Modifications to LL assignment rules!!!

[markr]

Originally Posted by Prime95

Rules for manual testing now implemented. Manual testers can now get cat 2 assignments if they've set the "I want smallest exponents" option and have a decent track record over the last 120 days.

It only gives me cat 3 LL or DC e.g. 73559879 and 38854451. I took the first time LL just as a test; my manual work is all DC & P-1.

[proxy2222]

Hi, first of all I applaud the new LL assignment rules, they work really well

However, I have found a bug with the expiration date for my exponent M67597393

I am aware of the assignment rules that Cat0 should be started within 7 days and finished within 30 days.

I have registerd the exponent to perform LL using cudaLucas. Prime95 was used to reserve the exponent. On the status page, the expiration date showed 7 days - which is correct.

The exponent ran for about 15 minutes and I let it communicate with primenet: expiration date changed from 7 days to 30 days and stage changed from empty to LL - still correct.

Exponent is currently running on cudaLucas (75% completed) but to my surprise it expired yesterday!

Now I have an expired exponent and cannot get any cat0/cat1/cat2 anymore for 4 months

So my question is 2 fold:

1) Fix the shown expiration date if exponent was started BUT has 0.00 % completed: it should show 7 days instead of 30.

2) @Madpoo or @Prime95: would it be possible to remove my expired exponent so I can get cat0-2 again ? I trusted the expiration date on the status page.....

Many thanks!!

Kind regards,

Hans

[Prime95]

Your assignment has been purged.

The SQL query checks for %done > 0. Your brief prime95 run changed the state to "LL" but the %done value did not get above 0 (I think you need to get to 0.1% for prime95 to send a non-zero %done).

Sorry for the trouble.

[proxy2222]

Originally Posted by Prime95

Your assignment has been purged.

The SQL query checks for %done > 0. Your brief prime95 run changed the state to "LL" but the %done value did not get above 0 (I think you need to get to 0.1% for prime95 to send a non-zero %done).

Sorry for the trouble.

Thanks!

[Madpoo]

Originally Posted by Prime95

Your assignment has been purged.

The SQL query checks for %done > 0. Your brief prime95 run changed the state to "LL" but the %done value did not get above 0 (I think you need to get to 0.1% for prime95 to send a non-zero %done).

Sorry for the trouble.

My little function to figure out the expiration date might be looking at whether it started at all, in which case even a 0.0% complete would qualify... if the code that does the actual expiration is looking not only at whether it started but whether it's moved the needle beyond zero, I should adjust.

Apologies for that... my function works backwards from what the expiration code does and I'm sure I just missed that nuance.

[retina]

Originally Posted by Madpoo

My little function to figure out the expiration date might be looking at whether it started at all, in which case even a 0.0% complete would qualify... if the code that does the actual expiration is looking not only at whether it started but whether it's moved the needle beyond zero, I should adjust.

Perhaps while you are at it, you could make it say something like "7+23" instead of "7" when not started and then "30" after starting. Or maybe "30(7)", or at least something to show the independent expiry durations. It is so confusing to see the expiry date change suddenly.

[petrw1]

There are LOTS and LOTS of assignments much larger that are WAY WAY WAY over a year old; and have no progress and no updates in almost 5 years in some cases....

Shouldn't these too have expire in a reasonable (granted longer) time period:

Sort this by Last Update:

http://www.mersenne.org/assignments/...&exp1=1&extf=1

[Prime95]

Originally Posted by petrw1

There are LOTS and LOTS of assignments much larger that are WAY WAY WAY over a year old; and have no progress and no updates in almost 5 years in some cases....

Shouldn't these too have expire in a reasonable (granted longer) time period:

Sort this by Last Update:

http://www.mersenne.org/assignments/...&exp1=1&extf=1

Assignments for exponents this large are not expired at all. Feel free to poach one if you are so inclined.

[S485122]

Originally Posted by Prime95

Assignments for exponents this large are not expired at all. Feel free to poach one if you are so inclined.

I would not publicly encourage anyone to poach, and you as owner of the project should be especially cautious.

In my opinion, assignments should be updated regularly as stated by the rules ("Since PrimeNet began, the server has recycled exponents where the client computer is 60 days past due in updating the server. This rule has been fairly effective and will continue.")

Manual reservations should be renewed annually (with no possibility to extend such reservation three years in the future.) This would not prevent somebody to work those exponents with sometimes expected completion dates more than 60 years in the future ! They could even retain their reservation but would have to do something about it at least once a year.

With the current non application of some of the rules those exponents will not be recycled until they are in CAT 1 territory !

Jacob

[henryzz]

Currently those exponents aren't even tier 1 for 100M digits. Once testing gets near that level then the position can be reconsidered.

[Mark Rose]

Originally Posted by henryzz

Currently those exponents aren't even tier 1 for 100M digits. Once testing gets near that level then the position can be reconsidered.

The problem is that some of the abandoned exponents have weeks of TF work done on them.