mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Data > Marin's Mersenne-aries

Closed Thread
 
Thread Tools
Old 2017-04-16, 20:16   #1420
Madpoo
Serpentine Vermin Jar
 
Madpoo's Avatar
 
Jul 2014

29·113 Posts
Default

Quote:
Originally Posted by kladner View Post
Oh, well. I suppose user "WWU-CS" is the one with the spotty record, not that I should sneer. I have my own rap sheet of proven bad results.
Well, you win some, you lose some. Like I said, this list of smaller exponents includes computers that, for any particular month, may have an almost equal # of bad and good, so it's more of a crapshoot.

Over time I expect we'll still come across computers where they have far more bad than good, but we've actually done a pretty good job of finding and testing the worst of the worst. There are still some unknowns out there that simply don't have enough double-checked work and it's a safe bet there are some terrible machines out there waiting to be discovered.

For those, there's really no other way to know except run at least one double-check of any machine that's never had one done yet... I've done some of that on old systems (version 4 clients) but the rest will come up naturally here and there. I guess we could have a project to tackle those, and I think we've talked about some ways of doing that (a special assignment type of some sort).
Madpoo is offline  
Old 2017-04-16, 21:12   #1421
kladner
 
kladner's Avatar
 
"Kieren"
Jul 2011
In My Own Galaxy!

67·151 Posts
Default

Quote:
Originally Posted by kladner View Post
I took-
Code:
40260631    2    2    1    0    1    0    DoubleCheck=40260631,72,1
And did not match the first run.
kladner is offline  
Old 2017-04-16, 21:21   #1422
Mark Rose
 
Mark Rose's Avatar
 
"/X\(‘-‘)/X\"
Jan 2013

2×31×47 Posts
Default

Quote:
Originally Posted by Madpoo View Post
For those, there's really no other way to know except run at least one double-check of any machine that's never had one done yet... I've done some of that on old systems (version 4 clients) but the rest will come up naturally here and there. I guess we could have a project to tackle those, and I think we've talked about some ways of doing that (a special assignment type of some sort).
How many of those machines are there?
Mark Rose is online now  
Old 2017-04-16, 22:09   #1423
ric
 
ric's Avatar
 
Jul 2004
Milan, Ita

22×32×5 Posts
Default

Quote:
Originally Posted by Mark Rose View Post
How many of those machines are there?
My educated guess... a few more than 5330 in the range 0-100M

Last fiddled with by ric on 2017-04-16 at 22:14 Reason: added range
ric is offline  
Old 2017-04-17, 14:58   #1424
Madpoo
Serpentine Vermin Jar
 
Madpoo's Avatar
 
Jul 2014

1100110011012 Posts
Default

Quote:
Originally Posted by Mark Rose View Post
How many of those machines are there?
There are a total of 13,683 systems that have never had a single successful (or even unsuccessful) double-check run on them with a combined total LL test count of 85,197 exponents.

Of those 13,683 machines, if I run a report to get the smallest available exponent from each of them, I get a list of 12.466 exponents (some systems already have all of their results assigned for DC...usually systems with only a single check to their credit).

About 3200 of those exponents are below 50M, another 6100 are between 50M-60M, then another 1900 between 60M-70M, and then 1130 from 70M-80M. The rest are a wild mix of 80M+ even including a bunch of 100 million digit stuff that I don't care about if they're wrong.

So if any of you just wanted to do some double-checks to "chalk one up" for these systems, knowing that you'll probably match 95% of them (based on the overall odds), I can spit out a list of those. Maybe I could even periodically generate a "lowest 100 exponents" list based on that, just to get the smallest stuff included.

I guess that would still qualify as a strategic double-check since it has a more specific goal in mind than a typical DC.

With that in mind... if interested, there are the first 100 from that list. Note that several of these are Category zero so if you want it, you'll have to assign it by starting work and then checking in to the server to get the assignment, and there's also a good chance by the time you see this someone will already have it auto-assigned.

