LLR Version 3.8.17 released [deprecated]
 

Hi All,

I uploaded today the version 3.8.17 of the LLR program.
You can find it now on my personal site :

[url]http://jpenne.free.fr/[/url]

The 32bit Windows and Linux compressed binaries are available as usual.
The Linux 64bit binaries are also released here.
I uploaded also the complete source in a compressed file ; it may be used to
build the Mac-Intel executable and also the 64bit Windows binary.

This LLR version is linked with the Version 28.8 of George Woltman's gwnum library
in which George have fixed the issue that occurred rarely while using AVX code.
I released also a cllrd binary, which is linked with the debug version of this
gwnum library.

Three new features has been added in the small primes ranges (< 1000 digits), all of them using the GNU-MP library and the APR-CL codes ; they are :
- The search for Wieferich primes in the range and with the base stated by the user.
- The test of Wieferich prime candidates with the base stated by user.
- The APR-CL primality test of general digit strings.
Please, look at the Readme file for details.
On Linux and Mac-Intel platforms, the relevant code is linked in the llr binary.
On the other part, on Windows platforms, three companion executables are required, which are compiled under cygwin system, and launched by the llr program as child processes. They are "llrwfsrch.exe", "tw.exe" and "aprcl.exe" ; the "cygwin1.dll" dynamic library is also needed.
To build these executables on 64bit Windows platform, using a cygwin shell, you can find all the sources in the aprcl sub-directory.

There are further improvements in Roundoff errors processing :
To avoid endless retries, AbortOnRoundoff option is forced when testing with the next FFT length has been done more than MaxFFTinc times (5 by default). Then, the default is continuing the test with the next term in the input file. But the user can override this behavior by setting -oStopOnAbort=1, and then, manually deciding to continue.

At the request of an user, the backup of the .ini file is now implemented, and is done each time a ks<number>= or ns<number>= is added in the .ini file.

As usual, I need help to build the 32bit Mac Intel binary, and also the 64bit Mac Intel and Windows ones.
Please, inform me if you encountered any problem while using this new version.
Best Regards,
Jean

Hi Jean,

I have sent you the files for win32, win64 and win64 GUI.

One question about the roundoff errors:

If one CPU has a roundoff error most of the time a double check without any roundoff errors has still different residues. Why is that so?

Can anyone point to exactly what has changed between gwnum 28.7 and 28.8 ? I had a brief search but I didn't find any release notes for 28.8.

Thanks!

[url]http://www.mersenneforum.org/showpost.php?p=399390&postcount=2[/url]

Thanks, George!

Hi,

Thanks to Iain Bethune, the Mac-Intel 32bit and 64bit binaries, are now available.

Thanks to Rebirther, the 64bit Windows binaries (both GUI and console), are also available.

However, for these 64 bit Windows applications, the users must also build the companion binaries (aprcl.exe, tw.exe and llrwsrch.exe) using cygwin, if they want to use the small primes features implemented in this LLR version ; I think (but am not sure) the 32bit companion binaries cannot be used here... Let me know if you have difficulties!

Best Regards,
Jean

[QUOTE=rebirther;426089]Hi Jean,

I have sent you the files for win32, win64 and win64 GUI.

One question about the roundoff errors:

If one CPU has a roundoff error most of the time a double check without any roundoff errors has still different residues. Why is that so?[/QUOTE]

Thank You for this work, Reb!

About the Roundoff errors, I think that roundoff errors, even if they where recovered by square-carefully, occured because the test ran near the FFT limit, so, a run using the next FFT length is certainly more reliable, and must be taken into account, if it yields a different residue...

Regards,
Jean

Parsing errors on ABC input formats
 

Hi LLR users,

Primegrid LLR users found that parsing errors may occur when using input file with some ABC formats headers. Many thanks to Jmblazek and to Rogue, wo warned me about this bug!
This is now fixed, and I tested successfully all ABC format headers presently accepted by LLR version 3.8.17.
Also, I updated and uploaded all the binaries I could build, and, indeed, the source, on my personal site.
I would be great if Iain Bethune and Rebirther accepted to rebuild the MAC OS and Win64 binaries, would you excuse me for this drawback...
Thank you by advance, and Best Regards,
Jean

Parsing errors on ABC input formats
 

Hi LLR users,

Thanks to Iain Bethune and Rebirther, Mac OS and Win64 binaries have been rebuilt, so, the source and all the binaries are now updated and uploaded, with this bug fixed.
Note that this error may occur in all previous versions of LLR, causing that some ABC format input files could not be processed at all, but not yielding false results on processed files!

Best Regards,
Jean

I noticed that this version of LLR ( on Win and Linux) produce file [B]save_llr.ini [/B]when prime is found. I wasn't noticed this behavior before. Is this new "feature" and what function have?
Thanks for reply

[QUOTE=pepi37;432557]I noticed that this version of LLR ( on Win and Linux) produce file [B]save_llr.ini [/B]when prime is found. I wasn't noticed this behavior before. Is this new "feature" and what function have?
Thanks for reply[/QUOTE]

Yes, this is a new feature ; it was required by users searching k*b^n+c candidates for the first n giving a prime, in a set of several k values. These users are using the "StopOnPrimedk=1" option, so, the .ini file may become large, and need a backup...

Regards,
Jean