mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > PrimeNet

Reply
 
Thread Tools
Old 2019-11-05, 23:10   #1
David703
 
Oct 2019

17 Posts
Default Expiration date contradiction

So, I manually reserved an exponent to test, selecting that I only wanted world record exponents to LL test, and going in the manual extensions page (https://www.mersenne.org/manual_extension/) I see that after I extended the assignment several times I should have one year to complete it.

While if I go in the assignments page (https://www.mersenne.org/workload/) it says I only have 90 days to complete it.
Which of the two numbers of remaining days to complete the work is the actual one?
David703 is offline   Reply With Quote
Old 2019-11-06, 04:29   #2
PhilF
 
PhilF's Avatar
 
Feb 2005
Colorado

5×131 Posts
Default

I may be wrong, but I'll take a stab at this.

I think the 90 days refers to how long you have to start the exponent. Once the first progress report gets sent to the server, the expiration date changes to a year.
PhilF is offline   Reply With Quote
Old 2019-11-06, 13:05   #3
David703
 
Oct 2019

17 Posts
Default

Quote:
Originally Posted by PhilF View Post
I may be wrong, but I'll take a stab at this.

I think the 90 days refers to how long you have to start the exponent. Once the first progress report gets sent to the server, the expiration date changes to a year.

Ok, but how can I let PrimeNet know that I have started to work on the exponent if I'm manual testing?
David703 is offline   Reply With Quote
Old 2019-11-06, 14:39   #4
PhilF
 
PhilF's Avatar
 
Feb 2005
Colorado

5×131 Posts
Default

Normally Prime95 checks in with the server every day, reporting any progress on all of the exponents in the worktodo.txt file, including manual checkouts. If you have Prime95 set to not do that, or the machine doesn't have internet connectivity, then that becomes problematic.

I think all the server needs in order to give you that year for completion is that first progress report, even if you have to use a different machine to report it.
PhilF is offline   Reply With Quote
Old 2019-11-06, 14:46   #5
David703
 
Oct 2019

1710 Posts
Default

Quote:
Originally Posted by PhilF View Post
Normally Prime95 checks in with the server every day, reporting any progress on all of the exponents in the worktodo.txt file, including manual checkouts. If you have Prime95 set to not do that, or the machine doesn't have internet connectivity, then that becomes problematic.

I think all the server needs in order to give you that year for completion is that first progress report, even if you have to use a different machine to report it.

Ok, now Prime95 sent the server an extimated completion date. But the server says that the days to go are 276 while it will expire in only 120 days. What is the difference between these 2 values?
David703 is offline   Reply With Quote
Old 2019-11-06, 14:49   #6
Mark Rose
 
Mark Rose's Avatar
 
"/X\(‘-‘)/X\"
Jan 2013

2·5·293 Posts
Default

Quote:
Originally Posted by David703 View Post
Ok, now Prime95 sent the server an extimated completion date. But the server says that the days to go are 276 while it will expire in only 120 days. What is the difference between these 2 values?
The client estimated it would take 276 days to complete the test. The client can be somewhat inaccurate depending on other things using CPU, etc. Occasionally it will calculate the time remaining while the system is under heavy load an estimate a time taking even a hundred times longer than it will.

The 120 days is the time the server is allowing to finish, according to the rules here.
Mark Rose is offline   Reply With Quote
Old 2019-11-06, 15:04   #7
David703
 
Oct 2019

100012 Posts
Default

Quote:
Originally Posted by Mark Rose View Post
The client estimated it would take 276 days to complete the test. The client can be somewhat inaccurate depending on other things using CPU, etc. Occasionally it will calculate the time remaining while the system is under heavy load an estimate a time taking even a hundred times longer than it will.

The 120 days is the time the server is allowing to finish, according to the rules here.

Ok, got it! Thanks to you and PhilF for the help.
David703 is offline   Reply With Quote
Old 2019-11-11, 20:59   #8
moebius
 
moebius's Avatar
 
Jul 2009
Germany

607 Posts
Unhappy

Quote:
Originally Posted by Mark Rose View Post
The client estimated it would take 276 days to complete the test. The client can be somewhat inaccurate depending on other things using CPU, etc. Occasionally it will calculate the time remaining while the system is under heavy load an estimate a time taking even a hundred times longer than it will.

The 120 days is the time the server is allowing to finish, according to the rules here.


Sorry guys but this "expires (days)" at https://www.mersenne.org/manual_extension/ is disgusting since only the value at https://www.mersenne.org/workload/
really matters.
Why extending a Exponent 60 days, if that has no effect at expireing?
moebius is offline   Reply With Quote
Old 2019-11-17, 12:14   #9
moebius
 
moebius's Avatar
 
Jul 2009
Germany

607 Posts
Default

I hate early expiring and unclear data about it on the website, because it leads to what now really means wasted computing time. (and no it's not an early triple check,).

49299013
moebius is offline   Reply With Quote
Old 2019-11-17, 18:37   #10
Madpoo
Serpentine Vermin Jar
 
Madpoo's Avatar
 
Jul 2014

63618 Posts
Default

Quote:
Originally Posted by moebius View Post
Sorry guys but this "expires (days)" at https://www.mersenne.org/manual_extension/ is disgusting since only the value at https://www.mersenne.org/workload/
really matters.
Why extending a Exponent 60 days, if that has no effect at expireing?
"Disgusting" ? I'm hurt.

I added the days-to-expire estimates to certain pages as a general guideline, and it works by estimating the time-to-live based on the various assignment rules.

One thing it does not include (and I've made this clear repeatedly) is exponents that have an extension applied through that page. Reason being, it's not commonly used, and it was going to add to the complexity of the estimation calculations.

I've said that at some point I may modify it to take that into consideration, but it's just not something that has bubbled up into my to-do list.

Anyway, don't worry - if you've manually extended an assignment, that will work just fine.

The reason it was hard to estimate an expiration date in advance is that all of the code that does the actual expiration is "point in time". It analyzes each assignment to see "should this be expired?" and if so, expires the assignment and makes it available again.

Looking at an assignment and making a binary decision "expire: yes or no" is one thing. Predicting in advance exactly *when* it will expire is another thing entirely. Lots of rules, lots of "if this then that, or if this, then another thing, but if this then not that" etc.

For example, a brand new assignment has a shorter expiration time if it hasn't started yet. But once it does start, the expiration date changes. That's the simplest example, but you get the idea. Others are things like what category was it when expired, and what category is it in now? If it's moved into certain categories then it may expire sooner.

I hope that helps, and don't be frustrated, just finish your assignments in a timely manner.

Oh, and it's worth pointing out that "large" exponents (like 20-30 million above whatever current category 4 exponents) aren't going to expire anyway. There's just not a need. So people doing 100M digit checks are going to be safe for many years. That also includes people doing double-checks like 70-80M and higher. Of course a lot of those are first time checks that expired and were converted to double-checks when someone else turned in a result.
Madpoo is offline   Reply With Quote
Old 2019-11-18, 12:12   #11
moebius
 
moebius's Avatar
 
Jul 2009
Germany

25F16 Posts
Default

Thanks for the attention and explanation, no offense.
moebius is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
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
Exponent expiration drh PrimeNet 25 2011-10-24 04:51
Expiration date Italian PrimeNet 4 2003-12-13 19:54

All times are UTC. The time now is 01:14.


Mon Aug 2 01:14:23 UTC 2021 up 9 days, 19:43, 0 users, load averages: 1.21, 1.16, 1.14

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.