Error in Worktodo file

[CADavis]

Originally Posted by GARYP166

2) What is the trigger that adds a new exponent to the queue? We've established that having one of my exponents reaching 75% was just a co-incidence, but how far does the test have to progress before another one is allocated?

When the amount of time left in Test > Status is less than your setting for "Days Of Work to Queue Up" in Options > Preferences.

Example: Test > Status shows your test to finish in about 12 days. If you setting is to queue up 10 days of work, it will get another test from PrimeNet in about 2 days.

[GARYP166]

Yes that sounds reasonable.

I missed Prime95 off the list in my previous message - sorry.

G

[cheesehead]

Originally Posted by GARYP166

1) What should I have in my worktodo file so that 46,266,683 is the next to start?

Can you post what's in the worktodo now?

[GARYP166]

This is it.

[Worker #1]
DoubleCheck=44C4F6F2CB1AA38A4941A98466710304,23110891,67,1
Test=A5996856A4F73D28634EFA6CB510066F,exponent,68,1
[Worker #2]
Test=BDA22BB964CF82416AE126334DFA234B,45890281,68,1
DoubleCheck=B94D34BF558295F697581568AD2E14AF,exponent,66,1

G

[cheesehead]

Originally Posted by GARYP166

[Worker #1]
DoubleCheck=44C4F6F2CB1AA38A4941A98466710304,23110891,67,1
Test=A5996856A4F73D28634EFA6CB510066F,exponent,68,1
[Worker #2]
Test=BDA22BB964CF82416AE126334DFA234B,45890281,68,1
DoubleCheck=B94D34BF558295F697581568AD2E14AF,exponent,66,1

Originally Posted by GARYP166

What should I have in my worktodo file so that 46,266,683 is the next to start?

A) Stop and exit Prime95.

B) Get rid of the two lines with "exponent", so the worktodo is back to:

[Worker #1]
DoubleCheck=44C4F6F2CB1AA38A4941A98466710304,23110891,67,1
[Worker #2]
Test=BDA22BB964CF82416AE126334DFA234B,45890281,68,1

C) Do you still have a copy of your 46266683 test assignment line somewhere?

If so, then copy your 46266683 test assignment line into the worktodo file. Put it either just after the " DoubleCheck=44C4 ..." line or just after the " Test=BDA2 ..." line, so that the worktodo looks like either:

[Worker #1]
DoubleCheck=44C4F6F2CB1AA38A4941A98466710304,23110891,67,1
( >>> Here place your original assignment line for 46266683 <<<)
[Worker #2]
Test=BDA22BB964CF82416AE126334DFA234B,45890281,68,1

if you want the 46266683 test to resume when the DC on 23110891 completes,

or

[Worker #1]
DoubleCheck=44C4F6F2CB1AA38A4941A98466710304,23110891,67,1
[Worker #2]
Test=BDA22BB964CF82416AE126334DFA234B,45890281,68,1
( >>> Here place your original assignment line for 46266683 <<<)

if you want the 46266683 test to resume when the test on 45890281 completes.

D) Restart Prime95.

- -

If you don't still have a copy of your original 46266683 test assignment line, we'll explain further.

[GARYP166]

Thanks Cheesehead, but I'm afraid I don't have the original files - they disappeared into the ether. See

http://www.mersenneforum.org/showthread.php?t=11476

I don't mind restarting the exponent as only about three days' work was done.

G

[cheesehead]

Originally Posted by GARYP166

See

http://www.mersenneforum.org/showthread.php?t=11476

Oh ... that's why this looked a bit familiar.

I just looked at 46266683's status. PrimeNet now has it assigned to you for double-checking:

AssignedDouble-checking to "GARY P" on 2009-11-20

What happened, apparently, is that back in the summer PrimeNet reassigned it for first-time testing to someone else, "linded", after you hadn't reported any progress for six months. Then "linded" reported a first-time test result in September.

Recently (2009-11-20), apparently you communicated with PrimeNet while either your worktodo file had a line for 46266683, or your prime.spl file still had a progress report for 46266683 left over from February. Whichever way it happened doesn't matter!

The important point is that PrimeNet has now re-reserved 46266683 for you, as a doublecheck this time.

So, add the following line to your worktodo file:

DoubleCheck=,46266683,68,1

[Note that there must be a comma following the = sign.]

