Exponent expiration

drh · 2011-08-19, 16:17

I have a few questions about the expiration of exponents. I know that manual assignments expire after 6 months, but does that happen automatically? Is this also true of PrimeNet assignments?

In the range that I'm working on at the moment, I've found over 30 exponents that were assigned over 6 months ago, and couple that will be 12 months in a few days. In the next month, this number will double to over 60 exponents.

90% of these are assigned to Anon, the rest to actual users, or at least previous users.

Can someone verify that exponents get released after 6 months? I'd like to get these cleaned up, yes, I'm a little OCD about it and I'm being patient. I could just go ahead and test them, but I sure don't want to be thought of as a Poacher.

Thanks,
Doug

Brian-E · 2011-08-19, 22:08

Others will correct me if I'm wrong, but no I don't think exponents generally expire after 6 months if their progress is regularly updated (and I think this update is supposed to happen at least every 60 days).

You don't mention which work type and which range it is, but note that LL-testing of 100M digit numbers will always take well over 6 months, generally a few years in fact. And some slower systems like mine will take a good 6 months or more even for testing an exponent in the current general first-time LL testing range. I'm not currently doing first time LL testing, but I certainly wouldn't be happy if I was and someone pinched my work just because I took over 6 months to finish.

cheesehead · 2011-08-20, 00:53

Quote:

Originally Posted by drh

I have a few questions about the expiration of exponents. I know that manual assignments expire after 6 months, but does that happen automatically? Is this also true of PrimeNet assignments?

You're looking at the wrong elapsed time.

Let me emphasize something Brian wrote parenthetically that may not stand out:

The most important factor in whether an exponent expires is not how long it's been assigned! It's how long it has been since the most recent progress report.

The idea is that as long as an assignee is regularly reporting progress, PrimeNet doesn't expire the assignment. The contributions of what some folks consider "slow" systems are still valued as long as they report regular progress. As I've sometimes put it: 1001 MPH is faster than 1000 MPH. (There are some exceptions made where there is especially slow progress on an exponent that has become conspicuous for being the last one under some "milestone", but this is, or should be, at the discretion of GIMPS administrators rather than a matter that impatient vigilantes poach on their own initiative.)

Quote:

Can someone verify that exponents get released after 6 months?

Not 6 months after assignment, but 60 days after most recent progress report.

(Maybe there's a 6-month guaranteed minimum -- i.e., expiration is at 60 days after last progress report, or 6 months after assignment, whichever is later.)

Quote:

I'd like to get these cleaned up, yes, I'm a little OCD about it and I'm being patient. I could just go ahead and test them, but I sure don't want to be thought of as a Poacher.

Thanks for asking us. Now, just remember that it's not how long an exponent's been assigned -- it's how long it's been since the last progress report.

drh · 2011-08-20, 03:00

Good, now I have a much better understanding ... and you are right, I wasn't all that clear. I'm doing TF's in the 76M range, bringing everything up to 73 bits.

So, is there a way to tell when the last update was on an exponent being TF'd?

Thanks,
Doug

Take for example 76106197, assigned for TF on 8/29/10

KingKurly · 2011-08-20, 03:23

Quote:

Originally Posted by drh

Good, now I have a much better understanding ... and you are right, I wasn't all that clear. I'm doing TF's in the 76M range, bringing everything up to 73 bits.

So, is there a way to tell when the last update was on an exponent being TF'd?

Thanks,
Doug

Take for example 76106197, assigned for TF on 8/29/10

http://mersenne.org/assignments/?exp...et+Assignments

To me, that looks like someone has the assignment on a V4 client, because it is not updating a percentage. The user's machine is claiming to the server that the test will be done in two days' time. Whether you believe it, and how long it has been claiming this, well, that's up to you...

A side note, please be careful when using that assignments site near the top of the hour; it is very CPU-intensive for the server. And try to restrict your queries to specific exponents, instead of large ranges. The server does a lot of work in the first minutes of every hour, so I try to avoid using the assignments page in the last 5 minutes and first 10 minutes of each hour.

