World record sized double-check?

Siegmund · 2016-05-09, 16:36

A machine of mine set to "world record sized numbers to test" recently completed M74557619. I was quite startled to see it say "C-Verified" on my results page. Oh well, I thought, I got a recycled exponent and the guy who expired finished it after all.

But then I looked at the exponent status page, and saw that in fact the first-time test had been completed 2005-11-25, long before I was assigned the exponent.

Apparently "world record sized numbers to test" does not mean "world record sized numbers for first-time testing": is assigning world-record-sized double checks an intended behavior, or a bug? I can't remember ever seeing a double-check assigned to one of my workers that was working on world-record-sized numbers before.

GRB

Mark Rose · 2016-05-09, 17:01

I believe with probably bad results the high exponents have been thrown back into the first time pool.

Uncwilly · 2016-05-09, 17:01

World Record means that the exponent is larger than the exponent of the largest known prime. If the original test had a non-zero error code, it might have been mark as available for another first time check. But, when you match and you don't have a "bad" error code, it is marked verified.

henryzz · 2016-05-09, 17:24

In 2005 that test would have taken a long time to finish(My 2007 era Q6600 is 4x slower than a modern skylake at LLR and probably similar for LL tests). The longer it takes to run a test the more likely it is to have errors.

Madpoo · 2016-05-09, 20:45

Quote:

Originally Posted by Uncwilly

World Record means that the exponent is larger than the exponent of the largest known prime. If the original test had a non-zero error code, it might have been mark as available for another first time check. But, when you match and you don't have a "bad" error code, it is marked verified.

Uncwilly is correct... it was marked suspect during the first run and results like that are thrown back into the assignment pool as another "first time check". 50% of those do turn out to be bad, on average, so it's understandable.

www.mersenne.org/report_ll/?exp_lo=74557619

You can see the error code using that page.

Siegmund · 2016-05-09, 22:04

That makes sense, and is a good reason to retest it right away.

I didn't see the error code on the regular exponent report (http://www.mersenne.org/report_expon...4557619&full=1) and never crossed my mind that report_ll and report_exponent gave different details.

Thanks all for explanation. (And 2005-11-25 vs. 2015-11-25 was my typo, sorry to sidetrack you, henryzz.)

GRB

Uncwilly · 2016-05-09, 22:39

Quote:

Originally Posted by Siegmund

But then I looked at the exponent status page, and saw that in fact the first-time test had been completed 2005-11-25, long before I was assigned the exponent.

Quote:

Originally Posted by henryzz

In 2005 that test would have taken a long time to finish

2005 was a typo, should be 2015.

2016-05-09, 16:36	#1
Siegmund Mar 2014 52₁₀ Posts	World record sized double-check? A machine of mine set to "world record sized numbers to test" recently completed M74557619. I was quite startled to see it say "C-Verified" on my results page. Oh well, I thought, I got a recycled exponent and the guy who expired finished it after all. But then I looked at the exponent status page, and saw that in fact the first-time test had been completed 2005-11-25, long before I was assigned the exponent. Apparently "world record sized numbers to test" does not mean "world record sized numbers for first-time testing": is assigning world-record-sized double checks an intended behavior, or a bug? I can't remember ever seeing a double-check assigned to one of my workers that was working on world-record-sized numbers before. GRB

2016-05-09, 17:01	#2
Mark Rose "/X\(‘-‘)/X\" Jan 2013 3×1,033 Posts	I believe with probably bad results the high exponents have been thrown back into the first time pool. Last fiddled with by Mark Rose on 2016-05-09 at 17:42

2016-05-09, 22:04	#6
Siegmund Mar 2014 2²×13 Posts	That makes sense, and is a good reason to retest it right away. I didn't see the error code on the regular exponent report (http://www.mersenne.org/report_expon...4557619&full=1) and never crossed my mind that report_ll and report_exponent gave different details. Thanks all for explanation. (And 2005-11-25 vs. 2015-11-25 was my typo, sorry to sidetrack you, henryzz.) GRB Last fiddled with by Siegmund on 2016-05-09 at 22:06

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Sieving freakishly big MMs (was "World record" phone number?)	davieddy	Operazione Doppi Mersennes	284	2021-10-24 13:53
World Record Factorial Prime Found	rogue	Lounge	8	2012-03-02 16:41
70 billion pixels Budapest (world record)	R. Gerbicz	Science & Technology	0	2010-07-28 01:50
First check and double check llrnet servers.	opyrt	Prime Sierpinski Project	3	2009-01-02 01:50
Double-check check?	M0CZY	Software	15	2008-10-30 14:20

2016-05-09, 17:01	#3
Uncwilly 6809 > 6502 """"""""""""""""""" Aug 2003 101×103 Posts 3·3,631 Posts	World Record means that the exponent is larger than the exponent of the largest known prime. If the original test had a non-zero error code, it might have been mark as available for another first time check. But, when you match and you don't have a "bad" error code, it is marked verified.

2016-05-09, 17:24	#4
henryzz Just call me Henry "David" Sep 2007 Liverpool (GMT/BST) 13617₈ Posts	In 2005 that test would have taken a long time to finish(My 2007 era Q6600 is 4x slower than a modern skylake at LLR and probably similar for LL tests). The longer it takes to run a test the more likely it is to have errors.