which was the default in 4c-r7a)
      is that each line of the file is processed, before the next loop
      is done. In the "default" mode of 5.0c, each line is processed as
      many loops as were required, prior to the next candidate being
      processed. In 4c-r7a, you had to use the -- switch to get the
      "depth-first" processing.
      8. -ve n has been added to "verbosely" show expressions shorter than n
      characters when looping.

      Some differences between ECM 5.0 and ECM 4c-r7a

      5.0 has different command line methods of entering items such as X and A and
      sigma.

      5.0 is considerably faster in the B2 stage, and slightly faster in B1 stage.

      5.0 added P-1 and P+1 factoring.

      5.0 added "save after B1" / "resume from saved B1"

      5.0 changed -l n to -c n (looping switch). It still performs the same
      function.

      5.0 changed from a "breadth-first" to a "depth-first" processing of a file (by
      default.)

      5.0 changed from a -deep search to a shallow -1 search by default.

      5.0 does not have -rho factoring (at this time).

      5.0 has a "enhanced" expression parser over 4c-r7a. Primorials, multi-
      primorials and factorials have been added.

      5.0 Many other enhancements were added to 5.0.


      At this time, I am not releasing the source code to this forum. I am working
      with PaulZ, to get these changes implemented into the "core" GMP-ECM package.
      At that time (which hopefully will be soon), the source can be obtained from
      the "official" source tarballs.

      NOTE, the release I have just placed up on Yahoo groups file folder:
      http://groups.yahoo.com/group/primenumbers/files/FactoringPrograms/GMP-ECM/e
      cm-5_0c.zip
      should be treated as a "beta" release. Even though the ECM-GMP 5.0 has been
      "officially" released, there is a significant amount of new code in this
      5.0c release. Also, most of this new code is a re-write. There was little
      code done in the 4c-r7a that was used without modification (the only thing
      was the expression parser, and even that was extended). A word of warning is
      to not remove the ecm4c-r7a from your system until you are satisfied that
      this 5.0c is working correctly for you.

      Also, the .zip file in the "files" folder, contains numerous exe files. These
      are CPU specific. Also, there is NO need for Cygwin1.dll, as these exe's
      have been built with MinGW, and are 100% self-contained.

      Jim.
    • andrew_j_walker <ajw01@uow.edu.au>
      ... the changes ... Things added ... ecm 5.0 The -t option appears to be broken. -t 10000 for instance gives an error. ... Is -oe replaced or redundant? ... It
      Message 2 of 4 , Mar 2 7:32 PM
      • 0 Attachment
        --- In primenumbers@yahoogroups.com, "Jim Fougeron." <jfoug@k...>
        wrote:
        > I have uploaded the latest GMP ECM release. I have placed most of
        the changes
        > I had previously done to the ECM 4c-r7a into this new release.
        Things added
        > "beyond" the stock 5.0 release are:
        >
        > 5. -t n (trial division). The -rho switch has not been added to
        ecm 5.0

        The -t option appears to be broken. -t 10000 for instance gives an
        error.
        > (yet).
        > 6. -i file will use file as the input (instead of requiring stdin
        > redirection).

        Is -oe replaced or redundant?

        >
        > 5.0 added P-1 and P+1 factoring.

        It would be nice to have this as an option to use before ecm
        as the trial and rho factoring work in 4c.
        > 5.0 has a "enhanced" expression parser over 4c-r7a. Primorials,
        multi-
        > primorials and factorials have been added.
        >

        Again, nice!

        Andrew
      • Christ van Willegen
        ... Do you have the source uploaded as well? Or, can you direct me to the source files for this? Christ van Willegen
        Message 3 of 4 , Mar 5 7:41 AM
        • 0 Attachment
          Jim Fougeron. wrote:

          >I have uploaded the latest GMP ECM release. I have placed most of the changes
          >I had previously done to the ECM 4c-r7a into this new release. Things added
          >"beyond" the stock 5.0 release are:
          >
          >

          Do you have the source uploaded as well? Or, can you direct me to the
          source files for this?

          Christ van Willegen
        • jim_fougeron
          The source will not be released by me like before. I am now a member of the official GMP-ECM development group. When that group releases ECM 5.1, then
          Message 4 of 4 , Mar 5 9:21 AM
          • 0 Attachment
            The source will not be released by me like before. I am now a member
            of the "official" GMP-ECM development group. When that group
            releases ECM 5.1, then there will be the sources for that. At this
            time, treat the 5.0c as an early release beta. Also, I will probably
            post 5.0d binaries today. 5.0d is what I have checked in to the CVS
            source control repository for ECM.

            5.0d fixes several bugs in the 5.0c release, and adds Phi(N,B) to the
            expression parser syntax.

            Jim.

            --- In primenumbers@yahoogroups.com, Christ van Willegen
            <christ@i...> wrote:
            > Jim Fougeron. wrote:
            >
            > >I have uploaded the latest GMP ECM release. I have placed most
            of the changes
            > >I had previously done to the ECM 4c-r7a into this new release.
            Things added
            > >"beyond" the stock 5.0 release are:
            > >
            > >
            >
            > Do you have the source uploaded as well? Or, can you direct me to
            the
            > source files for this?
            >
            > Christ van Willegen
          Your message has been successfully submitted and would be delivered to recipients shortly.