Enjoy!

drh · 2011-08-20, 12:48

Quote:

Originally Posted by KingKurly

http://mersenne.org/assignments/?exp...et+Assignments

To me, that looks like someone has the assignment on a V4 client, because it is not updating a percentage. The user's machine is claiming to the server that the test will be done in two days' time. Whether you believe it, and how long it has been claiming this, well, that's up to you...

A side note, please be careful when using that assignments site near the top of the hour; it is very CPU-intensive for the server. And try to restrict your queries to specific exponents, instead of large ranges. The server does a lot of work in the first minutes of every hour, so I try to avoid using the assignments page in the last 5 minutes and first 10 minutes of each hour.

Enjoy!

Thanks, that's exactly what I was looking for.

Christenson · 2011-08-20, 14:52

King Kurly,
Are you stating that some time in the last 60 days, a V4 client has made a report to the server stating it is still working on the exponent involved, and it said it would be done in two days?

E
I'd let these guys finish, and move up and on...small irregularities like this are the norm on large distributed projects like this. Try the clean-up just monthly or so.

cheesehead · 2011-08-20, 15:19

I see now on the Manual Assignments page (Manual Testing -> Assignments) (http://mersenne.org/manual_assignment/) that "Manual assignments are good for only six months." So, there may be six-month expirations that are enforced only if the assignment was made manually through this page rather than through the automated PrimeNet process. This page doesn't mention progress reports.

Brian-E · 2011-08-20, 15:31

Quote:

Originally Posted by cheesehead

I see now on the Manual Assignments page (Manual Testing -> Assignments) (http://mersenne.org/manual_assignment/) that "Manual assignments are good for only six months." So, there may be six-month expirations that are enforced only if the assignment was made manually through this page rather than through the automated PrimeNet process. This page doesn't mention progress reports.

It mentions them in a negative way if you bring up the page without logging into PrimeNet first:

Quote:

[Failure to log in before filling out the form] also means you will not be able to manage the reservations such as extending the six month expiration date.

drh · 2011-10-20, 22:06

Ever since I got this information, I've been patiently watching all the exponents in the 76M range that I'm working on, catching ones that expire, get assignments, completing them, and turning them in. Yesterday, there were only 23 left in the 65 - 67 bit range, all assigned, a few by me, the rest by many others, some in the process of actively being worked on, and this morning I see that the user with the ID of "rduerr" poached them all. I'm not sure if this is SOP for this person or not, but it sure creates a lot of extra work in that I now have to do to find them all and eliminate them from my worktodo file. Guess there isn't anything that can be done to prevent it, but maybe after some number of complaints about an ID doing this, they could be penalized in some way ... very fustrating.

Doug

Christenson · 2011-10-21, 02:48

75M range is fairly high... you might want to join Mr P-1, me, and some others in getting a little better TF job done on exponents that are about to be assigned to LL tests.

A month is a very long time for one of my GPU TF tests to be out at 67 bits, since it takes perhaps an hour to do one on a bad day and my mid-range GT440 GPU.

Do send Prime95 a message about having your TF assignments from the server poached...that is very annoying, and there's sufficient work of all kinds not to need to squabble about it or duplicate it needlessly. Prime95 has some special talent in that area, and you don't need the run-in you might get if you messaged that user directly.

