PFGW 3.2.0 has been Released
 

You can d/l Windows, MacIntel, and Linux versions from these links:

[URL="http://openpfgw.svn.sourceforge.net/viewvc/openpfgw/pfgw_win_3.2_20090725.zip?view=log"]PFGW for Windows[/URL]
[URL="http://openpfgw.svn.sourceforge.net/viewvc/openpfgw/pfgw_mac_3.2_20090725.zip?view=log"]PFGW for MacIntel[/URL]
[URL="http://openpfgw.svn.sourceforge.net/viewvc/openpfgw/pfgw_linux_3.2_20090725.zip?view=log"]PFGW for Linux[/URL]

Please do not use IE to get these files. IE seems to want to d/l the files as text instead of binary which will mean that you cannot unzip them after d/ling as the zip archive will be corrupted. You should use a better browser such as Safari or FireFox.

Enhancements to: v3.2.0 RC 1c[list][*]Built with version 25.12 of gwnum. This version of gwnum addresses SUMOUT errors that occur on some tests.[*]Modify tray icon to show current number and its progress.[*]Generate 64 bit residues instead of 62 bit residues.[*]Added -N option for command line version. When used PFGW will execute with NORMAL priority (1.2 behavior). Without it, PFGW will execute in IDLE priority. WinPFGW will execute in IDLE priority.[*]Added -V option so that command line version of PFGW can output diagnostic information.[*]Only show max round off line if MAXDIFF < 1e10 because a large MAXDIFF could wrap.[*]When using -q in WinPFGW, ensure that the results of the test are output when the number is composite.[*]Use special modular reduction code when PRP testing (k*b^n+/-c)/d forms.[*]Call gwset_larger_fftlen_count instead of gwset_safety_margin as it will guarantee that gwnum will use the next larger FFT size.[/list]
Thanks to various people who have found numbers that cause PFGW to get ROUNDOFF or SUMOUT errors.

Many thanks to George for tracking down and addressing some problems that PFGW uncovered in his gwnum library.

Thanks to Steven Harvey for providing the Linux build. I cannot test it, so I hope that I didn't botch the file when zipping and adding to the repository.

Enjoy,
Mark

Finally! PFGW now produces true 64-bit residuals that can compare perfectly with those produced by LLR, Phrot, and Prime95! :w00t: :w00t: :w00t:
[code]110042066*3^25001+1 is composite: RES64: [A03D235977C0541E] (3.0068s+0.0011s)
110042066*3^25001+1 is not prime. RES64: A03D235977C0541E OLD64: E0B76A0C6740FC57 Time : 17.125 sec.
110042066*3^25001+1 is composite LLR64=a03d235977c0541e. (e=0.03613 ([EMAIL="0.0641379~3.23992e-016@0.000"]0.0641379~3.23992e-016@0.000[/EMAIL]) t=7.75s)[/code]

It is indeed fantastically fast for the PRPs of (k*b^n+/-c)/d form!
I got a ~5-6x speedup, immediately.

[QUOTE=rogue;182840]Please do not use IE to get these files. IE seems to want to d/l the files as text instead of binary which will mean that you cannot unzip them after d/ling as the zip archive will be corrupted. [/QUOTE]It seem that with latest patches IE8 is correctly downloading those files :tu:

Hello, I have ever a problem to download this pfgw version 3.2 from this source.
(An Internet problem)

Can everyone send me this zip file to [email]nluhn@yahoo.de[/email]

Many thanks.

Best

Norman

[QUOTE=Cybertronic;184053]Hello, I have ever a problem to download this pfgw version 3.2 from this source.
(An Internet problem)

Can everyone send me this zip file to [email]nluhn@yahoo.de[/email]

Many thanks.

Best

Norman[/QUOTE]

I will.
Which version?

Luigi

[QUOTE=Cybertronic;184053]Hello, I have ever a problem to download this pfgw version 3.2 from this source.
(An Internet problem)

Can everyone send me this zip file to [email]nluhn@yahoo.de[/email]

Many thanks.

Best

Norman[/QUOTE]

Don't use IE to d/l the file. Others have experienced problems with IE, but nobody else (to my knowledge) has experienced problems with other browsers.

In your case if other browsers also fail, maybe it is an OS issue. Do you have the latest Windows patches?

Thanks to Luigi. This was the correct version.

Both, IE and Firefox , the download don't work.

But I have it now :-) . Very fine !

I got the following error when playing with WinPFGW 3.2.0:

[QUOTE]-----------------------------------------------------------------------
Error occuring in PFGW at Tue Oct 20 20:27:20 2009
Expr = 2*15^1600-1
Failed at bit 484 of 6252
Msg = Detected in MAXERR>0.45 (round off check) in prp_using_gwnum
Iteration: 484/6252 ERROR: ROUND OFF 0.46875>0.45[/QUOTE]

Error is repeatable and also occurs on 2nd PC.

C2Q 6700 and C2Duo 4300.

edit:
Sorry. I followed a link to get to this thread. Fixed in version 3.2.3 below.

[QUOTE=Flatlander;193384]Sorry. I followed a link to get to this thread. Fixed in version 3.2.3 below.[/QUOTE]

This can still happen in 3.2.3, but should be fairly rare. When it does occur, just re-run the failed test with -a1.

[quote=rogue;193437]This can still happen in 3.2.3, but should be fairly rare. When it does occur, just re-run the failed test with -a1.[/quote]



Where can I find version 3.2.3? I checked the sourceforge link and it's still 3.2.0. I'm running 3.2.0 and getting a lot of the round off errors at the low n-ranges on a very large conjecture...both Window's and Linux versions. I'm hoping that 3.2.3 corrects most of the problem. I'm going to need to rerun my whole test because I've found 2 primes in the list in pfgw-err.log out of about the first 100 erroneous tests.

It seems to work great at the higher n-ranges. The tests that I'm doing are for n=1 to 1000 on Riesel base 40...conjecture k>3M.

Edit: Would it slow it down much if I simply used the -a1 switch when testing the entire file?


Thanks,
Gary