WorkToDo File

[storm5510]

In the worktodo file, there is a huge hexadecimal number near the front of each line:

Test= big hex number, exponent, power, 0

I searched around and didn't find anything relative. What is the meaning of that number?

[henryzz]

Originally Posted by storm5510

In the worktodo file, there is a huge hexadecimal number near the front of each line:

Test= big hex number, exponent, power, 0

I searched around and didn't find anything relative. What is the meaning of that number?

It is the assignment id that stops people stealing your assignment.

[storm5510]

Okay, it's just a random identification number. Thanks!

[storm5510]

I still have questions about the worktodo file. Sorry.

My old computer is smoothly chugging away and has a lot of exponents to test. At the end of each line, there is something like this: "66,67", "67,68", or "68,69". I can tell by looking at the main screen what these are, but why are the values only in the upper 60's?

At the beginning of the lines, I've seen "Test", "Factor", and "DoubleCheck". What else is there?

[Kevin]

Originally Posted by storm5510

I still have questions about the worktodo file. Sorry.

My old computer is smoothly chugging away and has a lot of exponents to test. At the end of each line, there is something like this: "66,67", "67,68", or "68,69". I can tell by looking at the main screen what these are, but why are the values only in the upper 60's?

At the beginning of the lines, I've seen "Test", "Factor", and "DoubleCheck". What else is there?

As you search higher and higher for factors, your chances of finding one decrease, and it also takes longer to get between bit levels (ie, from 2^65 to 2^66, then 2^66 to 2^67, etc.). At some point your chances of finding a factor become so small and the next stage of factoring takes so long that (on average) you save time by going straight to the LL test instead of doing more factoring.

As an example, say you're at a point where factoring to the next bit level would take 1 day and give you a 1% chance at finding a factor, and the LL test would take 10 days. If you continue factoring, 1% of the time you'll find a factor and it will only take 1 day to finish off the exponent, and 99% of the time it'll take 11 days to finish off the exponent (1 day of factoring, 10 days of LL testing). Thus, the expected time it would take to finish off an exponent would be 10.9 days. However, if you just go straight to the LL test and don't continue factoring, it'll only take 10 days to finish off an exponent.


There is one main additional work-type, which is P-1 factoring. Unlike trial factoring, which is an exhaustive search of all possible factors, it's a way to search for potential factors, P, where P-1 has a lot of small factors. It's one more step of factoring that's beneficial to do between trial factoring and LL testing, in the sense of "expected time to finish an exponent" I described above.

[Mini-Geek]

If you meant, "why do these numbers only start in the upper 60's", (or "why do these numbers start and end in the upper 60's") and not "why do these numbers end in the upper 60's", (as Kevin probably thought you asked) that's because the numbers have already been checked up to some amount, usually in the 60's bit levels. (and only should be checked a few more bit levels)
As Kevin said, the other main type of work is P-1 factoring, but there are a few other types that Prime95 supports that GIMPS doesn't use in its main work. From commonc.c in Prime95's source:

/* Handle Test= and DoubleCheck= lines.                    */
/*    Test=exponent,how_far_factored,has_been_pminus1ed        */
/*    DoubleCheck=exponent,how_far_factored,has_been_pminus1ed    */
...
/* Handle AdvancedTest= lines. */
/*    AdvancedTest=exponent */
...
/* Handle Factor= lines.  Old style is:                    */
/*    Factor=exponent,how_far_factored                */
/* New style is:                            */
/*    Factor=exponent,how_far_factored,how_far_to_factor_to        */
...
/* Handle Pfactor= lines.  Old style is:                */
/*    Pfactor=exponent,how_far_factored,double_check_flag        */
/* New style is:                            */
/*    Pfactor=k,b,n,c,how_far_factored,ll_tests_saved_if_factor_found    */
...
/* Handle ECM= lines.  Old style is: */
/*   ECM=exponent,B1,B2,curves_to_do,unused[,specific_sigma,plus1,B2_start] */
/* New style is: */
/*   ECM2=k,b,n,c,B1,B2,curves_to_do[,specific_sigma,B2_start][,"factors"] */
...
/* Handle Pminus1 lines:  Old style:                */
/*    Pminus1=exponent,B1,B2,plus1[,B2_start]            */
/* New style is:                        */
/*    Pminus1=k,b,n,c,B1,B2[,B2_start][,"factors"]        */
...
/* Handle PRP= lines.                            */
/*    PRP=k,b,n,c[,how_far_factored,tests_saved][,known_factors]    */
/* A tests_saved value of 0.0 will bypass any P-1 factoring        */

PRP means "probable prime".
For info on ECM factoring see http://mersennewiki.org/index.php/Elliptic_curve_method
AdvancedTest simply LL tests the given Mersenne number, (ignoring any sort of prefactoring) and is used by Prime95 when you use the Advanced > Test option.

[cheesehead]

Originally Posted by storm5510

At the beginning of the lines, I've seen "Test", "Factor", and "DoubleCheck". What else is there?

Read the readme.txt and undoc.txt files that came with your copy of prime95.

Be patient, work your way carefully through them and take notes to re-organize the info in your preferred manner -- the author was a professional programmer and a brilliant organizer of this first successful widescale Internet distributed computing project, but not a professional technical writer AFAIK.

[Uncwilly]

Cheesehead,

Maybe someone like you and/or others can build a good wiki entry for the structure of the worktodo. Also, one could work on it here and have a sticky that has several entries, the first a brief listing of entry types, then a posting for each type that gives a rundown on the arguements of each entry.

[cheesehead]

Ooh, good idea. On my official to-do list now.

[storm5510]

I now have firsthand experience of why they start, and end, in the upper 60's!