mersenneforum.org  

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

Reply
 
Thread Tools
Old 2020-10-13, 02:15   #45
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

4,733 Posts
Default

Quote:
Originally Posted by ATH View Post
Would it be hard to make the server deny any TF assignments in like 53M-59M and 93M-108M ?
Set it to deny excessive-tf-depth assignments globally. Then TF assignments can't tie up any exponents that are sufficiently factored.
kriesel is offline   Reply With Quote
Old 2020-10-13, 15:13   #46
ATH
Einyen
 
ATH's Avatar
 
Dec 2003
Denmark

2·3·7·71 Posts
Default

Quote:
Originally Posted by kriesel View Post
Set it to deny excessive-tf-depth assignments globally. Then TF assignments can't tie up any exponents that are sufficiently factored.
I doubt there is a limit set on all exponents that is up to date with the GPU factoring limits.
There was back in the day when only Prime95/mprime was used for factoring.

Plus it would block everyone who wants to factor deep for example in <50M range like the "under 20M unfactored exponents" project.
ATH is offline   Reply With Quote
Old 2020-10-13, 15:20   #47
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

4,733 Posts
Default

Quote:
Originally Posted by ATH View Post
I doubt there is a limit set on all exponents that is up to date with the GPU factoring limits.
...t would block everyone who wants to factor deep for example in <50M range like the "under 20M unfactored exponents" project.
Ok, if that application of excessive TF is accepted, cap TF assignments for p>40M to no more than one bit higher than the corresponding tf target in the gpu72 line of mersenne.ca lookup for the exponent concerned, or the approximate linear interpolation of that. It's common to see random exponents at the wavefront wastefully TF 2 or 3 levels above optimal, and/or P-1 skipped, before primality testing commences.

Last fiddled with by kriesel on 2020-10-13 at 15:23
kriesel is offline   Reply With Quote
Old 2020-10-13, 15:29   #48
firejuggler
 
firejuggler's Avatar
 
Apr 2010
Over the rainbow

1001101010112 Posts
Default

That would be about half the GhZ cost of a prp, right? Or is it a quarter?
firejuggler is offline   Reply With Quote
Old 2020-10-13, 17:49   #49
S485122
 
S485122's Avatar
 
Sep 2006
Brussels, Belgium

1,597 Posts
Default

Quote:
Originally Posted by kriesel View Post
Ok, if that application of excessive TF is accepted, cap TF assignments for p>40M to no more than one bit higher than the corresponding tf target in the gpu72 line of mersenne.ca lookup for the exponent concerned, or the approximate linear interpolation of that. It's common to see random exponents at the wavefront wastefully TF 2 or 3 levels above optimal, and/or P-1 skipped, before primality testing commences.
Much simpler, instead of having the lowest unverified exponent for double-check TF and the lowest untested exponent, for first time checks factoring, set it to the lowest CAT2 exponent in each category (respecting the fact that CAT 0 and CAT1 "Not available for manual testing." (CAT2 should only be available for reliable users having signed up for getting the smallest exponents.) Why not use the existing rules (and already implemented code) applied to DC, LL, PRP and PRP-CERT ?

Jacob
S485122 is offline   Reply With Quote
Old 2020-10-13, 18:34   #50
Uncwilly
6809 > 6502
 
Uncwilly's Avatar
 
"""""""""""""""""""
Aug 2003
101×103 Posts

2×5×7×127 Posts
Default

Quote:
Originally Posted by S485122 View Post
Why not use the existing rules (and already implemented code) applied to DC, LL, PRP and PRP-CERT ?
That would prevent things like this assignment (in the bottom half of Cat 0) 95000533
Uncwilly is online now   Reply With Quote
Old 2020-10-13, 18:42   #51
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

127D16 Posts
Default

Quote:
Originally Posted by Uncwilly View Post
That would prevent things like this assignment (in the bottom half of Cat 0) 95000533
What does "that" refer to? (And firejuggler, same question.)
95000533 is already TF 2 levels higher than goal, and P-1 done rather thoroughly too. https://www.mersenne.ca/exponent/95000533
Justifiable TF level is lowered by P-1 integration into gpuowl primality testing, resulting in higher economic P-1 bounds than previously. And lowered by PRP proof reducing primality effort by ~50%.

Last fiddled with by kriesel on 2020-10-13 at 18:45
kriesel is offline   Reply With Quote
Old 2020-10-13, 18:43   #52
S485122
 
S485122's Avatar
 
Sep 2006
Brussels, Belgium

1,597 Posts
Default

Quote:
Originally Posted by Uncwilly View Post
That would prevent things like this assignment (in the bottom half of Cat 0) 95000533
Until now the rules state "Not available for manual testing."

And it could concern only untested exponents in the first test range and the unverified exponents in the double check range (for the time it will exist.)

And if adding those conditions would be too difficult, the TF of verified exponent can wait until the wavefront passed.

Then it is possible that the size of the different categories could benefit from a review.

Jacob
S485122 is offline   Reply With Quote
Old 2020-10-13, 18:46   #53
chalsall
If I May
 
chalsall's Avatar
 
"Chris Halsall"
Sep 2002
Barbados

246E16 Posts
Default

Quote:
Originally Posted by Uncwilly View Post
That would prevent things like this assignment (in the bottom half of Cat 0) 95000533
Yup. And because this is a manual assignment, it won't expire until March of next year.

Further, because of the way TF assignments are handled, even if someone did the TF'ing to 78 (which would be silly) the assignment /still/ wouldn't be released.
chalsall is online now   Reply With Quote
Old 2020-10-13, 21:16   #54
ATH
Einyen
 
ATH's Avatar
 
Dec 2003
Denmark

56468 Posts
Default

Server should block TF in Cat 0+1 and probably Cat 2 (and maybe Cat 3?) and it should give no AID on the TF assignments either from manual assignment or through Prime95/mprime. Particularly since all the El Dorkos and morons and idiots out there are using this strategy to annoy and block progress.

Before handing out AID the server already checks the category anyway to see if the users computer requirement are met for Cat 0/1/2/3 and denies the AID if not.
I do not know how this is programmed, but logically it seems reasonable that an additional check could be put in like: IF (Cat<3 AND Assignment == TF) DENY!

Last fiddled with by ATH on 2020-10-13 at 21:27
ATH is offline   Reply With Quote
Old 2020-10-13, 21:44   #55
Uncwilly
6809 > 6502
 
Uncwilly's Avatar
 
"""""""""""""""""""
Aug 2003
101×103 Posts

100010101110102 Posts
Default

There is this batch down in the DC Cat 1 range (just above the Cat 0 level).
https://www.mersenne.org/assignments...first=1&exp1=1

ATH and S485122 have suggested server rules that would prevent this.
Uncwilly is online now   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
How many bits does/did the server trial factor to? Jayder Information & Answers 6 2015-01-25 03:29
Trial Factor Bit Depth lavalamp Operation Billion Digits 8 2010-08-02 18:49
trial division over a factor base Peter Hackman Factoring 7 2009-10-26 18:27
P95 Trial Factor speeds 40M vs 100M harlee Software 3 2006-10-15 04:38
Shortest time to complete a 2^67 trial factor (no factor) dsouza123 Software 12 2003-08-21 18:38

All times are UTC. The time now is 02:24.

Mon Nov 30 02:24:34 UTC 2020 up 80 days, 23:35, 3 users, load averages: 0.95, 1.25, 1.29

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, 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.