mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Information & Answers (https://www.mersenneforum.org/forumdisplay.php?f=38)
-   -   Why can't I extend something that's certain to not complete in time? (https://www.mersenneforum.org/showthread.php?t=26509)

drkirkby 2021-02-18 10:28

Why can't I extend something that's certain to not complete in time?
 
2 Attachment(s)
Maybe I am stupid, but I set up a manual assignment to do a LL test with an exponent of 332,311,079


[URL]https://www.mersenne.org/report_exponent/?exp_lo=332311079&full=1[/URL]


According to my CPU screen, this has been running one day, completed 1.5%, and will take 94 days to complete. However, according to the "Assignment Expiration Details", this will expire in 60 days. I have tried to extend the expiration date, but every time I do this, the server reports it has done it, but it is still shown as expiring in 60 days. It's clear I can't complete this in 60 days. 120 days should give me long enough, which does not seem an excessively long time given the exponent is large. See screenshots



User 'yangyang6666' attempted this exponent before. It was assigned in 2016, he/she got 34% done 9 months later, and then let it expire in 2020. So yangyang6666 had it for 4 years, but I am limited to 60 days, which is too short.



Dave

Viliam Furik 2021-02-18 12:18

Don't worry about the expiration, it will not expire. AFAIK, the assignment rules are not displaying correct expiration up there, and so far nobody bothered to fix it, because it is not at all a pressing matter. Someone else might correct me.

And also, please, stop the test, and restart it as PRP. As Kriesel already mentioned, in the previous thread, that it has better error checking capabilities, and most importantly, a proof of correctness is generated afterwards so that no double-check test is needed (unless it is a prime, of course).

This should be a correct worktodo line for the test:

[CODE]PRP=N/A,1,2,332311079,-1[/CODE]

You should keep the original LL assignment, and run it as a PRP.

Uncwilly 2021-02-18 13:53

Until your machine checks in, PrimeNet does not you are have actually started working on it. Try putting your assignment ID (that long random string in your line in your worktodo.txt file in place of the N/A in Villiam's example and run the PRP. The assignment type is close enough that PrimeNet should be ok with it.

drkirkby 2021-02-18 16:31

Em,

I'm confused by this. I thought it wise to unassign the exponent, as the server might get confused with me requesting to do one test, then changing it to something else. Now I get the message below when I try to request that exponent. The funny thing is, if I put that exponent as a minimum, and leave the maximum open, it finds me a larger one to do.



What sort of resources will I need for this - particularly in terms of disk space? I have 2 GHz CPU, 26 cores and 64 GB RAM.



Error text: No assignment available meeting CPU, program code and work preference requirements, cpu_id: 2399764, cpu # = 0, user_id = 244634

drkirkby 2021-02-18 16:34

I was expecting if I set my exponent as the minium, and maximum, it would give me that one to do, given nobody else is doing it, and I have unassigned it myself.



Perhaps I would have been better not unassinging it to myself, but then I did not want to confuse the server, which is why I did it.



I don't really want to test an even bigger one! Is PRP slow/faster than LL on exponents of that size?

kruoli 2021-02-18 16:36

[QUOTE=drkirkby;571885][URL]https://www.mersenne.org/report_exponent/?exp_lo=332311079&full=1[/URL][/QUOTE]

The exponent is still assigned to you. Maybe try unassigning it again?

drkirkby 2021-02-18 18:35

[QUOTE=kruoli;571915]The exponent is still assigned to you. Maybe try unassigning it again?[/QUOTE]
Thanks. I seem to have got rid of it, although I would have liked to have done it, as there are no factors up to 2^80. But despite it is now not assigned to anyone, I can't get it for manual testing - I get a message

"Error text: No assignment available meeting CPU, program code and work preference requirements, cpu_id: 2399764, cpu # = 0, user_id = 244634"

Yet for some strange reason, asking for a 100 million digit candidate to test, I get allocated a larger one! I can't really see the logic of that.

I'm now doing 332646233

[URL]https://www.mersenne.org/report_exponent/?exp_lo=332646233&exp_hi=&full=1[/URL]


which is larger, and has not been tested for factors of 2^77 or greater. Testing a smaller number, which had been tested for more factors seemed more likely to find a prime to me.
I'm quite puzzled how the assignments are given out. Why should I not have the resources to do a certain size exponent, but have them to do a larger one?
Dave


All times are UTC. The time now is 17:57.

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