Put it immediately after the

DoubleCheck=44C4F6F2CB1AA38A4941A98466710304,23110891,67,1

line

or immediately after the

Test=BDA22BB964CF82416AE126334DFA234B,45890281,68,1

line, depending on whether you want your test of 46266683 to resume after your 23110891 doublecheck completes or after your 45890281 LL test completes.

Then, the next time your Prime95 sends a progress report to PrimeNet, PrimeNet will (I think) fill in your 32-hex-digit assignment ID key (between the = sign and the first comma) on the 46266683 line in your worktodo file. From then on, it will look just like (and, indeed, be) a normal assignment line.

[ckdo]

Originally Posted by cheesehead

The important point is that PrimeNet has now re-reserved 46266683 for you, as a doublecheck this time.

Somehow I fail to see the point in running a 43.6m doublecheck right now unless one has a savefile with considerable progress. In the same time that test takes to complete one can run about four 23m doublechecks...

[cheesehead]

It depends on how Gary feels about finally finishing 46266683 himself versus getting some other, first-time, assignment and leaving 46266683 to someone else to doublecheck a few years from now.

Originally Posted by ckdo

In the same time that test takes to complete one can run about four 23m doublechecks...

The iterations in each case (92M fast ones or 46M slower ones) will have to be done sometime by somebody either way. Either choice benefits GIMPS in the long run.

Personally, I think the most important factor is satisfaction of the participant, who should do whatever he'll enjoy most.

Yes, there are some folks (I'm one of them) urging users to take on more DC assignments, in order to close up the gap between DC and first-time results, but only if they find it interesting. To me, each particular user's satisfaction is more important than some race between DCs and F-Ts. GIMPS has a variety of tasks available for different tastes, and all those contributions are valuable.

Suppose the first result for 46266683 turns out to be in error, and the DC finds it to be a Mersenne prime? That is a very, very long shot (one chance in multimillions), but Gary may wish that he'd been the one who performed that DC.

OTOH, suppose Gary takes some other first-time assignment instead of DCing 46266683 and that other one turns out to be a Mersenne prime. Then he'll be really glad he dropped 46266683. His chances are better in this second case than in the first case.

[GARYP166]

I didn't join for the investment opportunity, though of course if I discover anything I shan't be ungrateful. I think the answer is the good old British compromise - my PC can check two numbers at once, so maybe one should be 46,266,683 and the other should be something new. I mentioned 49,979,687 which is the 3,000,000th prime according to the list on Chris Caldwell's site and which has no factor below 2^68.

Question: what option do I choose from the drop down menu when I want to test it? I guess that an LL test would be the next that is needed.

Incidentally, what is an spl file (Cheesehead's posting of 24th November, 4:48 p.m.)? There are none with that extension in my Prime95 folder, or indeed anywhere on the PC apart from three that have the names of languages as part of the name.

G

[Mini-Geek]

Originally Posted by GARYP166

I mentioned 49,979,687 which is the 3,000,000th prime according to the list on Chris Caldwell's site and which has no factor below 2^68.

Question: what option do I choose from the drop down menu when I want to test it? I guess that an LL test would be the next that is needed.

It'd be best to just tell Prime95 what's done and let it handle what needs to be done. This line tells it everything:

Test=49979687,69,0

Put that in worktodo.txt somewhere and restart Prime95, and it'll get to it when it gets there in the file. In this case, it's already at the optimal TF depth, but doesn't have P-1 done yet, so it'll run P-1 at optimal bounds for the memory you allow (the more, the better), then run the LL.

Here a breakdown of the line:
Test tells it the sort of work it'll be doing, i.e. any prefactoring necessary and then an LL test
49979687 is, of course, the exponent (commas mean a very different thing here, so you need to make sure you don't make it 49,979,687)
69 tells it that it's known there are no factors below 2^69 (which is accurate, as far as PrimeNet knows anyway)
0 tells it that no P-1 factoring has been done (which is accurate, as far as PrimeNet knows anyway; 1 would mean that it has been done)

If your Prime95 is set to communicate with PrimeNet, it'll tell PrimeNet about what it's doing, fill in an assignment key, and report the work when it's done, just like any test that PrimeNet gave you. If not, you should use http://www.mersenne.org/manual_result/ to report any results (such as complete TF, P-1, and LL results) when they complete.

I'm pretty sure the prime.spl file is basically a file Prime95 makes to prepare to communicate with PrimeNet, listing what it'll say when it does communicate. Not 100% sure on that, or on the details of it.