mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Information & Answers (https://www.mersenneforum.org/forumdisplay.php?f=38)
-   -   Server bug -- new? (https://www.mersenneforum.org/showthread.php?t=15756)

Christenson 2011-07-12 03:38

Server bug -- new?
 
I asked the server for a TF assignment using the usual form, along the following lines, and, instead of getting an error message, got half a masthead....and a long, long delay. Time is now 23:32, GMT-4.

[url]http://www.mersenne.org/manual_assignment/?cores=1&num_to_get=20&pref=102&exp_lo=40000000&exp_hi=60000000&B1=Get+Assignments[/url]

Is this a new server bug?

The rationale for the request is my standard calculation that with GPUs at 10x faster than CPUs, they should do about 3 additional bit levels; at 100x faster, they should do about 7 extra bit levels....so in my mind it makes sense to revisit some exponents that already have a single LL test against them.

The server seems to agree, sometimes....in that a non-matching LL-D of mine has been assigned to TF. How do I get the server to assign it, though?

NBtarheel_33 2011-07-12 04:12

[QUOTE=Christenson;266164]I asked the server for a TF assignment using the usual form, along the following lines, and, instead of getting an error message, got half a masthead....and a long, long delay. Time is now 23:32, GMT-4.

[URL]http://www.mersenne.org/manual_assignment/?cores=1&num_to_get=20&pref=102&exp_lo=40000000&exp_hi=60000000&B1=Get+Assignments[/URL]

Is this a new server bug?

The rationale for the request is my standard calculation that with GPUs at 10x faster than CPUs, they should do about 3 additional bit levels; at 100x faster, they should do about 7 extra bit levels....so in my mind it makes sense to revisit some exponents that already have a single LL test against them.

The server seems to agree, sometimes....in that a non-matching LL-D of mine has been assigned to TF. How do I get the server to assign it, though?[/QUOTE]

I get nothing when I click on that link, even after a refresh (Wait - I just got half the page frozen on the screen). If I understand this right, you are asking for singly-LL'ed exponents between 40M and 60M to be handed to you as TF assignments, correct? Could it be that the server is perhaps set up to only hand out previously-LL'ed exponents as DCs?

What if you just asked for 20 DC candidates between 40M and 60M, and then did the extra bits of trial factoring "by hand", i.e. by inserting the appropriate "Factor=" lines in worktodo (or the analogous file in mfaktc)?

UPDATE! Weird...when I refreshed my assignments page...it looks like PrimeNet has granted me a single LL assignment in the 47M range...and it's not even had a first LL!

davieddy 2011-07-12 04:52

[QUOTE=NBtarheel_33;266167]
UPDATE! Weird...when I refreshed my assignments page...it looks like PrimeNet has granted me a single LL assignment in the 47M range...and it's not even had a first LL![/QUOTE]

If you watch the "Primenet Summary" page, you will see that
returned exponents are dished out at mightnight UTC.
My guess is that your exponent needs P-1 as well.

David

cheesehead 2011-07-12 05:28

[QUOTE=Christenson;266164]How do I get the server to assign it, though?[/QUOTE][U]If[/U] you were not insistent on an automatic procedure, and didn't mind a bit of manual labor, you could do it the LMH way, as explained in the "[B]How to LMH using Prime95 v25.8 and PrimeNet v5[/B]" thread at [URL]http://mersenneforum.org/showthread.php?t=11308[/URL]

However, it looks like you [U]are[/U] trying for an automated procedure.

OTOH, the semi-manual LMH procedure might actually take less clock time than the "long, long delay"! :smile:

Christenson 2011-07-12 11:35

[QUOTE=cheesehead;266176][U]If[/U] you were not insistent on an automatic procedure, and didn't mind a bit of manual labor, you could do it the LMH way, as explained in the "[B]How to LMH using Prime95 v25.8 and PrimeNet v5[/B]" thread at [URL]http://mersenneforum.org/showthread.php?t=11308[/URL]

However, it looks like you [U]are[/U] trying for an automated procedure.

OTOH, the semi-manual LMH procedure might actually take less clock time than the "long, long delay"! :smile:[/QUOTE]

Who said I [U]insisted[/U] on the "manual assignments" automatic procedure? In the long run, automatic procedures are *supposed* to be faster, or fail.....I'm just reporting that the semi-auto procedure is misbehaving and wondering what to try next, while busily ignoring RDS sage advice....I've done plenty of OBD assignments on full manual....

-- Science Man Christenson

cheesehead 2011-07-12 21:44

[QUOTE=Christenson;266194]Who said I [U]insisted[/U] on the "manual assignments" automatic procedure?[/QUOTE]No one.

The sentence, "If you were ..." used the subjunctive mood to express something about a possibility, not the indicative mood to state a certainty.

I was trying to do you a courtesy by presenting a possibility without presuming that it necessarily would or would not be to your liking.


All times are UTC. The time now is 11:32.

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