LLR beta Version 3.8.13 (deprecated)

[Jean Penné]

Hi All,

I uploaded today the beta version 3.8.12 of the LLR program.
You can find it now on my personal site :

http://jpenne.free.fr/

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 presently a beta one because it is linked with the beta Version 28.4 of George Woltman's gwnum library, which is its only difference from V3.8.11.
I released also a cllrd binary, which is linked with the debug version of this
gwnum library.
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

[pinhodecarlos]

Thank you.
I'll wait for the 64bit Windows binary.

[rebirther]

Originally Posted by pinhodecarlos

Thank you.
I'll wait for the 64bit Windows binary.

Is ready for testing. You can find it here http://www.bc-team.org/downloads.php?cat=7

[pinhodecarlos]

Originally Posted by rebirther

Is ready for testing. You can find it here http://www.bc-team.org/downloads.php?cat=7

Thank you.

[pinhodecarlos]

2% speed up on Ivy Bridge processor running two instances.

[Jean Penné]

Thanks to Rebirther, I uploaded the two Win64 binaries this morning.
Regards,
Jean

[Cruelty]

Why not name it 3.9.x? After all it's a whole new 28.x version of gwnum library. Just a thought

[pinhodecarlos]

Testing exponents n=1.5M, RPS case, the client is slower by 1.5 % on an Intel Xeon CPU E5-2630 v2 at 2.6 GHz but 2.5 % faster for exponents n=5M. I am running two instances of LLR on this machine, letting free 10 cores.

[KEP]

Originally Posted by pinhodecarlos

Testing exponents n=1.5M, RPS case, the client is slower by 1.5 % on an Intel Xeon CPU E5-2630 v2 at 2.6 GHz but 2.5 % faster for exponents n=5M. I am running two instances of LLR on this machine, letting free 10 cores.

Please remember that timings on tests with the same k and same n value can have a difference of +/- 5 to 10 %, because each core benefits differently from the increased speeds, due to memory constraints. So propably nothing to be alarmed about, since it appears that your n=5M test took up more priority than your n=1.5M test did

[pinhodecarlos]

Originally Posted by KEP

Please remember that timings on tests with the same k and same n value can have a difference of +/- 5 to 10 %, because each core benefits differently from the increased speeds, due to memory constraints. So propably nothing to be alarmed about, since it appears that your n=5M test took up more priority than your n=1.5M test did

I don't think so. I changed the client that was running the n=1.5M tests to previous version and timings dropped. Kept the client running the n=5M tests.

[Jean Penné]

Hi,
Thanks to Iain Bethune, the MAC OSX 32bit and 64bit binaries are also available now. While building them, Iain discovered a bug while testing a single number, using -q"expression". It is now fixed.
This was due to a stupid extra test in "lprime.c" file.
Indeed, only the command-line versions are concerned.
I rebuilt the Windows 32bit cllr and cllrd, and also all Linux 32bit and 64bit binaries, and updated my site accordingly (compressed source directory included).
In spite of this bug, cllr 32bit seemed to work fine, nethertheless, perhaps the 64bit Windows command-line application would need a rebuild...
Sorry for this drawback, and Best Regards,
Jean