Code:
worktodo
DoubleCheck=39954737,71,1
DoubleCheck=39977671,71,1
DoubleCheck=39980543,71,1
DoubleCheck=39991681,71,1
DoubleCheck=39994193,73,1
DoubleCheck=39996581,71,1
DoubleCheck=40004947,72,1
DoubleCheck=40025353,72,1
DoubleCheck=40030183,72,1
DoubleCheck=40030219,72,1
DoubleCheck=40034287,72,1
DoubleCheck=40046597,72,1
DoubleCheck=40049413,72,1
DoubleCheck=40059133,72,1
DoubleCheck=40068799,72,1
DoubleCheck=40097593,72,1
DoubleCheck=40098923,72,1
DoubleCheck=40130471,72,1
DoubleCheck=40130989,72,1
DoubleCheck=40139483,72,1
DoubleCheck=40140367,72,1
DoubleCheck=40162201,72,1
DoubleCheck=40175767,72,1
DoubleCheck=40182199,72,1
DoubleCheck=40200451,72,1
DoubleCheck=40202047,72,1
DoubleCheck=40233229,72,1
DoubleCheck=40242767,72,1
DoubleCheck=40252547,72,1
DoubleCheck=40259833,72,1
DoubleCheck=40268911,72,1
DoubleCheck=40291663,72,1
DoubleCheck=40293061,72,1
DoubleCheck=40301491,72,1
DoubleCheck=40340669,72,1
DoubleCheck=40342429,72,1
DoubleCheck=40351481,72,1
DoubleCheck=40373077,72,1
DoubleCheck=40396871,72,1
DoubleCheck=40407799,72,1
DoubleCheck=40441871,72,1
DoubleCheck=40451701,72,1
DoubleCheck=40452131,72,1
DoubleCheck=40453397,72,1
DoubleCheck=40465417,72,1
DoubleCheck=40482733,72,1
DoubleCheck=40483039,72,1
DoubleCheck=40745893,72,1
DoubleCheck=40759967,72,1
DoubleCheck=40766417,72,1
DoubleCheck=40770137,72,1
DoubleCheck=40770563,72,1
DoubleCheck=40780781,72,1
DoubleCheck=40793549,72,1
DoubleCheck=40793723,72,1
DoubleCheck=40854631,72,1
DoubleCheck=40859191,72,1
DoubleCheck=40905539,72,1
DoubleCheck=40965433,72,1
DoubleCheck=40968409,72,1
DoubleCheck=40968713,72,1
DoubleCheck=40973131,72,1
DoubleCheck=40974823,72,1
DoubleCheck=40983511,72,1
DoubleCheck=40984393,72,1
DoubleCheck=40986271,72,1
DoubleCheck=41021971,72,1
DoubleCheck=41037617,72,1
DoubleCheck=41043029,72,1
DoubleCheck=41048317,72,1
DoubleCheck=41048519,72,1
DoubleCheck=41049049,72,1
DoubleCheck=41056313,72,1
DoubleCheck=41072287,72,1
DoubleCheck=41077171,72,1
DoubleCheck=41085127,72,1
DoubleCheck=41097593,72,1
DoubleCheck=41100893,72,1
DoubleCheck=41105413,72,1
DoubleCheck=41119301,72,1
DoubleCheck=41119801,72,1
DoubleCheck=41130121,72,1
DoubleCheck=41131481,72,1
DoubleCheck=41139143,72,1
DoubleCheck=41140331,72,1
DoubleCheck=41140727,72,1
DoubleCheck=41155739,72,1
DoubleCheck=41164289,72,1
DoubleCheck=41169229,72,1
DoubleCheck=41181263,72,1
DoubleCheck=41181887,72,1
DoubleCheck=41181997,72,1
DoubleCheck=41182637,72,1
DoubleCheck=41183297,73,1
DoubleCheck=41188249,72,1
DoubleCheck=41194051,72,1
DoubleCheck=41199559,72,1
DoubleCheck=41200331,72,1
DoubleCheck=41209877,72,1
DoubleCheck=41210021,72,1
Madpoo is offline  
Old 2017-04-17, 15:55   #1425
Mark Rose
 
