Mac build 0.99 is now available: [URL="http://www.pyramid-productions.net/downloads/genefercuda_macintel_0.99.tar.gz"]http://www.pyramid-productions.net/downloads/genefercuda_macintel_0.99.tar.gz[/URL]

Cheers

- Iain

Wonderful work, it runs flawlessly after the first make :bow:

Now, I have a few questions... :smile:

1 - How would a input file looks like? Does genefer parse FermFact / NewPGen / npsieve files?
2 - Is there a website where reservations are made?
3 - Is there a genefer user manual (Mark?)?

Luigi

A question for Mr. Yamada
 

Is there a possibility to have a (generalized) Fermat test program like GMP-Fermat (or GFN from Geoff Reynolds) based on CUDA? :smile:

Luigi

[QUOTE=ET_;279865]1 - How would a input file looks like? Does genefer parse FermFact / NewPGen / npsieve files?
2 - Is there a website where reservations are made?
3 - Is there a genefer user manual (Mark?)?[/QUOTE]

1. ABC format. I don't recall which flavor format exactly.
2. GFN search projects are run by PrimeGrid using PRPNet. There are some n values not being actively searched yet, one is being searched by David Underbakke and PrimeGrid is sieving the rest.
3. No. All versions of genefer are based upon Yves Gallot's original code.

I'm afraid I did something wrong with genefer CUDA... :redface:

[code]
Testing 2868122^16384+1... 348160 steps to go
maxErr exceeded for 2868122^16384+1, 0.5000 > 0.4500
[/code]

Are there limitations on the chosen base?

[QUOTE=ET_;279866]Is there a possibility to have a (generalized) Fermat test program like GMP-Fermat (or GFN from Geoff Reynolds) based on CUDA? :smile:[/QUOTE]

Interesting idea. I think it would be easy to do, but I'll leave the real answer to Shoichiro.

[QUOTE=rogue;279870]1. ABC format. I don't recall which flavor format exactly.
2. GFN search projects are run by PrimeGrid using PRPNet. There are some n values not being actively searched yet, one is being searched by David Underbakke and PrimeGrid is sieving the rest.
3. No. All versions of genefer are based upon Yves Gallot's original code.[/QUOTE]

Thank you Mark! :smile:

I stumbled upon Yves Gallot and "The FatPhil" pages, but they seemed out of date.
So, the only way to actively search for such GFNs is joining BOINC and using their pre-sieved ranges. Too bad :sad:

Luigi

[QUOTE=rogue;279873]Interesting idea. I think it would be easy to do, but I'll leave the real answer to Shoichiro.[/QUOTE]

I tried to "adapt" Oliver's mfaktc, but with the limitation at 95 bits per factor the project would be useless. Anyway, it could be a good place to start from.

Luigi

Ver 1.01 source code with Linux64 exec file.:smile:

[QUOTE=msft;280075]Ver 1.01 source code with Linux64 exec file.:smile:[/QUOTE]

Differences with 0.99? :smile:

Luigi

[QUOTE=ET_;280076]Differences with 0.99? :smile:
[/QUOTE]
Ken Braziermay`s code changes that may produce a percent or two speed increase in GeneferCUDA.