2011-08-19, 16:17	#1
drh Jan 2011 Cincinnati, OH 2²·5² Posts	Exponent expiration I have a few questions about the expiration of exponents. I know that manual assignments expire after 6 months, but does that happen automatically? Is this also true of PrimeNet assignments? In the range that I'm working on at the moment, I've found over 30 exponents that were assigned over 6 months ago, and couple that will be 12 months in a few days. In the next month, this number will double to over 60 exponents. 90% of these are assigned to Anon, the rest to actual users, or at least previous users. Can someone verify that exponents get released after 6 months? I'd like to get these cleaned up, yes, I'm a little OCD about it and I'm being patient. I could just go ahead and test them, but I sure don't want to be thought of as a Poacher. Thanks, Doug

2011-08-20, 03:00	#4
drh Jan 2011 Cincinnati, OH 2²×5² Posts	Good, now I have a much better understanding ... and you are right, I wasn't all that clear. I'm doing TF's in the 76M range, bringing everything up to 73 bits. So, is there a way to tell when the last update was on an exponent being TF'd? Thanks, Doug Take for example 76106197, assigned for TF on 8/29/10 Last fiddled with by drh on 2011-08-20 at 03:05 Reason: More clarification

2011-10-20, 22:06	#10
drh Jan 2011 Cincinnati, OH 64₁₆ Posts	Poaching Ever since I got this information, I've been patiently watching all the exponents in the 76M range that I'm working on, catching ones that expire, get assignments, completing them, and turning them in. Yesterday, there were only 23 left in the 65 - 67 bit range, all assigned, a few by me, the rest by many others, some in the process of actively being worked on, and this morning I see that the user with the ID of "rduerr" poached them all. I'm not sure if this is SOP for this person or not, but it sure creates a lot of extra work in that I now have to do to find them all and eliminate them from my worktodo file. Guess there isn't anything that can be done to prevent it, but maybe after some number of complaints about an ID doing this, they could be penalized in some way ... very fustrating. Doug

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Unexpected exponent expiration	Siegmund	PrimeNet	6	2017-10-03 03:21
Expiration Dates	Fred	PrimeNet	3	2016-02-20 08:30
Expiration time	NormanRKN	GPU Computing	7	2013-06-28 23:53
LL and DC expiration times	Chuck	GPU to 72	8	2012-01-18 17:39
Expiration date	Italian	PrimeNet	4	2003-12-13 19:54

2011-08-19, 22:08	#2
Brian-E "Brian" Jul 2007 The Netherlands 110011001110₂ Posts	Others will correct me if I'm wrong, but no I don't think exponents generally expire after 6 months if their progress is regularly updated (and I think this update is supposed to happen at least every 60 days). You don't mention which work type and which range it is, but note that LL-testing of 100M digit numbers will always take well over 6 months, generally a few years in fact. And some slower systems like mine will take a good 6 months or more even for testing an exponent in the current general first-time LL testing range. I'm not currently doing first time LL testing, but I certainly wouldn't be happy if I was and someone pinched my work just because I took over 6 months to finish.

2011-08-20, 14:52	#7
Christenson Dec 2010 Monticello 5×359 Posts	King Kurly, Are you stating that some time in the last 60 days, a V4 client has made a report to the server stating it is still working on the exponent involved, and it said it would be done in two days? E I'd let these guys finish, and move up and on...small irregularities like this are the norm on large distributed projects like this. Try the clean-up just monthly or so.

2011-08-20, 15:19	#8
cheesehead "Richard B. Woods" Aug 2002 Wisconsin USA 7692₁₀ Posts	I see now on the Manual Assignments page (Manual Testing -> Assignments) (http://mersenne.org/manual_assignment/) that "Manual assignments are good for only six months." So, there may be six-month expirations that are enforced only if the assignment was made manually through this page rather than through the automated PrimeNet process. This page doesn't mention progress reports.

2011-10-21, 02:48	#11
Christenson Dec 2010 Monticello 5×359 Posts	75M range is fairly high... you might want to join Mr P-1, me, and some others in getting a little better TF job done on exponents that are about to be assigned to LL tests. A month is a very long time for one of my GPU TF tests to be out at 67 bits, since it takes perhaps an hour to do one on a bad day and my mid-range GT440 GPU. Do send Prime95 a message about having your TF assignments from the server poached...that is very annoying, and there's sufficient work of all kinds not to need to squabble about it or duplicate it needlessly. Prime95 has some special talent in that area, and you don't need the run-in you might get if you messaged that user directly.