Mark Rose's Avatar
 
"/X\(‘-‘)/X\"
Jan 2013

2×31×47 Posts
Default

If you could make that query a new work type, I would automate my systems to fetch that (with the usual DC category rules applying).
Mark Rose is online now  
Old 2017-04-17, 18:13   #1426
ric
 
ric's Avatar
 
Jul 2004
Milan, Ita

22·32·5 Posts
Default

Quote:
Originally Posted by Madpoo View Post
There are a total of 13,683 systems that have never had a single successful (or even unsuccessful) double-check run on them
Hmm... are you including all machines, or just the ones with at least one unverified expo?
ric is offline  
Old 2017-04-17, 18:34   #1427
GP2
 
GP2's Avatar
 
Sep 2003

2×1,291 Posts
Default

Quote:
Originally Posted by Madpoo View Post
Of those 13,683 machines, if I run a report to get the smallest available exponent from each of them, I get a list of 12.466 exponents (some systems already have all of their results assigned for DC...usually systems with only a single check to their credit).
I think the majority of these machines, maybe the vast majority, have only returned a single LL result, ever.

It won't be particularly helpful to double-check these ahead of time, because whether that single result is good or bad, it provides no clues about any other exponents. So these are by definition "unstrategic" double checks.

Maybe it's better to focus on machines that have two or more unverified results and no verified results. Then if any results are bad we can look at double-checking other exponents returned by the same machine.
GP2 is offline  
Old 2017-04-18, 01:11   #1428
bgbeuning
 
Dec 2014

22×32×7 Posts
Default Title

Quote:
Originally Posted by Madpoo View Post
There are a total of 13,683 systems that have never had a single successful (or even unsuccessful) double-check run on them with a combined total LL test count of 85,197 exponents.

Of those 13,683 machines, if I run a report to get the smallest available exponent from each of them, I get a list of 12.466 exponents (some systems already have all of their results assigned for DC...usually systems with only a single check to their credit).

About 3200 of those exponents are below 50M, another 6100 are between 50M-60M, then another 1900 between 60M-70M, and then 1130 from 70M-80M. The rest are a wild mix of 80M+ even including a bunch of 100 million digit stuff that I don't care about if they're wrong.
I here by dub thee "data miner extraordinaire".
bgbeuning is offline  
Old 2017-04-18, 13:09   #1429
airsquirrels
 
airsquirrels's Avatar
 
"David"
Jul 2015
Ohio

51710 Posts
Default

I've been quite busy with work lately, so I've just been grabbing a few manual exponents for DC lately (honestly I think I even have some old assignments that I still need to clean up).

If there would be an easy way to grab some of this specialized work I'd be happy to focus efforts on this instead.
airsquirrels is offline  
Old 2017-04-18, 14:26   #1430
Madpoo
Serpentine Vermin Jar
 
Madpoo's Avatar
 
Jul 2014

29·113 Posts
Default

Quote:
Originally Posted by ric View Post
Hmm... are you including all machines, or just the ones with at least one unverified expo?
That was a count of all systems that have returned at least one LL result, but so far none of their results have been verified (or proven bad).
Madpoo is offline  
Closed Thread

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Double-Double Arithmetic Mysticial Software 50 2017-10-30 19:16
Clicking an exponent leads to 404 page marigonzes Information & Answers 2 2017-02-14 16:56
x.265 half the size, double the computation; so if you double again? 1/4th? jasong jasong 7 2015-08-17 10:56
What about double-checking TF/P-1? 137ben PrimeNet 6 2012-03-13 04:01
Double the area, Double the volume. Uncwilly Puzzles 8 2006-07-03 16:02

All times are UTC. The time now is 18:57.

Tue Dec 1 18:57:03 UTC 2020 up 82 days, 16:08, 2 users, load averages: 1.42, 1.56, 1.75

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.