mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Marin's Mersenne-aries (https://www.mersenneforum.org/forumdisplay.php?f=30)
-   -   Semi-automated P-1 testing (https://www.mersenneforum.org/showthread.php?t=1175)

 GP2 2003-09-29 15:11

Semi-automated P-1 testing

Xyzzy,

After you finish any manual batch of P-1 tests, how about trying the method suggested by Complex33 for semi-automating it:

It would make sense to do P-1 testing this way (for Primenet credit to boot).

I can still post Pfactor= sets though for those who want them, in higher ranges well away from the leading edge of double-checking.

I have another idea though (not P-1) that will require some manual Mersenne-ary work. I'll describe it in another thread.

 GP2 2003-09-29 15:19

Actually, it occurs to me that the semi-automated method mentioned above would only do P-1 testing of exponents that had never been P-1 tested... but there are some exponents that have been P-1 tested with lower-than-average bounds.

For instance:
[font=courier]
10401191,145000,145000
[/font]

Someone with a machine with more memory could redo these P-1 tests with higher bounds. This would have to be done by manual testing, since the semi-automated method would skip any exponents that have had any P-1 test done at all no matter how low the bounds.

Still, the higher priority would probably be exponents that haven't had any P-1 test at all done, and there's still plenty of those.

 GP2 2003-09-29 19:01

So in summary:

I think we can have two kinds of P-1 testing going on in parallel:

1) The semi-automated method described in [url]http://mersenneforum.org/showthread.php?s=&postid=11671#post11671[/url]
whose purpose would be to do a clean sweep of all P-1 tests ahead of the leading edge of double-checking.

2) The manual method of posting sets of [font=courier]Pfactor=p,depth,1[/font] lines for worktodo.ini. These won't be consecutive, rather they'll be a mix of smaller and larger exponents. The idea is to thin out the overall herd of un-P-1-tested exponents rather than do an entire range at a time.

 All times are UTC. The time now is 19:33.